Feedback (from back in the day)

I switched over to a new, better feedback system that lets us vote for the ideas we like most.

I want to thank you all again for filling this page up with so many ideas and words of encouragement,
 
Ken
kendroberts@gmail.com

300 Responses to “Feedback (from back in the day)”

  1. Dad Says:

    Glad to see you’re making improvements to the site. I think the posting idea is really good…keep it up!

    Dad

  2. Kelly Campbell Says:

    Nice Work! You’ve done a great job with the visualization of the data.

    kc

  3. James Liang Says:

    Your site was mentioned on AJC.com and will probably be in the Atlanta paper tomorrow. Here is the tidbit:

    PLAYING THE PERCENTAGES: If you’re sweating the Thrashers’ playoff chances, don’t. Ken Roberts has figured it out, and they entered Thursday’s game with a 92.7 percent chance of making the postseason. Roberts, who writes software in Raleigh, computes the percentage by running a computer simulation millions of times each night, and updates it nightly on his Web site http://www.sportsclubstats.com.

  4. Johnny Horizon Says:

    Something’s wrong with the graph; looks like maybe the Rangers’ and Flyers’ plots are swapped? [Ed. This should be fixed, thanks.]

  5. Johnny Horizon Says:

    All kinds of weirdness with the NHL Eastern Conference graph. The mouseover stuff is flaky in Firefox 2 on WinXP. If I start at the top of list and move the mouse down over each team in succession, sometimes it not showing the right team. [Ed. This should be fixed, thanks.]

    Cool site though. Saw it from the AJC article James Liang sites above.

  6. Ken Roberts Says:

    James, that’s great, thanks for the heads up.

    Johnny, you are correct, thanks for the info. My mouseover stuff is bad. You can see what team the graph is really selecting by looking for the team’s name in that sentence above the graph. Often you have to move your mouse off to the side of the team you want to select and then move it back over the team to get the graph to update. And even if you do all that, if your looking at a Out or In team the graph sometimes highlights a different team than the sentence above the graph says is highlighted. This last thing I’m working on now. The mouseover stuff will take longer. Your right, it should just work like you expect.

    And another thing, if you move the mouse over the the teams really fast as the page is loading, sometimes you will be treated to a mystery generic bar graph. You can refresh to get the real graph back.

    [Ed. Update: Other than the "mystery generic bar graph", all these graph issues should be fixed.]
    [Ed. Update 2: 6 months later, the mystery graph bug is fixed. Where do the months go?]

  7. Mark Says:

    You clearly have too much time on your hands. Cool page! I’ll pass this along to a sports fanatic I work with who will enjoy it immensely.

  8. Ben Dover Says:

    Outstanding work. Please continue to keep it updated.

  9. Ken Roberts Says:

    Thanks Ben, I appreciate it.
    -Mike Hunt

  10. Dackle Says:

    Hi Ken, really enjoy the site, especially the idea of the upcoming big games. I notice your log goes back a few days, are you able to extend this and maybe have a page with the 10 biggest games of the season?

  11. Ken Roberts Says:

    Dackle, that is a great idea. Thanks.

  12. Dackle Says:

    Another variation might be “Chance of winning world series”, although I doubt the relative rankings would change much.

    Once again, I’m really enjoying the “big games” section of your site — I think that’s what makes it unique.

  13. Ken Roberts Says:

    Thanks Dackle. I moved today’s big games up to the top of the team pages.

  14. Dackle Says:

    The big games on the team pages looks great! It’s a nice way of sifting through the out-of-town scoreboard. I mean, generally a fan knows what out-of-town scores they care about, but … something pleasing about having those games ranked.

  15. Cubbyman Says:

    Thanks for letting me know about your site. I’m really impressed with the presentation of the data. I like the upcoming “big game” sections. It is neat to see the potential impact of winning or losing that important game or series.

    Nice job on the site, I’ll be checking back regularly.

    Mike

  16. Jimbo Says:

    I’m very impressed with your logic and your work, and as a Cubs fan, I check your calculation of their chances of making the playoffs every day.
    However, I really need something explained to me. Today, 7/3, you have stated that one of the games affecting the Cubs’ chances of making the playoffs is the Royals (+0.0) vs. the Mariners (-0.3). Now, for the life of me, I can’t imagine how either American league team winning this game will affect the Cubs chances, at all, of them making the playoffs in the National League. I’d love some help with this.
    And thanks again.
    Jimbo

  17. Ken Roberts Says:

    Mike and Jimbo, thank you.
    Jimbo, I just got back from the beach. I’ll take a look and the numbers you pointed out tomorrow. Thanks for the feedback.
    Ken

  18. au-s Says:

    Hello,

    In the Diagram on the F1-page you have (whan you hover your mouse over dots) wrong sponsor in the Brittish Grand Prix which was the last one.
    It is not Fosters since all alcohol and ciggarettes companies retired from advertising. It should be Santander Brittish Grand Prix.

  19. Ken Roberts Says:

    Thanks au-s, I’ve corrected the name.
    Jimbo, I have not forgotten your question.

  20. Jimbo Says:

    Whenever youre ready, Ken. Take your time. Enjoy the summer. But, I did notice the same occurrence for Saturday, 7/14, when one of the “big games” for the Cub’s playoff chances is the Devil Rays vs. the Yankees. Unless this is some indication that the Yanks want our Big Z (lol), I still cant figure out the connection.
    But again, Im a big fan of your site. Thanks for your work.

  21. Tbone Says:

    Ken – found your site yesterday through a link from the Detroit Tigers Weblog. Love it! Like Jimbo above, however, I’m stumped as to why any opposite League games have an affect on any given team’s chances to make the playoffs. Yesterday, one of the Tigers “Big Games” was the Brewers vs. Diamondbacks.

    I suspect that snce the Tigers played the Brewers in inter-league play that they somehow got factored into your program. Or is there an opponent’s record tie-breaker I ma unaware of? Any light you can shed would be appreciated. Thanks!

  22. Scrig Says:

    Wow, great site. Very interesting, will definately pass it on.

  23. Phill McCrevis Says:

    Outstanding work–very cool. Please continue to keep it updated.

  24. Scrig Says:

    Ken,

    I noticed that the Cardinals and Brewers had a double header this past Saturday (7/28) but under the Cubs “Big Games” section, the impact of only one game is shown. Are the two games simply factored together?

  25. Ken Roberts Says:

    Scrig,
    Thanks for pointing that out. It’s neat to see someone is actually looking at the numbers.

    Unfortunately the “Big Games” section does not handle makeup games. (One of the doubleheader games was a makeup of a game postponed back on April 13th.) My program did not know there was a doubleheader Saturday until it got the scores that night, and by then it had already run its “big game” logic. I’ve added that to my list of things to fix. I don’t think I’ll get to it this season. [Edit, this is fixed now]

    The rest of the site (the teams total odd of making the playoff, daily change, the graphs) handle makeup games correctly. So the subtotal in the big games section will be wrong and one of the games will be missing, but the numbers above in the Chance of making the playoffs section are correct.

    Thanks,
    Ken

    PS Phill: thanks for writing, I’ll pass.

  26. Lt Col Patrick Says:

    Great blog, Ken. I’ll be checking it probably more times than the wife would like :)

    Patrick

    http://dutyinthedesert.blogspot.com/

  27. robert bourne Says:

    good stuff on Nascar it’s helped in my fantasy league plus things to watch in upcoming races…

  28. Ken Roberts Says:

    Thanks Patrick and Robert.

    Jimbo and Tbone: Your right. There is no reason for a MLB game between 2 teams in the opposing league to have any affect on a teams playoff chances. I’m not sure, but I think what is happening is I’m reporting a result that is not “statistically significant”. I upped the number of seasons I simulate for the “what if” section from 250,000 to 400,000 to hopefully fix this. We’ll see. Thank you for clueing me into this.

  29. Dackle Says:

    Why is it that the likelihood of a team going 35-0 or 0-35 down the stretch is 1,000 in every 10,000,000 simulations? Seems like all of the other possible won-lost records follow the normal distribution, but the scenarios of winning or losing every remaining game occur at a fixed 1,000 times for every 10,000,000 runs. For example the Blue Jays:

    If finish: Chance in Chance team will finish the season at position:
    TW W - L PCT playoffs 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Count

    99 35 - 0 1.000 In 100 0 1,000
    96 32 - 3 .914 In 100 2
    95 31 - 4 .886 In 100 9
    94 30 - 5 .857 In 73 17 11 95
    93 29 - 6 .829 99.6 48 26 23 3 0 445
    92 28 - 7 .800 96.7 25 29 36 7 3 0 2,007
    91 27 - 8 .771 88.4 10 25 38 15 11 0 6,920
    90 26 - 9 .743 69.4 2 14 35 18 28 3 0 20,663
    89 25 - 10 .714 43.6 0 6 23 15 46 11 0 53,596

  30. Ken Roberts Says:

    Dackle: For 1000* of the seasons I force the Blue Jays to always win. And for 1000 of the seasons I force the Blue Jays to always loose. I’m trying to flush out the best and worse case possibilities.
    So for baseball, since there are 30 teams: 30 * (1000 win + 1000 loose) = 60,000 times I lock down a team (and let the rest of the teams be random) and 9,940,000 times everybody is random. 1000 is so small compared to 10 million that it does not noticeably effect the “real” chances.

    [Edit: *it's no longer 1,000. Instead it's some small percent of the total simulated seasons.]

  31. Tribe Fan Says:

    I think that you may be using an old schedule as the indians did play August 27. The Minn series was moved up to allow Cleveland and Seattle to play on Thursday to make up a snow game.

  32. Ken Roberts Says:

    Thank you Tribe Fan for finding this, and thanks for explaining why the schedule changed.
    I’ll rerun the program with the correct results now.

    On an unrelated matter, every teams big games section is blank for tomorrow because of a mistake I made last night.

  33. Tribe Fan Says:

    Thanks and great site.

  34. pstein Says:

    Ken…Don’t know if you saw the recent news item regarding Champ Car’s Phoenix race being cancelled: http://sports.yahoo.com/cart/news?slug=ap-champ-racecanceled&prov=ap&type=lgns

  35. Ken Roberts Says:

    Thank you pstein, I did not know the last race was canceled. I’ll rerun the numbers.

  36. bat Says:

    hey, how bout some college sports added in there! Great site!

  37. Ken Roberts Says:

    bat, thanks. I’ll look at that. If I did NCAA football what could I compute besides “chance will win conference”?
    [Edit: I added Men's and Women's NCAA basketball. I'm working on football for next season]

  38. Greg Groeper Says:

    Hi Ken:
    Your site is a monstrously great effort… and this is just your hobby! Most people don’t do this well at their “real” job. Quick thought on your MLB section.

    The rankings you display (1-14 American & 1-16 National)are a bit irrelevant because of the “Divisional” aspects of the competition. Teams are not competing with teams in other divisions, so teams with better records can lose out while teams with worse records in a different division can win (e.g. the Cubs would be in 5th place in the NL West right now, but are in 1st place in the NL Central.

    It would be helpful if your “digital competitions” took place in the relevant Division only and when viewing a single Division, were displayed as the “chance of finishing” (1-4, 1-5 or 1-6)instead of just comparing all 14 or 16 teams against each other. As mentioned, that comparison, while interesting, does not suggest who will finish at each position in each Division.

    That said, I found the statistical predictive analytical aspects of your site to be impressive and helpful. Maybe next you can predict the winning lottery numbers and retire in grace and comfort? ;-)

    Warm regards,

    Greg G

  39. kiloutimuk Says:

    Hello

    I am really excited. Keep up the great work. Good resources here.

    G’night

  40. Lorenzo St. DuBois Says:

    Probably the best put-together site of this kind on the web, and there are many. The graphs are superb. I really wish there was a way to make it more sophisticated than “coin-flip” though. Perhaps adding a feature that allows the visitor to customize the inputs (maybe adding strength of schedule, run differential considerations, weighting recent performance a little more than performance from 5 months ago). If that happens, this becomes unquestionably the best site of it’s kind on the web.

  41. Mike H. Says:

    The Mets play the Cardinals today in a make-up game with major postseason implications, and it’s not factored into the NL calculations.

  42. Ken Roberts Says:

    Mike, thank you very much for the heads up. I fixed it and started the simulation again. Not sure how long before the site is updated, maybe around noon Eastern time.
    The “odds in” and “what if” numbers are not effected by this, the simulation knew the game still had to be made up, it just did not know it was happening today. Hopefully I’ll fix this issue by next season.

    Ken

  43. Ken Roberts Says:

    Greg, kiloutimuk and Lorenzo, thank you for the encouraging words.

