Dugout Central Challenge: Photo finish

by KerryWhisnant

At the beginning of the season, Dugout Central readers and staff were challenged to predict the number of wins for each of the thirty MLB teams. The predictions were given in previous articles for the AL and NL. There were actually two contests, each with a different scoring method. One was simply to take the average win difference (AWD) between our picks and the final win totals. The other used root mean square error (RMSE), which penalizes really bad individual predictions.

In the AWD contest, it went down to the wire and the winner was decided by one game late on Sunday afternoon. Or, more accurately, the winners (plural) were decided, since when Oakland defeated Seattle 4-3, I pulled into a flat tie with Patrick Greco. The total win difference (TD) and average win difference (AWD) for each player:

Rank TD AWD
1 Patrick Greco 196 6.53
1 Kerry Whisnant 196 6.53
Keith Law 198 6.60
Over/Under 201 6.70
2009 Pythagorean 208 6.93
Average predictions 214.5 7.15
3 Richard Carver 218 7.27
ZiPS 218 7.27
Rob Neyer 221 7.37
4 Jon Ellis (Seven) 222 7.40
5 Sky Kalkman 224 7.47
6 Yu-Hsing Chen 226 7.53
7 Jacob Thompson 228 7.60
CHONE 228 7.60
Marcel 228 7.60
8 Adam White 234 7.80
2009 236 7.87
9 Randy Newsom 236 7.87
Oliver 238 7.93
10 Thomas Wayne 246 8.20
BaseballProspectus 264 8.80
81 Wins 266 8.87
11 John Bowen 276 9.20
12 Chuck Johnson 294 9.80

Also shown are the predictions using the 2009 records, the 2009 Pythagorean records, 81 wins for each team, and the average win predictions from all of our picks. The other picks, computer projections and those of Rob Neyer and Keith Law, are taken from the compilation at vegaswatch.net. The rankings are for predictions submitted to Dugout Central.

Last year’s AWD winner, Sky Kalkman, slipped to fifth this year. Thomas Wayne, who tied me for second last year, dropped all the way down to tenth. Chuck Johnson, after finishing next to last in 2009, “improved” to last place this year. Chuck and John Bowen did worse than simply predicting everyone would win 81 games.

The performance of the so-called computer picks — ZiPS, CHONE, Marcel, Oliver and Baseball Prospectus – was good to poor. Last year the previous year records and Pythagorean records did not do well, but this year they were better, especially the 2009 Pythagorean records. The Vegas Over/Under win totals, which beat everybody in AWD last year, did very well again this year. Keith Law almost matched Patrick and me, while Rob Neyer’s predictions were slightly better than our middle of the pack. As a group we did pretty well, with the average predictions beating everyone but Patrick and me.

The RMSE title was decided long ago, and I finished with the best RMSE score for the second year in a row:

Rank RMSE
1 Kerry Whisnant 8.16
2 Patrick Greco 8.70
Keith Law 8.72
Over/Under 8.94
2009 Pythagorean 9.01
ZiPS 9.08
CHONE 9.26
Average 9.28
3 Richard Carver 9.30
4 Sky Kalkman 9.34
5 Jacob Thompson 9.42
Rob Neyer 9.43
2009 9.48
Marcel 9.73
Oliver 9.82
6 Jon Ellis (Seven) 9.93
7 Randy Newsom 9.97
8 Adam White 10.13
9 Yu-Hsing Chen 10.14
10 Thomas Wayne 11.00
81 Wins 11.00
BaseballProspectus 11.13
11 John Bowen 12.52
12 Chuck Johnson 12.82

The order for RMSE was shuffled somewhat compared to the AWD standings. Sky, Jacob and Randy moved up, but Jon and Yu-Hsing dropped. The top and bottom of the standings remained pretty much the same.

In a future post we’ll look at some of the surprises and find out who came closest for those teams.

Recent Posts

13 Responses to “Dugout Central Challenge: Photo finish”

  1. Chuck Says:

    ” Chuck Johnson, after finishing next to last in 2009, “improved” to last place this year.”

    Smart ass.

  2. Cameron Says:

    Dare I ask, how do you do worse than a straight 81-win prediction? …And how could BP fuck up that bad?

  3. Chuck Says:

    “Dare I ask, how do you do worse than a straight 81-win prediction.”

    Dude, easy as pie.

    First, you pick Seattle to win the West.

    Then, you pick San Diego to have the worst record in baseball.

    Follow that up with Pittsburgh and Baltimore being the two most improved teams in the league.

    Then..

    Actually, that did it.

    Four teams cost me a top three finish.

    Can o’ corn.

  4. Kerry Says:

    @2, you do worse than 81 by getting on the wrong side of 81 for several teams. As Chuck mentioned, the Padres and Mariners were the worst (many people had problems with them). Also losing ground for Chuck versus 81 wins: the Jays, Cubs and Giants.

    “Four teams cost me a top three finish.”

    Well, to be fair, if other people get to take out their worst ones, they would improve, too. Or even just the four teams you mentioned. For example, Chuck missed by 91 wins on those four teams, while Richard missed them by 60, Jon 56, and Sky 58. Patrick may have done best on those four (I didn’t check everyone), missing by “just” 49.

  5. Hossrex Says:

    Kerry: “The other used root mean square error (RMSE), which penalizes really bad individual predictions.”

    I’m not a big fan of that model.

    The only time you’ll ever see a “really bad individual prediction” is if a team finishes either FAR above expectations, or FAR below expectations. You aren’t a particularly knowledgeable baseball fan if you predicted that San Diego, Cincinnati, and Texas would win 90 games… yet if you didn’t… you’re going to be penalized for making a “really bad individual prediction.”

    In a year where SO many goofy things happened (the aforementioned San Diego Texas and Cincinnati… along with Tampa, Los Angeles, Anaheim, Chicago, San Francisco, and especially Seattle), it seems like a lot of intelligent predictions are going to result in penalties.

    Since someone will feel the need to explain this to me… I fully understand that the whole reason this sort of thing is fun is because unpredictable things happen, and making silly predictions that pan out is the best part…

    But clearly the reason Baseball Prospectus did so poorly had more to