    Greg, I don’t think I could have created this solely in my spare time. I’m not that prolific. In 2006 I started working half time at my “real” job. Around April of that year I started making this site with the other half. Lately I’ve been working even less at my “real” job and more on the site. We’ll see where I can take it before my bank account dries up.
    I’ll add “odds of finishing in which place in you division” to my ideas list. I should maybe add “chance of winning the division” too because sometimes that is hard to tease out of the 1-4 seed numbers because of the 3-4 “flip if wc is from the same division as 1st seed” rule.

    Lorenzo, I agree about adding additional “algorithms.” Next season (hopefully this NFL season.) And maybe someday you will be able to uses the site to experiment with your own.

    Ken

    [Update: started working full time again around January. It's amazing how not paying off your credit card bill each month stretches out your bank account. :)]

  44. Mike H. Says:

    This is a great site, and thank you for doing it. I just wanted to point out one more thing: The Rockies are not out of the playoff race yet. If they go 2-0 and the Padres go 0-2, the Rockies will be tied with the Padres (and possibly the Mets and/or Phillies), putting them in a play-in situation.

  45. Mike H. Says:

    Oops, quick correction… if there’s a four-way tie, the Padres get the edge over the Rockies for the WC for some reason that I don’t really understand. But proof that the Rockies are still alive is provided by MLB.com:

    Here’s how it all shakes out:

    If there’s a four-way tie among the Mets, Phillies, D-Backs and Padres: The Mets and Phillies would play in Philly on Monday to crown the East champ while the Friars and Snakes go at it Monday in Arizona for the NL West crown. Monday’s losing teams would play Tuesday to decide the Wild Card; the Diamondbacks would have to go to Philly or New York. The Padres would either host the Phils or travel to New York.

    If there’s a three-way tie among the Padres, Mets and Phillies: The Mets and Phillies would play in Philly on Monday to crown the East champ. Monday’s losing team would play Tuesday to decide the Wild Card. The Padres would either host the Phils or travel to New York.

    If there’s a three-way tie among the Padres, Rockies and Mets: The three teams play a three-way playoff on Monday and Tuesday to decide the Wild Card. The Rockies hold all tie-breakers and get the choice of one on the road or two at home.

    Here are some other, more traditional scenarios:

    If the Mets and Phillies finish tied atop the East: A one-game playoff will be played Monday at Philly.

    If the D-Backs and Padres finish in a tie atop the West: A one-game playoff will be played Monday at Arizona, but only if a playoff spot is at stake

    If the Padres and Phillies finish in a tie for the Wild Card: A one-game playoff will be played Monday in San Diego.

    If the Padres and Rockies finish in a tie for the Wild Card: A one-game playoff will be played Monday at Colorado.

    If the Phillies and Rockies finish in a tie for the Wild Card: A one-game playoff will be played Monday at Colorado.

  46. Ken Roberts Says:

    Mike, Thanks. I’ll fix.
    [Update: fixed]

  47. Karl Says:

    Great site! I was recommending it to all my friends as we watched the Mets’ slide. Ugh.

    Nitpick: If I’m reading this right, your New England Patriots page shows that if NE loses its remaining 12 games, it still has a 51.4% chance to make the playoffs. Come on, the Jets, Bills, and Dolphins aren’t THAT bad, are they?

  48. Ken Roberts Says:

    Karl, thanks man. That can’t be right, let me see what’s up.
    [Update: I think I fixed the bug. I reran the numbers with just a few "coin flips", tonight I'll rerun them with a bunch. Thanks again Karl].
    [Update 2: Everything is working again. It is interesting that 2 "seasons" out of 10 million the Patriots win their division with a record of only 5-11.]

  49. MikeNoffke Says:

    Greetings,

    I like your web site. However, I am wondering why run 10 million random numbers? I suppose that is fun.

    Example: NFL
    When 2 teams play each other the result of that contest will have only 3 possible outcomes:

    W=win L=loss T=tie
    Team A WLT Team B WLT
    1) 100 010
    2) 010 100
    3) 001 001

    Since all 32 teams will play 16 games we can forget about the bye week.
    With 32 teams there will be 16 groups of 2 teams playing each other each week for 16 weeks.
    Thus the total number of possible win,loss and tie records is
    16 (2 team groups) * 3 possible out comes = 48 possible combinations each week.
    16 weeks times 48 possible combinations = 16 *48 = 768 total possible cominations for an entire 16 week season. Am I missing something?

    Thanks,
    Mike

  50. Mike Noffke Says:

    Never mind. I see the error of my ways the 16 pairs with 3 possible outcomes each is equivalent to a 16 digit trinary number. Thus, each week is 3^16 possible combinations that equals 43036721 possible outcomes each week. Thus, random numbers seems pretty good.

    Thanks,

    Mike

  51. MikedD1 Says:

    Ken,
    Love your site. Just curious, have you developed anything that would display championship results during the Petty and/or Earnhardt timeframes as to what the finishes would be if the chase were in play? Heck, even in the Gordon years leading up to the chase.

    Mike

  52. Ken Roberts Says:

    Thanks MikedD1,
    No I haven’t, but that’s a great idea, I’ll add it to the list. And likewise I could show this year without the chase.
    Ken

    Mike Noffke, thanks for the update. At some point I’ll add something about what you brought up on the “what is this?” page.

  53. Nascapper Says:

    Hi there, I am the webmaster at http://www.nascapper.com and would like to trade links with you, I can put a link to Sports Club Stats on my main page, in the left hand column under Nascar stats and info. Are you interested in trading links and increasing both of our google page ranks? Great site by the way, let me know!

  54. TBO Says:

    whats the deal? How come i can’t see any spreads on the saints?

  55. Ken Roberts Says:

    Nascapper: I’d be glad to. Thanks.
    TBO: Was the page blank? I don’t show betting spreads.

  56. Anwis Das Says:

    Why are you doing computer simulations for something that can be computed
    in closed form?

    There are about 8 teams that matter for the skins giving a total of 3^8 or ~500000
    combinations or 5 * 10^5. For three weeks, the total number of combinations is
    125 * 10^15 or 1.25 * 10^17. That’s piddly for a computer and you can find out
    the exact answer in a mere seconds.

  57. Ken Roberts Says:

    Hi Anwis, the computer has been doing it all season long, for hundreds of teams.

  58. Anwis Das Says:

    Hey Ken,

    But still, with about 6 weeks left for at least the NFL, it would be easy to get
    an exact answer for all 32 teams.

    Also, I would argue that for the first 10 weeks, for at least the NFL
    season, your numbers are as good as random because of your simplifying assumption that
    each game has an equal chance for all results. All predictions based on that will have high
    entropy, and so less meaning, imo

  59. Ken Says:

    Ken: Redskins @ Minnesota is now scheduled as a national game. 8:15pm ET Sunday 12/20. (Flex Scheduling).

    Regards,
    Ken R.

  60. Ken Roberts Says:

    Thanks for that Ken, I fixed,
    Ken

  61. Ron Says:

    Can you check your NFL tiebreakers? Yourresults after the Vikings-Bears game show that the Eagles have a non-zero chance of making the playoffs. That is incorrect. They have been eliminated. They can finish 8-8 at best. Best case scenario is that the Vikings lose both of their games and finish 8-8, which would also mean that the Redskins would be 8-8 (since Washington plays Minnesota). The Redskins would win the tiebreaker based on better conference record.

  62. Ken Roberts Says:

    Ron
    Let me take a look. Thanks for spelling it out.
    Ken

  63. Ken Roberts Says:

    Thank you Ron. I fixed the NFL tiebreaking bug you found. And you told me right where it was, my code was using division record instead of conference record.
    Since the Redskins and Eagles had equal division records it compared strength of victory ( and if that was tied it compared strength of schedule, and if that was tied it flipped a coin.) Somewhere along the line the Eagles where better than the Redskins.
    I’m rerunning next weeks big games now.

  64. Jimmy Says:

    Nice Work. Site is awesome.

  65. PlanetSteelers Says:

    This is a really unique site, and glad I visited it. I haven’t really seen anything like this before.

    Keep up the good work.

    Brought to you by: PlanetSteelers.com

  66. Phill McCrevis Says:

    Cool site dude!

  67. Ken Roberts Says:

    Thanks Phill, you and Ben should get together.

  68. TC Says:

    Awesome site! I can tell you have worked your butt off on this thing. Keep it up and you will be the next Jerry Palm.

    one thing that needs to be fixed…Tiebreakers! Carolina and Atlanta (NHL) should not have the same odds, as Carolina owns the tiebreaker for 3rd place. I don’t know how to fix it, but really, even though the record is the same, Carolina has a “phantom” 1 point (or half game) lead.

  69. Ken Roberts Says:

    Thanks TC.
    For the NHL I’m using the same tiebreaker rules the league uses, unless my code has bugs, which it always seems to have :)
    Right now I’m showing that Carolina has a .5% lead over Atlanta. Maybe that equals the odds that they will finish the season tied for the division title?
    Plus you’d have to factor in, to a lesser extent, who owns the tiebreaker with the Capitals, etc.
    Please yell if I’m missing something, and it still looks clear to you something is out of whack.
    Ken

  70. TC Says:

    Ken,
    you are correct. I overlooked the .5% because I was looking at odds of finishing in each position 3-8.

    Tough to work goal differential in though huh? It’s odd that NHL goes straight to Goal +/- after head-to-head instead of to divisional record etc.

    keep up the good work!

  71. Roger Says:

    WOW! After reading about all the “glitches” you have had to fix, if it were me doing this I would have given up LONG ago! lol
    Keep it up, I love the NASCAR stats. I did notice last year during the chase there wasn’t anyway to look at the race for #13 (1 million dollar prize for being top driver outside the chase at the end of the season.) Do you mind adding it to the ever growing list, right behind the “no chase” toggle. Thanks again!
    Roger

  72. Ken Roberts Says:

    Thank you Roger,

    Sometimes I let the glitches, or the lack of visitors, get me down. But surprisingly I’ve never entertained the thought of giving up. That is something I’ve had to learn (I’m 38), when I was younger it would have been much more of a struggle to persevere.

    If I’m following your suggestion, I think you where looking for this:
    http://www.sportsclubstats.com/2007/NASCARBottomDrivers.html
    That page was there last year once the chase started. I’ll make the link more visible.

    see ya,
    Ken

  73. GV Says:

    I really appreciate this website. I was wondering why in the NHL section, every team I looked at from the top of the division to the bottom was much more likely to go 16-1 than 1-16. I would think that they would be nearly identical. Any explanation for this?

  74. Ken Roberts Says:

    Thank you GV.
    Great question. I have to look into it to be sure, but here is some background info that might give you an idea of whats going on:

    I don’t show a line for each outcome I see because there are so many that it becomes harder to read, so I group the the records into “buckets” that are 2 points apart. And for the name of the “bucket” (what I show on the line) I use the record inside it with the most wins. That record typically has just 0 or 1 ot loss. Now, I also rig it that 0-17 and 17-0 are not buckets, but show only those exact records, because they are kind of interesting special cases.
    So, for example, a 10-7-0 “bucket” shows their odds when they finish
    out 10-7-0 or 9-6-2 or 8-5-4 etc… (20 points)
    and 9-7-1 or 8-6-3 or 7-5-5 etc… (19 points)
    It’s always the results that have the number of points in bucket “label” (what I show) and the results with 1 less point then the bucket label. I go one less instead of one more because it helps counteracts the rosy skew I think the chart has because of the 50/50 way I play future games.
    So, I show a row for exactly 34 points and a row for exactly 0 points, and in between I start from the top and work down, 33 or 32, 31 or 30, … 5 or 4, 3 or 2, 1 ( not 1 or 0, because 0 gets it’s own row). The 1 has no point buddy to add to its total. That is why they are not nearly identical. ( I think ).

    I’ll change it for tonights run to make the buckets 1 point instead of 2, It might be close enough to the end to make the extra lines usefully to see.
    [Edit: fell asleep at the switch, mañana]

    On a separate note, this is how I “play” a future NHL game:

    HomeScore = pick random number 0-3
    AwayScore = pick random number 0-3
    if HomeScore = AwayScore then they went to overtime, so I flip a coin
    to see who wins and add 1 or 2 goals to the winners score.
    That means in my simulation 25% of the games go to overtime. (which I hope is close to the real world).

    thanks again,
    Ken

  75. Ken Roberts Says:

    The NHL “what if” section is now broken down into 1 point buckets instead of 2.
    The remaining positive skew is due to that point you get for an overtime loss.
    For example, to earn 0 points total in all remaining games you have to lose every game in regulation. But to earn the max points you can win all remaining games in regulation OR overtime. So the simulation sees them max points outcome happen more often.

    Because 25 percent of my simulations games go to overtime, the expected value for points scored in a game is 1.125. Half the time you get 2 points, 3 out of 8 times you lose in regulation and get 0 points, and 1 out of 8 times you lose in OT and get 1 point.
    So the “normal curve” you see in the count column should be centered around:
    games left * 1.125 points, not .500 percent like every other sport.

  76. Brendan Gruen Says:

    Great site!

    One quick thing I think you should add in the Hockey section is a horizontal dotted red line between the 8th place team and the 9th place team, in the “Chance will make playoffs” section. I like the vertical red line between 8 and 9 in “Chance team will finish the regular season at position”. I think the cross section of the two red lines coming together would be great. Just a thought.

  77. Ken Roberts Says:

    Brendan, I think that is a great idea. Done.
    [Edit 3/20: I just noticed the line did not work in IE, fixed.]
    thanks,
    Ken

  78. shoelace203 Says:

    I just noticed something — it doesn’t look like you have it calculated where the division leaders in the NHL automatically get the first three seeds. I’m not sure if this really adds that much inaccuracy, but it might make some difference. I also don’t know how much code and memory it would take to compensate or even if it’s worth it. Still, it’s a good way to get a good, mathematically solid hypothesis as to what could happen. You’ve done a great job.

  79. Ken Roberts Says:

    Hi shoelace203, it handles that. Check out the What If section of Carolina, for example: http://www.sportsclubstats.com/NHL/Eastern/Southeast/Hurricanes.html
    3rd seed here we come, if we make the playoffs at all.

  80. Ali McG Says:

    This is a great site! I come here every day to check out the overnight impact on my teams and the various playoff races in the NHL and in soccer.

    One comment that I have is that your points total in Scotland is incorrect. Rangers actually have a 4 point lead over the 2nd place team.

    You have obviously put a great deal of work into the site and I share the URL whenever I can.

    Great site!

  81. shoelace203 Says:

    Okay, I see it now. Sorry, I was looking at teams in the Northwest, where all the teams (but one) are really close.

    Great site. Good job. I love it.

  82. Joe Says:

    Hey, Post these feedback comments from most reecent. Keep up the good work.

  83. BoKnows Says:

    Thanks for the hard work, Ken. I too will be visiting this site regularly. I am be moving out of the city and will be separated from my beloved sports teams. This site will allow me to see the standings and races as they develop throughout the year.

  84. Chris Says:

    What is the chance that you could also calculate standard deviation values for some of the percentages reported? Although given the battery of numbers already present, it might just clutter up the presentation of each page. Just a thought.

  85. Ken Roberts Says:

    Ali, thanks, I’ll fix Scotland. Thanks shoelace203. Thanks Joe, if I can figure out how. [Ed. done] Thanks BoKnows. Chris, my heads not working well enough to even begin to respond to you standard deviation idea :). Can someone help on how to compute?
    Thanks all,
    Ken

  86. Omega Says:

    First of all, congratulations on a great website with a great idea.

    Secondly, since you already have a lot of soccer and the 2008 season will be starting soon, can you add the MLS to this website? the league would fit perfectly for a website like this, because the final rush to the end last year was a mess with 4-5 teams having a realistic chance at the 8th spot in the final 2-3 weeks. if that happens again, there would probably be a lot of fans who could use this website to figure everything out.

  87. Frank Says:

    For the upcoming “Big games” in the NHL, would it be too much trouble to run scenarios where the losing team picks up a point due to OT or SO loss?

    Thanks

  88. Ken Roberts Says:

    Thanks Omega, I hope to get off me duff and add MLS.
    Frank, it runs those scenarios now. I don’t yet show all the numbers on the group pages, but you can see them on a team’s page. In the simulation 25 percent of the games are 3 point games.
    It’s not so obvious to see in the what if sections because I’ve been grouping them together by total points. This morning I turned off that grouping and fired off a 100,000,000 season simulation. So hopefully sometime this afternoon the what if sections will be updated with a line for every combination of w-l-ol record the simulation saw.
    Thanks,
    Ken

  89. Zach H. Says:

    Just wanted to say thanks for providing a very interesting site, and I’m excited to see the improvements you spoke of in the main page blog.

    I’ve always wondered how the odds might change if we weight the games different than 50/50.

    Cheers.

    - Zach

  90. drew w. Says:

    I noticed that you changed the NHL ‘games over .500′ chart to a ‘points over .500′. I understand why this was done, .500 being a difficult thing to compute in the NHL with the OTL point. (Is it a loss? Is it nothing?) What I’m wondering now is how you calculate ‘points over .500′. For example, you have the Flames sitting at 36 points over, but the two calculations I can think to run are: one point per game = .500, which places them at 13 points over; or actual points/potential points = points percentage: 89/150= 0.593

    Personally, I think the second calculation is the most accurate for reflecting a team’s performance in relation to other teams in the NHL.

    So, what calculation are you running to get a team’s points over .500?

  91. Ashley Says:

    Love the site looking forward to the new F1 season appearing at some point, very clever idea , working out chances of premiership relegation is particularly interesting.
    [Edit, Ken 4/03: F1 is up.]

  92. Ken Roberts Says:

    Drew, thanks for your brainstorming.
    I agree, the number in the graph is meaningless. What I did was for any day give a team 2 for win, 1 for ot loss, 0 for no game and -2 for loss. That is exactly what I want to see as far as the ups and downs of a team. But the number and the distance between lines and even the relative order of the lines gets out of whack.
    Your points/potential points idea is interesting. It might not be good because the changes are not “linear” over the season. Maybe points above/below the league average?
    Or I could just graph plain old points, but I like seeing the difference between a loss and no game.

    Thank you Ashley.

  93. drew w. Says:

    Points above and below league average could be good. Personally, I liked the old ‘games over .500′ chart as I’m one of those who advocate ditching the loser point, but I digress.

    I wonder how a points v. league average chart would look. Would it have the same problems of scale that a points/potential points chat would have? I guess your coding is probably good enough to tell the chart to recalculate the ‘average’ ever day … the issue then would be that you could see teams dropping on days off. If a team takes a saturday off when almost everyone else plays, then their points v. league average would fall, even though they didn’t play.

  94. Ken Roberts Says:

    Yea, good point Drew.

  95. Nabil Says:

    Wow, very good website for us hockey lovers. Saw it on a TSN webcast and your work is very well done, congratulations. Keep up the good work !!!

  96. Randy Says:

    http://watch.tsn.ca/nhl/clip40743#clip40743

    Congrats!

  97. Ken Roberts Says:

    Nabil and Randy,
    I can’t believe it was on SportsCentre!
    Thanks

  98. GV Says:

    I was looking at the Colorado Avalanche’s odds of making the playoffs today (March 29th) currently they are at 90% and the three games have a major influence. The best case scenario for each game is +7%, +4% and +2% respectively(rounded). The total best case scenario is listed as +13% which obviously can’t happen because that would put them at 103%. I understand how the +7 +4 and +2 can all be accurate but I’m not sure of how to interpret the +13. Thanks for all your good work, I check the site daily.

  99. Ken Roberts Says:

    GV, thanks.
    This post explains what’s going on, towards the bottom:
    http://www.sportsclubstats.com/?p=73
    I use to explain the issue in the little window that pops up when you hover over yesterdays big game total, but a while back that stopped working and I haven’t fixed it yet. Your right, it is big time confusing.
    Ken

  100. David Says:

    I love the site…I come here at least a few times a week (more frequently later in the season)…but the 2008 MLB season has started and it is not yet on your site. I am sure you are working on it, but there are those of us who can’t wait until you bring it up!

  101. Joe K. Says:

    Here’s a mention by Pierre LeBrun on the Rogers Sportsnet site: http://www.sportsnet.ca/hockey/2008/03/31/nhl_winding_down/ Congrats!

  102. Ken Roberts Says:

    Joe, thanks. Pierre is using “math geniuses” loosely :)
    David, 2008 MLB is finally up, thanks.

  103. Nathan Says:

    Something seems slightly wrong with the probability results for the game Blues at Predators tonight. It says that if the predators lose then the flames are “in”. But that isn’t correct, because if the Canucks win both of their games, including the one against the flames on Saturday, they will pass the flames on the basis of the season series between the teams.

  104. Nathan Says:

    Sorry, my bad, clearly the Canucks pass the Preds as well in that situation, and the Preds are “out”

    [Edit, Ken: No worries Nathan, thanks]

  105. Dackle Says:

    Hi Ken, what would you think about a shift to “Chances of winning World Series/Stanley Cup” etc in addition to the playoff odds? Sometimes a game is “big” because two teams are fighting for first overall in the conference and home ice advantage. It’s a big game, but not in terms of making the playoffs.

    Also, any chance of archiving the big games for an entire season? Would be nice to click on the Blue Jays and see a list of their biggest games of the season (something to watch during the winter). Even just on a league-wide level it would be nice to see a list of the five or 10 biggest games of the NHL season.

  106. Ken Roberts Says:

    Dackle, good good good, thanks. On the list. The past biggest games idea has been sitting on the list a long time, ever since you suggested it last year :), the league-wide list of past biggest games is a cool, easy, place to start.

    Chance of winning it all would be good too, especially once I add a “weighted” coin flip. But even with 50/50 I could copy the change in final seed big game numbers from the individual team pages up to the group pages, and maybe up the cutoff threshold to try and keep it from being too busy.

  107. Ali McG Says:

    Thanks for the updates to Scotland (March 18).

    I also wanted to mention that there are 38 games in the season in the Scottish Premier League. The season could extend to the later parts of May depending upon European results as well. Right now, the season shows as being complete.

    Thanks again for a great site!

  108. Scrig Says:

    Ken,

    I’m back again this baseball season. Great site, I’m enjoying the fine tuning you’ve done since last October. Keep it up…..I feel like you should be compensated for this somehow.

  109. Ken Roberts Says:

    Ali, I fixed Scotland again, thank you.
    I finally understand how the Scottish Premier League works. There are 12 teams and they play each other 3 times. Then they split into 2 groups, top 6 and bottom 6, and each teams plays the others in their group 1 more time. So each team plays 11 + 11 + 11 + 5 = 38 games. For next year I can build that logic into the simulation. [Update: next year already! Scotland works like it should now]
    I also added missing scores to Belgium, Turkey and Austria. Working on Greece and Denmark.

    Scrig is back! The sites traffic took a jump the last month of the NHL regular season. It seemed to be resonating with folks, a lot of return visitors and word of mouth visits from team forums and some Canadian press. Traffic is back down now, but if the general growth trend continues I imagine I’ll be able to earn money with (unobtrusive) advertising.

  110. roxrok Says:

    Just wanted to help out…you have DC United with one win too many (I think you might have unintentionally counted the game vs NY Red Bulls twice, as that result shows up twice in your list.)

    Thanks for doing what you do though…fun site!

  111. Ken Roberts Says:

    Thank you roxrok. MLS (and USL too) should be fixed. For now anyway :)

  112. Joe Says:

    Hey Ken!
    Wow! What a cool website! This is great, I can’t help but check all the MLB teams chances of making playoffs almost everyday. Recently I noticed you added a “weighted” calculation, I read about that and I understand how it works. But, it reminded me of something: the RPI (ratings percentage index). The RPI is used to help pick which teams go to the NCAA college basketball tournament. It’s a pretty cool calculation, taking in your winning % your opponents winning % and your oppponents oppponents winning %. If you want to read about it I found a website that explains it really well:

    http://kenpom.com/blog/index.php/weblog/rpi_help/
    http://kenpom.com/blog/index.php/weblog/an_rpi_with_road_rage/

    Thank You!

  113. Ken Roberts Says:

    Joe,
    I’ll run RPI on baseball (after I finish adding bike racing) and we can see how it looks.
    Thanks,
    Ken

  114. Ian Says:

    Hey Mike,

    great site – I just found your MLS section and love it. It really points out the importance of certain games.

    I have a couple of observations:

    1. There are 8 playoff spots. Why don’t the playoff percentages add up to 800% ?

    2. Looking at the win/loss/draw percencages in some of the mouse-overs, I notice that the frequency of draws is much lower than I would expect. Looking at the current MLS standings I see 35 draws from 117 games – about 30%. Your results seem to indicate about 10% draws.

    Keep up the good work.

    Ian

  115. Ian Says:

    Ken,

    Sorry about the name mistake.

    Ian

  116. Ian Says:

    Thanks Ken,

    I see my mistake – there are 3 points on the graph at ~39%, not just 2.

    It’s not my day. My mistake above means that there are 14 teams, not 13, and therefore 10 more games played than I originally counted. So the current draw rate is 35 from 127 games, i.e 27.5%. So your results are pretty accurate.

    Keep up the good work.

    Ian

  117. Ken Roberts Says:

    Ian, thanks, good questions.

    1. My calculator says 800.2, the .2 is just rounding error. What do you see?
    2. I don’t show a row for every result I see when simulating because I don’t want the list to be super long. So instead I group the records by points, and label it as the result in that group with the most wins. Soon I’ll switch to show every result I see. I’ll add a note about this in the “explain” links.

    When I “Play” a soccer game each team randomly gets 0,1,2, or 3 goals. (Except the 1000 times* I let them when out, where I give them 100 goals. Hey, it could happen. :) )
    So 25% of the simulated games are ties. (thanks for the 30% figure)

    Ken

    [Edit: *it's no longer 1,000. Instead it's some small percent of the total simulated seasons.]

  118. Nick Says:

    This is great stuff! Am I right in thinking that the Big Games predicts the effect on a team’s chances of only that game, and all other games (including other games that week) are simulated randomly? For example, the Thunder will be helped considerably if the Timbers lose to the Sounders this Thursday. They’ll also be helped if they win their own match against the Battery. But the magnitude of those effects are calculated separately for each game, right (and is not as simple as summing the two)?

    Also – I wonder if there’s a word to use besides “impact” in the Big Games section. Since Montreal’s USL team is called the Impact, it gets a little distracting!

  119. Ken Roberts Says:

    Nick, thanks for that post on USL Discussions.
    Yes (if I understand your question), the “impacts” in the big game section are calculated for that game only. Once a night, late enough for all that days games to be finished, my pc runs a program that looks on the web for new game results. If it finds any it runs the monte carlo simulation to find everybodies new current odds in. Then, for each upcoming game (the games next week plus any remaining games this week) it runs the monte carlo simulation 3 more times:
    1 pretend home always wins 1-0, do the rest random like normal
    2 pretend home always looses 0-1, do the rest random like normal
    3 pretend its a 0-0 draw, do the rest random like normal.
    after each run it records how each teams odds in differ from there current odds in.

    For each team, those are the 3 number I show for future games.
    When I show the impact of past games I say “ok, the home team actually won that game, so find the last time I computed big games before that game and show the “home always wins” difference.

    So you can’t straight add them up, but usually it is close. The total week change is the difference between the main monte carlo run after the last game of last week (or the last game of yesterday for sports with a lot of games each day) and the current main monte carlo run. I can’t just sum the individual game differences, I have to take them all as a whole.

    [Update: a better explanation:
    This is the correct value for the total change. It might not match the sum of the individual game changes above it because the individual impacts are calculated once before a days games are played. (Usually, sometimes it runs more often at seasons end). The earliest game's impact will be correct, but the latter games where calculated as if the earlier games outcome was unknown, when there "real" impact would need to take the earlier results into account. To take an extreme example, imagine it is the last day of the season and 2 game results can each knock you out of the playoffs, say take you from a 25% chance to a 0% chance. Loosing the first games makes the second game meaningless, but since I run the numbers the night before I show them both as possibly costing you 25%.
    This also effects the best and worst case totals because they are just the sum of the best or worst case for each game.
    The total is correct because I just take the teams current total chance and subtract their chance at the end of yesterday (or last week, depending on the sport).
    ]

  120. Nick Says:

    Say – is there a reason that the upcoming games are listed out-of-order? I would’ve thought they’d show up chronologically…

  121. Ken Roberts Says:

    Nick, the big games are ordered by importance. The sum of the absolute value of the impacts. So you can always see what the biggest game is for your team or league wide (on the league page). I’ll note that to the instructions. Thanks.

  122. Nick Says:

    Will you be running the simulation for this season of the English Championship League (which got underway last weekend)?

  123. Ken Roberts Says:

    Nick,
    Yes, thank you for the heads up. It’s good to hear it has an audience :)
    Check back Thursday.
    Ken

  124. Jim Says:

    Like the site….but I do have one question. In the NL, why do teams like the Mets have a very high percentage (58%) of coming in 3rd, whereas the Brewers have a very high percentage (80) of coming in second? I am not sure what you are attempting to show by the chance team will finish in this place for the season…..but I assume that 4th place would be the wild card slot regardless of their ranking for the overall record. If so….the Brwers should have a 80% chance of finishing in 4th place, as they will be the wildcard, and the D-Backs should be closing in on the third slot. If you are just showing rank by overall record and adjusting for the wildcard…then the Brwers should have the best percentage to finish in 2nd, not 3rd.

    Everything else looks great, but this is calculated wrong.

    As for the captcha, can’t you capture the text submital anyway, and if the captcha is wrong, generate the reentry page with the captured text already in the textarea?

  125. Ken Roberts Says:

    Jim,
    4th is the wildcard, unless 1st and 4th are from the same division (like the Cubs and the Brewers are now), then 4th swaps places with 3rd so they don’t play each other in the first playoff round. I think I should add a column for wildcard or something and explain it better on the page.

    thanks for pointing out the confusion and the CAPTCHA thoughts.

  126. Jim Says:

    Ken;

    The numbers look fine today….but they were different yesterday. It had the Mets as likely to finish in 3rd, not 2nd, and the Brewers to finish in 2nd, not 3rd. Check the copy of your probabilities that were posted yesterday.

  127. Ken Roberts Says:

    Jim, I looked through the last few days and they looked good too. I think one of us is sleep deprived :). Yell if you see more weirdness.

    [edit: and I just realized that I misread your original question]

  128. neil Says:

    Hello, love the site. Small thing…noticed that the system didn’t catch the double header results from the Chicago/Toronto games the other day.

    Toronto is 12 games above .500 not 11.

    Cheers.

  129. Ken Roberts Says:

    Thank you Neil, I’ll fix for tomorrow mornings numbers.

  130. Scott Snyder Says:

    I love your site and visit every day. I wanted to suggest, though, that there is an overlooked difference between fourth place in the league and the wildcard. Today’s win/loss scenarios have the Phillies in the playoffs if they finish fourth in the league — but, of course, since the NL West is a little bit pitiful, they will need to finish at least third to guarantee a playoff spot.

    Conversely, the Dodgers are almost certainly going to get into the playoffs and could end up ranked potentially as low as eighth in the league.

    I was also unsure how to reconcile the notion that the Phils were guaranteed a playoff spot if they go 11-1 or 10-2 the rest of the way. Since none of the three teams in the current playoff mix (Mets, Phils, Brewers) play each other any more, the Phils could go 10-2 and still lose out to perfect Mets and Brewers teams. Unlikely, of course, but I couldn’t reconcile the numbers.

    Again, tremendous resource — one of my favorite late-August and September websites. Thanks!

  131. Ken Roberts Says:

    [Update: this morning I reread Scott's message and it was clear as day. Sometimes I let my mind slip into slow mode, thinking about minor troubles, how the garage door just broke, I lost my glasses, I've got this to do and that to do... needless fog.

    The "Chance team will finish the regular season at position" section is smart, it include all the league logic involving division winners, wildcards, etc. It handles the tiebreaking rules to various degrees (I need to document this for each league, for baseball I think it handles everything but when 3 or more teams have to have a mini playoff tournament to decide who goes to the postseason). For example, in baseball if you have the 10th best record but still win your division you will correctly show up as finishing 3rd instead of 10th.

    It might be clearer if I said "Chance team will finish the regular season at SEED", but I'm not sure how universal the term seed is, plus it usually is only applied to the playoff seeds, not the teams that don't make it.
    End of Update]

    Scott, I am glad you like this enough to keep coming back.

    I’m not clear on your first question, could you ask it a different way if this does not help?:
    The 1-4 seeds are the playoff seeds, 3 division winners and 1 wildcard, but the wildcard is not always 4th because if the wildcard is from the same division as the best division winner the wildcard swaps places with the worst division winner. To start the playoffs the 1 seed plays the 4 seed and the 2 seed plays the 3 seed and they don’t want 1 and 4 to be from the same division. So, the odds I show for a team finishing 4th at the end of the regular season are the odds they will play the 1st seed in the first playoff round, which equals: (the odds they get wildcard and their divisions winner is not first) + (odds they win division AND are 3rd place among the 3 division winners AND the wildcard winner is from same division as the best division winner so they have move down to 4th and give the wildcard winner 3rd.)

    As for your second question, I see what you are saying and have 2 points to hopefully clarify:
    1. The “chance in playoffs” number is rounded to 1 decimal digit, 99.99 would show as 100.0. If I always see them make the playoffs I write “In” instead of 100.0. Likewise, when it is really 100% there will be no numbers (0 or otherwise, because 0 is just a small number that rounded down to 0) to the right of the red cut off line.
    2. Even when I write “In” they are not necessarily mathematically in. It just means in the millions of simulated seasons I never found a scenario where they did not make the playoffs. The case you gave of 2 teams both winning their last 10 games is a good example of an outcome the simulations likely could miss.

    Let me know if I’m not making sense, or if you have ideas for making the site clearer.
    Ken

  132. Alan Schmitt Mafra Says:

    Hey, man! Nice job, you’re doing great stuff! But I can’t understand how the Marlins’ game could affect the White Sox and Twins chance. It says +0.2 and -0.2%…

  133. Ken Roberts Says:

    Alan, your right, that’s a false positive. Sometimes they sneak in. I’ll up the number of seasons I simulate when computing the big games.

  134. Tim Says:

    Any way you can put the “biggest movers” daily post on a separate RSS feed from your normal blog posts?

  135. Ken Roberts Says:

    Good point Tim, thanks. I might not get to it for a week or so. I think I’ll make the daily highlights the home page and move the existing blog (without the new summary posts) to a separate page.
    [Update: done. The blog lives at blog.sportsclubstats.com now. If you use a reader it should hopefully automatically use the new address. I don't yet have a feed for the "recap" page.]

  136. Mike Lew Lamar Says:

    I was wondering about your odds for the AFC west. If San Diego goes 2-0 and Denver goes 0-2, then both would be tied, but you
    say that it would be a 50-50 odds for either team going to the playoffs. The first tie-breaker would be head-to-head, which each
    team would have won once. The next tie-breaker is record within the division, which SD would be 5-1 and Denver would be 3-3,
    giving the division to SD, as I understand it.

    I see that if you use the 50-50 method of probability, then for SD to get in they need to win vs Tampa, Denver needs to lose to Buffalo,
    and then SD would need to beat Denver. This would be a 1/8 chance for the Chargers, or 12.5%, which is pretty much what you say.

    Thanks for the great site, I use it a lot for all different sports.

  137. Mike Lew Lamar Says:

    I just figured out the answer to my own question, if SD goes 2-0, Denver could still go 1-1 and win the division.
    My bad.

  138. Ken Roberts Says:

    Mike,
    Thanks for the update. I know you are not the only person to be confused. I thought you where on to something myself when I started looking into it, before you explained it. It is easy to get confused when you try to make sense of 2 team’s what if numbers at the same time.

    I’m working on a super duper what if section that lets you hover your mouse over the numbers and see how your competitors had to have finished for you to get there.

    Keep writing when you see things that look wrong. Thanks.

  139. James Anderson Says:

    Re – your 03/24 post on the NHL loser point…

    I also think you should rework how you do the “points over .500″ graphs.

    Current: win = 2, otl = 1, no game = 0, loss = -2
    Suggestion: win = 1, otl = 0, no game = 0, loss = -1

    This is based on overtime losses being worth half as much as wins, so they are in essence “half-wins” (or ties). This would be equivalent to “games over .500″ – it would be (wins-losses), and (total points-games played).

    Just my thoughts. (keep up the good work!)

  140. Ken Roberts Says:

    James, that’s a better idea, thank you.
    I’ve done what you suggested.

  141. Travis Says:

    Hey Ken,

    Just wanted to say I really enjoy your site. I use it for all the leagues I am interested in and appreciate all the hard work you do.

    Was just wondering if you would ever consider doing a CFL (Canadian Football League) tab?

  142. Ken Roberts Says:

    Thank you Travis.

    I’d like to get off my rear and do every league in the world :)
    I’d like to make a way that lets you send in your daughter’s soccer league and have it show up as a tab. Or your friend’s fantasy league.
    My productivity has been less then great of late.

    I’ll move CFL towards the top of my list.

  143. Travis Says:

    That sounds great Ken!

    I know how much work must go into this site and I appreciate every second you put into it.

    Keep up the good work!

  144. Chris Palombi Says:

    Hey Ken! Thanks for leaving a comment on my blog; I was psyched out seeing you did! This site is incredible! The data, precise statistics (even for the effects of upcoming games) is top notch. After posting it on my blog, many friends and co-workers of mine checked out the site and made it the topic of conversations today.

    Thanks again!

  145. Greg Says:

    Quick question: Why are there no teams below .500 on the “Games above .500″ graph on the NCAA Men’s Basketball Standings page? I understand it says games ABOVE .500 (that’s what the graph is called) but on other pages it shows teams below .500 as well.
    Thanks

  146. Ken Roberts Says:

    Thanks Chris and Travis.
    Greg: It only graphs the top 20 teams (based on RPI rank). I’ll change the title, thanks.
    I’m not sure how many I should graph on that page.

  147. Edwards Says:

    How come in the NHL, all the teams have the most counts for sligtly over 0.500 records? If you are using 50/50, shouldnt there be highest probability to go exactly .500?

    for example, Canucks have a 14-9 remaining record happening 1,803,139 times, the highest of all records. Shouldn’t the 11-11-1 record, at only 999,453, be most often?

    thanks!

  148. Ken Roberts Says:

    Hey Edwards, good question. Does this answer your question?:
    http://blog.sportsclubstats.com/?page_id=6#comment-5324

  149. Kevin Says:

    Two questions, unrelated:

    1) When will the NASCAR stats start?
    2) Is there any way to see what the league percentages were for a certain week?

  150. Ken Roberts Says:

    Kevin,
    1. NASCAR: I don’t know, hopefully sometime next week. I still need to do some things to get it set up again.
    2. If you click on a graph to show it in “full screen” mode you can hover over the dots and the tooltip will show this historic numbers. When there are a lot of lines in the graph I tend to graph by week. On the teams page (when there is just 1 line to graph) I tend to show it by day.

  151. Kevin Says:

    What does “Total swing” on the front page, under Biggest Games, mean?

  152. Randy Says:

    Love the site but for the last couple of weeks the “Chance Will Make Playoffs” graphs for NHL teams seem to be messed up (e.g. Chicago Blackhawks graph shows the Hawks having around a 100% chance since week 3 of the season). It didn’t look like this until recently.

  153. Ken Roberts Says:

    Kevin, good, I need to explain total swing. I’m having a hard time expressing it in words, but here is what it means:
    total swing = sum of [ team's best case outcome chance in playoffs - team's worse case outcome chance in playoffs ] for all teams in the league.
    I’ll add that to the page.

    Randy, yes, I agree, it looks messed up. I’ll rerun the old numbers to fix it.
    This post sort of explains what is happening: http://blog.sportsclubstats.com/?p=216

    This is good feedback, thank you.

  154. Krach911 Says:

    I love this site. I’m a daily visitors. I play in a street hockey league and I want to use this system. How I can use this for my league ?

  155. nav Says:

    hey,
    the chance of making the playoffs for nhl teams is messed up i think. canucks won the game against columbus and went down 4.5 where as columbus lost to the canucks and only went down 0.2???

  156. mog Says:

    hey gotta agree, something is wrong with today’s calculations.

    Everything went the canucks way, they won in regulation, and the eastern teams beat the teams in the nucks confrence in regulation as well. vancouver should have gone up the max in chance to make the playoffs and instead they went down the max.

  157. wxray Says:

    Yes, something is wrong today, March 2nd. Maybe the southern snow storm glitched the servers. :-)

    Unlike the Canucks, nothing went the Cane’s way yesterday, yet their chances are way up. I wish it were so…

  158. Ken Roberts Says:

    What did I break this time :)
    Let me fix and get back to you.

    Update:
    I think its fixed now, thanks for letting me know things where amiss.

    Krach911, funny you should ask, I’m working on adding a new tab to the top called “You”.
    It will have spreadsheet templates you can download and fill in with your teams and schedule. Mail it to me, and auto-magically your site will show up on SportsClubStats just like the professional leagues.
    When you get new results you just add the scores to the spreadsheet and remail it.
    You’ll have to paypal me something. I don’t know what to charge, maybe $25 per league for the season.

  159. wxray Says:

    Thanks Ken. That looks better. Not for my team, but mathematically, it looks better. :-(

  160. Kevin Says:

    In soccer, what does “demoted” mean?

  161. Ken Roberts Says:

    Kevin, At the end of the season the bottom teams are “Demoted” (I’ve seen it called “Relegated” also) down to a lower league. The top teams are “Promoted” to a higher league if there is one.
    Here is a “World soccer for Americas 101″ post from a while back: http://blog.sportsclubstats.com/?p=60

    Clearly its time for me to put some explanation on the Recap page. :)

  162. Brian Burke Says:

    Great job, Ken. Love the gorgeous graphs in particular. I’ve been doing live in-game win probabilities for the NFL, and now NCAA basketball, but my graphs are just from the Google Charts API. I’d love to be able to do the great flash stuff you do, but unfortunately I’m just a self-taught plodder when it comes to coding.

    Brian

  163. Jeff Says:

    Hey Ken, is it possible to have instead of an ‘RPI Rank’ graph on the NCAA Basketball page, just a graph that shows varying team’s RPI as the season progresses?

  164. Ken Roberts Says:

    Jeff, I’ll swap it out tonight so we can see how it looks. Good feedback. I love feedback.
    [Edit: Not swapped out yet.]

    Brian, your live win probabilities are cool: http://wp.advancednflstats.com/bball/
    I think the graphs look great. Should the home be on the top? Is it that way in the score? I think home should be on the top for both.
    My graphs are pretty easy to do, I don’t have to code any flash, I just use this: http://www.maani.us/xml_charts/
    It’s cheap and elegant, you just add a bit of code to the page and point it at an xml file with the data. It has a “live update” feature to poll for the latest and redraw (and I bet it would redraw without the flicker.) But I don’t know how it could make them look any “better” then they are.

  165. Brian Burke Says:

    Thanks, Ken.

  166. victor79 Says:

    Hey All FLAMES Hockey Fans out there.

    I got good news for you. Fly your FLAMES flags with Calgary Pride.

    The FLAMES will WIN the Stanley Cup this year !

    Although they lost to the Blues tonight and the Canucks are biting on their heels, they will be the champions this year. Wait and see.

  167. Stephanie Says:

    I stumbled upon this site recently when I was lurking a hockey forum. What an absolutely incredible, useful, amazing site.

    Thank you, and thank you a million times over.

    Stephanie

  168. Dude Says:

    Great site, I use it all the time for the NHL. I was wondering if you could add a 3rd data set using the weighted method, but only using the last 10 – 15 games. This would give a better idea of how teams would fare based on how they have been playing.

  169. Ken Roberts Says:

    Thanks Dude, let me think about your idea. I’m torn between adding another dataset and just finding out a good amount of extra credence to give recent performance and adding that to the existing weighted dataset.

    Stephanie, what a cool compliment. I appreciate that.

  170. Stacey Ross Says:

    Amazing stuff for the stats and numbers junkies. And a well-constructed visual representation of those numbers with nice mouse-overs and useful charts. So good content AND a good interface: that’s rare. Gold star!

  171. Zack Heikkala Says:

    AMAZING WEBSITE! This site is amazing Ken! I can’t believe I just found it! I’m not gay, but if your ever in Portland I will buy you dinner.

  172. Aaron Says:

    Great site. Just one little thing. There are now 15 teams in the MLS. You’ve got fifteen teams sharing fourteen spots in your projections.

  173. Ken Roberts Says:

    Aaron: good catch. I think the numbers that are shown should be correct. I’ll add the missing column.

    Zack: I’ll see you in Portland.

    Stacey: Gold star :)

    Thanks all

  174. Jon Ciarlo Says:

    Just wondering why Ottawa is not shown as being 100% eliminated? Even if they win-out it shows 0.0% chance. Why not just say “out”? We lose the tie breaker to Montreal for head to head play so it’s official.

  175. Ken Roberts Says:

    Jon,
    Good question. It does not show Out because it still sees them make the playoffs 3 out of 50 million runs. I bet those 3 time where MULTIWAY ties. Even if they loose the tiebreak to Montreal, they win a tiebreaker between Montreal, them, and some other team (or teams). Does anyone see a case where that could occur? It’s either something interesting like that or it’s a bug :)

  176. don Says:

    you list the canucks as ‘in’ but it looks to me like it is still possible for them to finish 9th (extremely unlikely I agree) if St Louis, nashville and the ducks all win out. They don’t play each other so it looks possible to me.

  177. Ken Roberts Says:

    Your right Don,
    Sometimes you can jump to the 50/50 version to see if it catches a time they don’t make it in it’s 50 million runs. This time it did:
    http://www.sportsclubstats.com/NHL/Western/Northwest/Canucks2.html

    That is a limitation of the site. In and Out mean it never saw them make it. It’s not smart enough to know for sure.

  178. Ken Roberts Says:

    Update on Jon’s question: “Are the Senators eliminated?”
    http://blog.sportsclubstats.com/?p=270

  179. Mike Says:

    Great site, Ken!!

    Please consider adding World Cup qualifying to the football section of your site, then the WC tournament itself once it gets underway.

  180. Chad Says:

    Is there a chance that there could be a embedded code in the future? I love your website by the way.

  181. Kevin Y Says:

    The MLB season starts on Sunday. Will the first update be on Monday?

  182. Ken Roberts Says:

    Kevin Y, thank you for the reminder. I would have missed the start, again. I’ll try to have it up Saturday.

  183. Marc Brooks Says:

    Amazing site, thanks. Are you going to be tracking Formula 1 again this year? We’re one race in and I’m already jonesing.

  184. Mike Bendiksby Says:

    Thanks for an interesting and useful website. In the Scottish Premier League section you list the team ‘Dundee’ when it should be ‘Dundee United’. These are, in fact, two distnct teams, both with long histories.

    Mike

  185. Mark Says:

    Ken, awesome website! I’ve got a question about the Blues and Preds weighted odds of making the playoffs though. Both teams currently have 86 pts but the Preds hold the tiebreaker with more wins yet your site shows the Blues chances to be 52.1% with the Preds at 38.4%. This doesn’t seem to jive since the Blues will have to finish with more pts than the Preds to finish ahead of them.

  186. Ken Roberts Says:

    Mike, good, thanks, I fixed.
    Marc, I’ll start up F1 sometime soon. It will be crazy inaccurate until I get a weighted version up for racing.
    Kevin Y, I working on it now.
    Chad, would you explain “embedded code” some more?
    Other Mike, World Cup qualifying. That just might be a brilliant idea. I’ll look into if I can get the data I need for that. The late rounds of the last world cup was one of the first things I ran numbers for: http://playofftime.blogspot.com/
    Time flies.
    Thanks friends.

  187. Ken Roberts Says:

    Mark, do the Preds have a considerably harder schedule ( or, to a lesser extent, considerably more road games left)? That would cause what you are seeing. The 50/50 version shows you the odds with those 2 factors removed.

  188. Kevin Y Says:

    Got a couple questions about MLB. 1) What are PYT and RPI, and how are they calculated? 2) For the league page (http://www.sportsclubstats.com/MLB/American.html), why are all teams ranked #2? Is the rank the MLB rank, even on the AL page?

  189. Daniel Wood Says:

    Hi Ken,

    Sportsclubstats.com is fantastic. I have one small suggestion, why not include a probability of playing a certain opponent in the playoffs section? E.g., take the Vancouver Canucks right now, who are probably going to be playing the Hawks, and have a smaller chance of the Blue Jackets, and then a smaller chance still of playing somebody else. Thanks for putting in all the work that you do, I am checking your webpage religiously going down the stretch.

    Cheers,

    DW

  190. Ken Roberts Says:

    Daniel, great idea. I’ll do that (next season)

    Kevin Y, good, more explanation needed on the page. :)
    PYT is Pythagorean expectation*: http://en.wikipedia.org/wiki/Pythagorean_expectation
    I use it to help weight the teams for the weighted method.
    * Every league (except soccer) has there own tweaked version of the formula. For baseball PYT shows Pythagenpat. I have not found a good version for soccer, so it just uses the teams record for weighting.

    RPI is Rating Percentage Index. It ranks the teams based on record, opponents record, and opponents opponents record. It’s used in college basketball, here is a faq: http://www.collegerpi.com/rpifaq.html
    I’m showing RPI for kicks, it does not factor into the weighted method. The ranking are league wide (1-30 for MLB) regardless of which page I’m showing them. The first view days of the season lots of them are tied for the same rank.

    When I’m finished redoing the way I generate the pages I’ll start showing RPI on all leagues. And I’ll move it off to the far right because it’s not that interesting in leagues where everybody plays everybody else a bunch.

  191. Shaun Says:

    There’s something funky with the MLB projections. You are giving every team a 1 in 10,000 chance of winning out. Some kind of bug in the simulator…

    Shaun

  192. Ken Roberts Says:

    Hi Shaun, you found a “feature”:
    To help flush out each teams highest and lowest possible seeds, I force them to win or lose all their remaining games for 1,000 of the simulation runs.

    I do this for all the leagues. You can’t notice it in hockey now because there are so few games left.

    [Edit: it's no longer 1,000. Instead it's some small percent of the total simulated seasons.]

  193. Tim Says:

    Hi Ken,

    First of all, this is a great website… thanks for your efforts.

    Second of all, as a lifelong Toronto Maple Leafs fan (I know, its been a tough 42 years) all I have to cheer for this year is a lower finish, and therefore a better chance for a higher draft pick in the summer. With only one game left for most teams, the way I see it is if the Leafs lose to the Sens tonight and both of Phoenix and Los Angeles win tonight, all three teams will finish with 79 points. But the Leafs would have the least amount of wins among those three teams so wouldn’t they finish lowest of the three? Your model shows the Leafs finishing ahead of both other teams regardless of the outcomes and I was curious as to why.

    Thanks again,
    Tim

  194. Ken Roberts Says:

    Tim,
    Its doing that because the main league wide pages are not showing what you would expect them to show. The just order the teams by “chance in” (which is good, I think) and then by “average seed in conference” (which is irrelevant on the league page). I’ll eventually change that second sort to “average seed in league” like you would expect. Or something. I have to look at the lottery rules for the different leagues to figure this out. Like, for example, is a playoff team forced to get a worse draft pick than a non playoff team with more wins?
    Thanks

  195. Nick Says:

    Hi, Ken —

    Are you planning to follow the USL-1 season again this year?

  196. Ken Roberts Says:

    Heck yes Nick. Go RailHawks!
    I’ve got to do some other stuff first, do my taxes, crap like that. :)

    [Update. Taxes are done, and so is USL]

  197. Geoff Kraemer Says:

    I think there’s a bug. When you mouse-over an American League team, it doesn’t map correctly to the “games over .500″ graph (it does map correctly for the chance of making playoffs graph). FYI

  198. Ken Roberts Says:

    Thanks Geoff, I’ll fix.
    [Update: a proper fix will have to wait, but I changed it so that it is the weighted page that is always correct instead of the 50/50 page. Nobody looks at the 50/50 page. :) The bug happens whenever the "make playoffs" order of the teams on the page differs between weighted and 50/50. A real fix will happen when I redo the way the javascript figures out what to tell the chart to highlight. I've been wanting to do that anyway to add highlighting to the sidebar and list, and to try and make the highlighting work when you use google translate. Thanks again.]

  199. Ben Says:

    Just noticed that you’ve got a game coded incorrectly on the MLS page. On Wednesday, 5/6, RSL tied the Galaxy 2-2. You have it coded as an RSL win, 2-1. Thanks for the site–I love it!

  200. Ken Roberts Says:

    Thanks Ben. I fixed the score and the numbers are rerunning now.

  201. Wayne Says:

    I’ve been tinkering with Pythagorean Expectation for soccer, and from a quick Excel spreadsheet, it looks like using an exponent around 1.4, and multiplying the resulting value by 3 gives a fairly accurate point total for teams in the English Premier League.

    It would take more work to nail down the exact exponent, especially to make sure it works for all leagues.

  202. Ken Roberts Says:

    Wayne, thank you, all try out 1.4 for soccer.

  203. Kevin Y Says:

    For baseball, when you have all teams lose all their games at least 1,000 times, does that run with the other teams losing all 1,000 in the same simulations? In other words, would the simulations run for both the Dodgers and Nationals to each lose all their games 1,000 times (if they weren’t to play each other any more this season), thus putting the Nationals 16th and the Dodgers 15th in the NL?

  204. Ken Roberts Says:

    Kevin, nope, currently it never “rigs” more then 1 team at a time.
    For baseball, since there are 30 teams: 30 * (1000 win + 1000 loose) = 60,000 times I lock down one team’s games (and let the rest of the games be random) and the remaining 9,940,000 times every game is random.

    [Edit: it's no longer 1,000. Instead it's some small percent of the total simulated seasons.]

  205. Anonymoua Says:

    Hey Ken, it looks like the tables on the MLB pages of how recent games (a/e?)ffected a team and who to root for today are a little wonky. There are two columns for Wins and the Best and Worst labels are not really labeling anything.
    Thanks.

  206. Kevin Y Says:

    % Chance to make playoffs for MLS Soccer needs to be fixed. Currently it has all teams having the current percentage for all 16 weeks of the MLS season.

  207. Kevin Y Says:

    Actually, it’s the same for all leagues, the problem with that graph.

    Also, is there any sort of graph that shows how the average finish of all simulations changes over time?

  208. Ken Roberts Says:

    Things should be working now. I’ve been away from the site for a while, visiting family in Delaware (and checking out NY City with the kids) for a week, and then wallowing off in weeds for another week.

    The flat lines on the graph was a bug I introduced ripping out some unnecessary “week of game” distinctions in the code. The front page was wrong and baseball out of date because the a site I scrape scores from subtly changed their HTML. I’ll make the code email me if it sees that happen in the future.
    Anonymous, this may have caused what you saw, I’m not sure.

    Kevin Y, that would be so cool. I could put a slider under the league’s odds in list to let you fly through the season, all slots filled with the same number at the seasons start, shrinking down to a single diagonal line of 100′s at the end. What do you think? Is that what you had in mind?

    Also, your question about how I let the teams win/lose out got me thinking that I should do what you mentioned, being smarter in how I allocate the rigged seasons to better root out interesting cases. (All the while keeping it “fair” to all teams and not altering a teams total odds more then a tenth of a percent or so.) It’s on my todo list.

  209. Kevin Y Says:

    Yeah, a slider would be a cool idea. I just think that it’d be nice to see that also since every game impacts the % chance AND Avg. finish, so it’d just be cool to see how that changes, especially when you have teams like the Rockies go on those huge runs. Like right now, they’re the NL Wild Card leader, and nobody expected that a couple months ago when they fired Clint Hurtle. It’d be awesome to see how much they’ve really improved under Jim Tracy.

    Another note: the MLB page currently has it sorted by Win%, not by % Chance to make playoffs. Because of this, of the 8 teams above the line, 5 are AL teams; 3 are AL East teams, and none are NL Central. On one hand, that’s not the way it’s ever been, and it doesn’t show which 8 teams are currently in the playoffs (although we have ESPN for stuff like that). But that does get me thinking… would it be a good idea to have an option where we can sort by best record, and then see how the % Chance differs? For example, the Giants are 1.5 games ahead of the Cardinals. But even though the Giants have the better record, they’re 29.1%, and the Cardinals are 60.5% because of the division they play in, despite having a worse record. That’s quite a strange quirk indeed.

  210. Ken Roberts Says:

    I think I see. Adding graphs are easy. If I added a 3rd graph what would be coolest?:
    -Average final seed over time
    -Projected total wins (total points for soccer/hockey) over time
    -something else

    Your correct about the league sort. I need to tell y’all when I change things (and update the darn helps). Here was my thinking, for all sports where the “playoff level” is below the league level (sports that have conferences) I started sorting the league wide page by winning percentage (point percentage for soccer/hockey). For the reasons you talked about. Conference and division pages are still sorted by odds in and then average seed.

    I just realized If I do it this way I shouldn’t show a horizontal line on the league wide page! Thanks.

  211. Kevin Y Says:

    I think the Average final seed and Projected total wins graphs would basically be the same line, wouldn’t they? Just a different scale, since these are leagues where the seeding is determined by wins/points.

    But now that you bring that up, projected wins would be interesting if it’s so the team that wins Opening Day would be projected to go 162-0, and then the next day could bomb down to 81-81. I’d probably have to see a sample graph before I could say which I’d prefer more.

  212. bgix Says:

    Biggest flaw I see with the predictions is that is grossly underweights the chances of a draw on the final graph. The MLS is currently running something like 40% draws, yet for the remaining games, it looks like a draw rate of less than 20% is used.

  213. Ken Roberts Says:

    Bgix: Great points about MLS.
    Currently all soccer leagues use the same numbers (that I got from the English Premier League): 25.8% of games tie, and there is a 16% home field advantage.

    Using the numbers you emailed me:
    Home Team:
    W – D – L
    46% 34.3% 19.7%

    I’ll make 34.3% of games draw in MLS. I have to think about how I’m factoring in home field advantage a bit before I change that too.

    And more importantly, I’ll look into basing every league’s “end in tie”, “end in ot”, and “home field advantage” numbers on some kind of multi year rolling average, kind of like you mentioned in email.

    Question to all: It seems like if x% of games end in tie league wide, then for games between equally matched teams >x% of games would tie, and for best vs worse games <x% would tie. How do I scale x for each game based on the games log5 chance the home team will win? The site has a growing number of past years data to use for analysis.

  214. wxray Says:

    Hmmm. On MLB this morning, clicking on a team and examining the “chance to make playoffs” shows all scenarios as “Out” for all teams (at least in the NL). Looks like a glitch.

  215. Ken Roberts Says:

    Thanks wxray, it’s fixed.
    I broke it adding a feature inspired by bgix’s comment, so I think we should blame him. :)
    That “what if” section now only consolidates similar records when the total record count gets above 80ish. And then it consolidates as little as possible to get the count below 80. When a record really is just the most common of multiple consolidated records I put a * by it. For example:
    http://www.sportsclubstats.com/USA/USL/Whitecaps.html

  216. Chris Says:

    Hi Ken,

    Love the site, but is there a problem with the rss feeds? Specifically, http://www.sportsclubstats.com/MLB/National/East/Phillies_rss.xml doesn’t appear to have been updated since 5/26.

    Thanks for all the hard work!!

  217. Ken Roberts Says:

    Chris, yea, when I redid the way the site’s pages get generated I never got around to porting over the rss. I’m adding them back now. I trimmed back on some of the info in the feeds and spiffed up the formating.
    Thanks.

  218. Chris Says:

    Saw them come in today! Thanks so much :)

  219. Kevin Y Says:

    Will the English Football League Championship be starting soon? All teams have already played one match. Also, the Premier League starts this Saturday, so when will that first update be?

  220. Ken Roberts Says:

    Kevin Y, starting again already! I’ll probably get to it withing the week. Thanks for the heads up.

  221. Jake Says:

    Hey Ken,

    Love your site. Would it be possible to include an archive of all past predictions for a certain sport to see at what point / week of the season your predictions are nearly the exact outcome of the season? In other words, would there be a way to see when and how often you’re right?
    Basically, I’d like to know when your predictions are on average 51+% correct.

  222. Ken Roberts Says:

    Jake, Thanks.

    I can show an archive of all past predictions, (the graphs do that in a way), but I’m confused about what I would need to show to answer your type of question?

    This post sort of kind of tries to speak to how accurate the site is:
    http://blog.sportsclubstats.com/?p=296 (without really answering it)

  223. Kevin Y Says:

    On the English Football League Championship page, there is a green line separating the top-2 teams, which gain automatic promotion to the Premier League. Teams #3-6 have a playoff to determine the third team to be promoted, I know, but what does the red line after the 3rd-place team indicate?

    And darn, Jake asked my other question before I got the chance to. But that’d be something I’d like to see.

  224. Ken Roberts Says:

    Those are the teams that get demoted :<
    Thanks for pointing that out, I’ll fix.

  225. Jake Says:

    As of now, my question doesn’t make a lot of sense. I guess what I’m asking for is for you to make a prediction, let’s say for MLB, every week. Maybe to make that prediction you take the teams with the highest “chance of getting it to the playoffs” but you pick one team form each division and a wildcard and you say ‘I believe these teams will make the playoffs.’ Then, after the regular season is over, look back on the predictions you made each week and judge about at what point in the season your predictions were true. If you can do this for may years, you can create a stat that tells on average what week you are usually x% correct.
    Hopefully that clarified things.
    Thanks again!

  226. Ken Roberts Says:

    Yes, that’s clear as a bell Jake. Thank you for that idea.

  227. Kevin Y Says:

    Will Premier League be starting soon?

  228. Patrick W Says:

    I’m with Kevin Y. Looking forward to your EPL models.

  229. Allison Andrews Says:

    Your USL page seems to be missing one game, the 2-1 win by the Charleston Battery over the Vancouver Whitecaps on August 7, 2009.

    But well done on the site, I enjoy looking at how different matches affect the probabilities.

  230. Ian Says:

    Serie A has begun, you know…

  231. TurfToe Says:

    Fabulous site! Very cool to see how things progress. The CFL has not been updated to include last weeks games (Calgary 23-20 Toronto; Hamilton 30-31 Edmonton). Is there a way to contact the ‘admin’ of that league or to submit results if that admin isn’t around?

  232. Ken Roberts Says:

    Thanks all.
    Saturday will be “add all the new schedules and fire off the new seasons” day for me. I’m liable to get sidetracked changing the program to make the process more automatic or something, so it will likely carry over to Sunday.

    Allison, thanks for pointing out the missed game, it’s fixed. (The historic data in the odds in graphs between the missed game and when I fixed it is still wrong, but all the current numbers are correct).

    TurfToe, that is the million dollar question.
    I have not figured out how to approach it. Ideally everyone could add scores, (and Allison could fix the USL mistake), with some way to undo/guard against malicious folks entering crap. For now, I added your email address to the list of CFL admins, so you can add those scores if you want (I hope you do).

  233. Patrick Says:

    Your MLB page is missing one game between the Texas Rangers and the Toronto Blue Jays. There was a double header on September 1 between those two teams, and the Rangers won both games 5-2. You only counted one of those two games. The Rangers should be 76-59, and the Blue Jays should be 60-75. All the other MLB teams are correct.

  234. Ken Roberts Says:

    Thank you Patrick, I’m rerunning the numbers
    Evidently my “screen scrape” code gets confused when both games in a double header have the same score.

  235. Kevin Y Says:

    I don’t know exactly how to phrase this, but I’ll try: in MLB, it seems virtually certain that the Angels and Red Sox are gonna meet in the ALDS… again. For Angels fans such as myself, this isn’t something we’re looking forward to. But I wonder if it’s possible to show the chances of a particular playoff matchup to occur. In other words, I’d like to know if we have any hopes of facing another team in the postseason.

    Hope my question is understandable enough.

  236. HH Says:

    S league does not have demotion, pls review

  237. sidereal Says:

    The EPL simulation seems to have given each team exactly 1,250 instances of a perfect record, which seems pretty unlikely. The rest of the outcomes are normally distributed.

  238. Ken Roberts Says:

    @sidereal, that’s a “feature” :)

    @HH, thanks for the heads up.

    @Kevin Y, good idea. I’ll add a table (next season :<) showing all first round match up probabilities.

  239. sidereal Says:

    @Ken

    Ah, that makes sense. Maybe it’d be an improvement to run the best/worst case simulation separate from the main simulation, and only use its output to determine whether a team is mathematically out (so it can have the ‘Out’ attribute, dashes instead of 0′s on the grid, etc), but not populate the probabilities with it.

    Also, the MLS Quakes/Rapids game from last night isn’t in. I’m impatient and shrill! I’m willing to help enter MLS data, if that’s supported :).

  240. Ken Roberts Says:

    @sidereal, that is a good idea.
    And thanks you for offering to help. Let me think on that.
    In the mean time, I always appreciate the ping when you find a missing score.

  241. Paul Says:

    so the long skate has started, …. looking forward to the nhl updates ….
    thanks (as always, a great site)
    Paul

  242. Ken Roberts Says:

    Thanks Paul. I just fired them off.

  243. amoeba Says:

    How come it says the Rockies can’t finish in fourth in the league? If they lose both games, and St. Louis and Philadephia win, then they are fourth? Maybe I don’t understand what the positions mean.

    BTW I love the site!!!!

  244. Ken Roberts Says:

    @amoeba: The wildcard (normally the 4th seed) gets the 3rd seed when they are in the same division as the 1st seed. (3 and 4 swap so 2 division teams don’t play each other in the first round of the playoffs).
    Does that explain what you are seeing?

  245. Kevin Y Says:

    For NHL teams, what does “This row combines multiple infrequent records” mean? It doesn’t make any sense to me. I just know it has something to do with the OTL, but other than that I’m clueless.

    Go Kings!

  246. Kevin Y Says:

    Also, would there be any interest in keeping track of minor leagues also? I know you probably don’t want to overload the site with a thousand different leagues globally, but like in England you have Premier League, the Championship, and Leagues 1 and 2. Granted, those are relevant due to promotion and relegation, but why not the AHL as well?

  247. sidereal Says:

    Hey, Ken. It’s not a huge thing, but right now NHL games are dominating the ‘biggest games’ stats, even though we all know the 3rd and 4th games of the season really aren’t that big. I’m guessing it’s because it’s using the weighted predictions, and so a 2-1 team is being given a ton more wins down the road in the simulation than a 1-2 team. It might be worth pushing those teams back towards .500 somehow. A hacky way would be to just tack on 20 games worth of a .500 season at the beginning, to dilute the effect of early games. A smarter way would be to weight in last season’s record during the first part of the season.

  248. Joel Says:

    Is it just me, or has the site not been updated in 3 or 4 days now?

  249. Kevin Y Says:

    No, it’s not just you. Please come back soon!!

  250. Paul Says:

    An issue is up on the MLS listing that throws great doubt on the accuracy of a system that the statistician in me wants to believe in. There is only 1 game left in the season for the relevant teams. You show two teams at 39 points (DC United and New England Revolution) having a small probability of finishing in 4th place. 4th place is mathematically impossible. The maximum points for both of those teams is 42 and there are 5 teams that already have 44 or more points, therefore out of reach. The absolute best that either of these teams can do is 6th place.

    I want to recommend you site to sports and stats-minded friends, but I can’t when I see that kind of problem.

  251. Ken Roberts Says:

    Hello all.
    @Joel, @Kevin Y, @sidereal, I was off in the weeds a few days/weeks. I’m back. Thanks for asking.

    @Paul, I think you’ve been misled by my poor presentation of the MLS numbers. Let me know if this does not explain what your seeing:

    Major League Soccer playoff seeding works a bit differently than other sports, and for now I’ve just shoehorned the numbers into my existing infrastructure. Here is how to read:

    * 1 = chance of winning Supporters’ Shield
    * 1 + 2 = chance of getting 1st seed in conference
    * 3 + 4 = chance of getting 2nd seed in conference

    The first number in each pair means the team was better than the team with that seed from the other conference, which is meaningless for playoff seeding but that’s where it comes from.

    * 5 – 8 = wildcard seeds
    * 9 – 14 = did not make playoffs

    Buenas noches

  252. Paul Says:

    Thanks for the explanation regarding MLS. It does make more sense now. Great site, I recommend it widely.

  253. Erwin Says:

    One of my daily automatic hits. I love how this helps me follow the NHL, and track the bigger picture of upcoming games! (GO CANUCKS!… *sigh*)

    Anyway, here’s something I was thinking about when checking out the various teams. I noticed that playoff chance rises and falls slightly even on the days you don’t play. I wonder how much of that could be attributed to “destiny,” having your fate even slightly determined by the other teams. I’m not even sure if this topic has been discussed early on this feedback page.

    I think tracking the overall seasonal effect during non-game days would be interesting near the end of the season, should a few teams come up close in rankings come playoff time, one being prodded ever-so-slightly higher by the chance effect of other teams losses bumping up their chances. I understand this effect is persistent even on game days, and is affected by your ranking (higher rank would tend to have more outside games push your playoff chances higher anyways: “creating your own destiny?”)

    Would tracking this along the season become a moot point come playoff time? I can’t wrap my head around the numbers enough to see if every team ends up with the same result: “destiny is affected by your own standings”.

    Excellent site,
    E.A.

  254. Ken Roberts Says:

    Thanks folks for all this feedback.

    @Kevin Y, “This row combines multiple infrequent records” means it saw lots of different w-l-ol records that got us to that point total, but I don’t want to show all of them because it’s info overload, so I combine all the times I saw that point total and just label it with the most common record that got us there. As the season winds down and there are fewer possible point totals it progressively stops combining records. If I can’t explain it better I might be better off just not noting it because all those *s get annoying.

    And, @Kevin Y, I absolutely want to overload the site! I’d be lucky to have that problem. I’ve set it up so you can add leagues by sending in the schedule and score updates. The “Your League” tab sort of kind of explains how. In the email you send in you say something like (from Russian Hockey that Juris added):
    Kind: pro
    Sport: hockey
    Gender: male
    Country: Russia
    Level: A
    Season: 2009

    So for AHL it would be:
    Kind: pro
    Sport: hockey
    Gender: male
    Country: USA
    Level: B
    Season: 2009-2010

    When Kind is pro, college, or high school it automatically shows up in the proper place on the site.

    @sidereal, It does weight the record less, “regressing towards the mean”, a bunch to start and less and less as the season progresses. You can see the actual weight number used in the sim by hovering over the Pythag column. The math I use has room for improvement, I may not be doing it enough to start. But I suspect, because even the 50/50 early games have a big impact, that early games have a bigger impact than one would expect. It sort of makes sense to me when I look at it backwards, from the end of the season. What percent of teams that missed the playoffs would be in if you “spotted them 2 points?” It seems like there is often a few on the edge, and that percentage is related some how to how much your odds go up if you win your first game.
    I’m currently not using your idea of weighting in last seasons records at the start. Some day maybe.

    @Paul, thanks. Speaking of bugs in the site, yesterday my Uncle Lance pointed out that the NFL number for last season show Tampa Bay making the playoffs instead of Philadelphia! I’ll have to see what the heck is up with that. I don’t know if it was broken back then or if I recently broke it. And I don’t know yet if the bug impacts other leagues.
    [Update: I still don't know why it was broken before, but rerunning the number gave the correct results.]

    @Erwin, we’ll have to ponder that one. I think you are on to something interesting.

  255. Kevin Y Says:

    Yep, that’s clear as a whistle (the “multiple infrequent records” thing). If only you could put that on the pages instead, but that’d take too much space.

  256. Ken Roberts Says:

    Thanks to Mr. Kevin Y, AHL is in the house!

  257. Kevin Y Says:

    Ever thought about an option where you could see the team’s full season schedule on their team page? That way you don’t have to go to the league sites or other stats sites.

  258. Question Says:

    In week 11-1-2009 NFL….how can the 49ers beating the colts be bad for the Houston Texans? The percentage has the texans minus if the 49ers win. They are in the NFC. How can them beating the colts be bad for the texans in any way?
    I read the explain this but i didn’t really see why that game was a minus for the Texans chances. Could you please explain why this percentage came out negative? Seems like it would be positive for the 49ers to beat someone in our division that we have beaten already (we = Texans team)

    Thanks!

    Confused

  259. Ken Roberts Says:

    @Kevin Y, that is a good idea, Dackle has been asking me about something like that for years.

    @Question, very interesting. I’ll get some coffee and think about that.

  260. Question Says:

    @Question, very interesting. I’ll get some coffee and think about that.

    Thanks for the quick response. You may need a whole pot of coffee to figure this one out. Ha! Me and my Texan buddies have been pondering it all week!. ps. Love your site. Been visiting for 2 seasons and i really like the concept.

  261. Unknown Says:

    Why are almost all NHL teams’ pythag value in the red?

  262. Kevin Y Says:

    This might be completely irrelevant and just plain pointless, but for the English Premier League, for Wolverhampton Wanderers, they are usually referred to as “Wolves”, not “Wolvers”. Just thought I’d bring that up since that kinda was bothering me a bit :)

  263. Neo Says:

    Great website, I really like this. I’m checking my team after every match.

    One small bug, in the English league one Southampton were docked 10 points at the start of the season for enterign administration, but this isn’t in the table.

  264. Ken Roberts Says:

    Thank You!
    @Neo, I added the 10 point penalty.
    @Kevin Y, Wolves it is
    @Unknown, they are almost all red because something is busted. :) I’ll take a look.

    @Question, I’m still unsure about the 49ers-Colts game affect on the Texans:

    Colts -1.4 (-0.11) 49ers -0.9 (-0.04)

    Here are the Texans raw numbers for that game:
    If Colts win: top 6 seeds = 1486,4873, 4959, 4581, 104914, 123391 = 244204
    better chance to win division
    If 49ers win: top 6 seeds = 2299,9046,11734,14510, 93808, 117905 = 249302
    better chance to win wildcard

    And I’m confused about a the idea of both outcomes being bad or good in general. It seems the current odds in would adjust to keep that from happening.
    Anyway, I bet one key is that when I assume the game outcome for the “big game” section, before I rerun the numbers I recompute the teams “strengths”. But I don’t recompute strengths after normal simulated games. Or something.

  265. Question Says:

    Thanks. I still didn’t understand the raw numbers part. I’m glad you agree that the outcome should not be either both positive or both negative for a particular game.

    So the raw numbers part says it is a better chance to win the division. Do you mean for the Colts? Because the colts winning certainly is not creating a better chance to win division for the Texans. It’s all very confusing. Ha! I’m glad you understand it though. Coffee good! Thanks again for taking a look at it and trying to figure out why the numers both came out negative. The 49er’s winning should have been a positive number. Of course none of us thought the 49ers would actually win that game anyway, but we like to try to root for the right teams each week. Your response is much appreciated.

  266. Question Says:

    Oh, i forgot to let you know….this week the Miami and Patriots game both came out negative for the Texans this week. And the Bengals Ravens game…Ravens negative- Bengals came out zero. So both of those match ups are basically bad for the Texans no matter who wins this week. Maybe if you figure out why this happens , your mathematical formula might work the way we think it should work.

    But then again, it sounds pretty complicated. You might not ever find a “formula” that renders the numbers correctly. There are a lot of variables when there is still half the season to play out. But check it out. This week nfl Texans playoff chances- Patriots game and Bengals game have odd numbers for the Texans chances. I’d think the Bengals winning would be good, And Miami winning would be good for the Texans. But thats just a hunch.

  267. Jeremy Says:

    you haven’t updated the NHL in days

  268. Ken Roberts Says:

    @Jeremy, Dohhhh! thanks, I broke the auto run thing. I’ll fix and run now.

  269. Justin Says:

    I was just having a peek at the Australian A-League football stats to see if Wellington are likely to qualify for the playoffs/finals and I noticed that you’ve set the playoffs as a top 4. This season the league expanded and the playoffs have now changed to a top 6. Is it possible for you to re-run the stats with a top 6 so I can take heart that Wellington’s chances are on the up?

    Cheers

    Justin

  270. Frootmig Says:

    Seems that the probability numbers are not being calculated for NFL, NHL & NBA. Results seem to be up to date.

  271. Ken Roberts Says:

    @Frootmig, thank you, It is fixed now.
    I revamped the program to make it faster and broke it along the way.

    @Justin, cool, I changed from 4 to 6. Wellington up from 19.2 to 45.5!
    Listen, if you want to help us enter scores for that league just email me and I’ll add you to the list of “updaters”

    @Question, about the big game weirdness: One of the things I did to make the program faster (see above) was change how I compute big games. Instead of re running the sim for each of tomorrows game outcomes I just keep track of each outcome’s effect during the main sim run. This could make the weirdness go away (because all the calcs are based off the same pythag team strength numbers). We will see. Keep letting me know what you find.

  272. Unknown Says:

    It would be kind of cool if the odds of winning of both teams were shown for the next game (thinking about NHL)

    thanks, i love that stuff

  273. Ken Roberts Says:

    @Unknown, I think that is a great idea. I’ll add that. Thanks

  274. Max Says:

    Thank you! Your work is highly appreciated!!

  275. Question Says:

    Well, looking at the next few games it seems to have fixed the numbers. For this week anyway.

  276. Unknown Says:

    I wonder, what kind of hardware are you using? How much time does it require to do the simluations and how many simulations do you do each day (generally) ?

    thanks

  277. Ken Roberts Says:

    @Question, cool, thanks for looking.

    @Unknown, the program runs on my home PC, an Intel Q6600 (2.4GHz, 4 cores). The 50/50 and weighted versions run in parallel. Last night the NHL took 28 minutes to tally 10 million seasons, about 6 thousand a second. It took 8 hours to run 100 million seasons for the NFL, about 3500 a second. The NFL tie breaking code is much slower than the NHL’s.
    There are 70ish leagues, but most do not have new results on any particular day.
    A short while back I rewrote the “big games” logic and that made everything maybe 4 times faster, so I’ve had time to run each league multiple times a day.

    I can still use my pc when all this is going on because the sim runs at a lower operating system priority. That’s a fancy way of saying “Windows, make sure my web surfing gets all the CPU cycles it needs, the sim can wait. I have to see this You Tube vid now, dammit!

  278. Bug Says:

    Hey man love the site. Recently I’ve noticed a bug where the graph for “Chance of Making the Playoffs is messed up, at least for the NHL. If you just look at this graph before clicking it: http://www.sportsclubstats.com/NHL/Eastern/Southeast/Capitals.html, it looks like the Capitals have had less and less of a chance of making the playoffs (thought they’ve recently had a 6-game win streak and lead the league in points.) Upon going full screen however it says they have a ~100% chance. Problem exists on both the 50/50 and weighted methods.

  279. Ken Roberts Says:

    Thanks @Bug for the heads up. It’s fixed now. The bug was causing the small odds in graphs to show a team’s odds at each DAY instead of each WEEK, even though the label says week. (Evidently the Capitals did not have a stellar first 12 days of the season.)

    I added the bug with a careless copy and paste as I was speeding up the code.

  280. Tom Says:

    Any thoughts on running simulations for the 2010 Olympic Hockey tournament?? Not sure how your simulations would handicap games before the tournament, and I don’t know if the tournament would be too short to produce statistically valid results.

  281. Jeremy Says:

    Hi, just had a question… I think theres a possibility that all your NHL code is out of whack because on the main page for nhl it compares the graphs of games over 500 to playoff chances… in the nhl games over 500 does not determine playoffs.. its on a points system like soccer. Just wanted to let you know

  282. Question Says:

    I just thought i’d come back and point out another mistake in the mathematical formula. lol

    Don’t hate me..Ha!

    Last week the Houston Texans were formula-ed out at zero percent chance to make the playoffs. That may be true considering their chance of beatign the patriots the last game and getting help from ravens,broncos,or jets. However, the zero percent chance that was listed last week was wrong. Maybe you can figure out how come this came out zero when obviously they still have a chance that should hav ebeen above zero.

    Great site. Sorry i always come in and post about the bad things i see. 99 percent of your site is awesome! Thats a pretty good number.

  283. Ken Roberts Says:

    @Question, I never saw the Texans at 0% change of making playoffs. At the END of each day they got as low as .43%:
    http://www.sportsclubstats.com/NFL/AFC/South/Texans_ChanceWillMakePlayoffs.html
    They may have been lower in the middle of some day. It may have shown 0.0 but really been 0.0499999 or something, that rounds to 0.0. When I never see them make it I wright “Out”.
    If you saw “Out” then I bet what happened is you looked at the site right after some games finished and saw a 1 million season sim. What I’ve been doing for the NFL is fire off a quick million whenever I see new results, so the site update just minutes after the game ends. And 1 million might have missed the times the Texans got in. Then I bump it up to say 5 million, 10 million, etc until I get new results and start over. This week should hit a billion before Sunday. Anyway, my hunch is that’s what you saw. No matter how many seasons I simulate the site will never be able to “prove” they are in or out. It can only prove they are “not in” or “not out” (by finding examples).
    A good example of “wrong” things NFL has been showing when I’ve run less then 20 million or so seasons is tie outcome games showing up as “big games” when they should not, caused by seeing so few ties. I can filter most of those false positives out but I have not gotten around to it yet.

    @Jeremy, I graph “Games Above .500″ instead of points because I want to see the ups and downs. I want to see the loses as a downward line rather then a flat line. Even though, like you say, it’s not “correct”, and the numbers on the Y axis are kind of meaningless. But don’t worry, the way I graph it does not affect the other results.

    @Tom, I haven’t though about Olympic Hockey. Do they have a “group” stage before single elimination? If so you could add it yourself, like the
    Jonatan added the World Cup:
    http://www.sportsclubstats.com/WorldCup.html
    I agree about the rankings, but even just looking at 50/50 would be useful if it is not single elimination. I don’t yet handle a single elimination type tournaments, I always need to know who all the opponents are for all the games.

  284. Larry Green Says:

    You mention that teams in international football in the top league will not be able to be promoted, as there is nowhere to go to, but there are various forms of Champions leagues. For example, in the English Premier League, the top 4 teams qualify for the UEFA Champions League the next season. It would be good to see a % chance of qualifying for this, as the monetary difference is HUGE making CL, and especially this year the race is wide open in the EPL, and you could use the promotion code to do this.

  285. Question Says:

    Okay you are right. It was 0.04 percent that week. I would like to suggest you round up to 1% since the round off to zero is misleading. I get it now i wasn’t looking at the entire percentage number and just seeing the 0% at the top of page. Well. next week i guess we will see whether the Texans at .04% make the playoffs or not. Thanks for the great site.

  286. Ken Roberts Says:

    @Question, thanks, at the top of the page I want to show more digits, like “odds up 0.0003 to 0.0054%” I’ll keep an eye on that.

    @Larry, the number are here: http://www.sportsclubstats.com/UEFA/Qualifying.html,
    and that is a great idea, I should move the numbers to the promotion column of the main graph.

  287. Larry Green Says:

    I see it now, although it’s almost 3 months old, so not useful. Maybe that chart could use the results from the various leagues for it’s dataset?

  288. Ken Roberts Says:

    Larry, 3 months old! Dohhhh! I’ll fix and add to league pages. Thanks for the heads up.

  289. Larry Green Says:

    No problem. I absolutely love this site, btw.

  290. Stace Says:

    Ken, your site is awesome but I’m seeing something in the NHL standings that I don’t understand. Right now (~1:30pm PT 1/11/2010) I see 9 teams in the Western Conference with a greater than .500 chance of making the playoffs. But the playoffs only takes 8 teams. The 9th-place Detroit Red Wings are listed as having a .664 chance of making the playoffs. In order to make the playoffs they must supplant one of the teams above them. But every team above them has a greater than .700 chance. I’m not saying this is wrong, but I’m having a hard time understanding how that can be right. Thanks!

  291. Ken Roberts Says:

    Stace, thanks, good question. Think of it this way, there is a 66.4% chance that _ONE_ of the top 8 teams will be replaced by the 9th place Red Wings. But the top 8 teams are sharing that risk. Say you are a gazelle in the herd, there is a 66.4% chance the cheetah is going to catch somebody, but as long as you don’t spaz out and fall it probably won’t be you.

    As a sanity check you could add up all 15 team chances. They should add up to 800%. Every point below 100 for a top 8 teams is a point above 0 shared among the bottom 7. And right know the Red Wings hold 66.4 of them.

  292. Stace Says:

    Got it. As soon as you put it in terms of something eating something else, I understood immediately. ;) Good work, keep it up! I’m a Dallas Stars fan and watching their playoff chances plummet with each passing day is like watching a train wreck; I can’t tear my eyes away from it. Can you do something to prop them up a bit, for the purposes of depressed fans such as myself? :)

  293. Alex Says:

    First off, Ken, I would like to say that this is a great site. I have only one small suggestion. It would be nice to see a summary of all the records that a team got in the “future” and not simply just the records that made the playoffs or a specific position. This would enable you to see the most likely record that a team will get down the stretch. Technically this is already possible by adding up all the records from the individual positions but that is highly tedious. Once again thanks for supplying this awesome site.

  294. Ken Roberts Says:

    Thank you very much Alex.
    What your looking for, I think, is on the individual team pages, at the bottom.

    Maybe, on the team pages, I should put that “what if” section above the graphs to make it more noticeable. And I wonder how many people follow a link to a league page and never even realize the team pages exist?

  295. Alex Says:

    Oh cool thanks. I had clicked on the teams before but never actually scrolled down past the graphs.

  296. Danno Says:

    Great site! But I noticed today there are no charts for the
    NHL — Eastern Conference – Chance Will Make Playoffs — section.
    Thanks again for a great site.

  297. Burak Says:

    Ankaraspor from Turkish super league is disqualified from this season. Can you please correct your tables according to this information.
    In addition can you update the tables for Turkish Super League in real time?

  298. Neal Says:

    Great Site!!

    Could you put up a legend to decipher the table?

  299. Unknown Says:

    Hi, I think it would be interesting to see the stats of Western NHL teams vs Eastern NHL teams (how many of the west vs east games did each the western teams group win and how many the eastern teams won)
    I think it would be an interesting stat (which is unfortunately not displayed anywhere that I know), because it seems like the western teams are a fair amount better… (perhaps you could even consider that in your algorithm)

  300. Ken Roberts Says:

    Thanks all.
    @Unknown, that’s a good idea. I’d like to see that too.
    @Neal, most of the tables have an “explain” link that opens up to show a legend.
    @Burak, I think it is straight now. The table may still look different from other places because my table goes ahead and gives out the points for the forfeited games ahead of time. And I set it up to fire off more often. If you want, email me I’ll set you up to be able to send in scores yourself. The site would update 30 minutes or so after you sent new scores.
    @Danno, I don’t know what’s up with the missing chart you found. Let me know if you see some kind of pattern in it happening.

    Again, thanks for all this feedback.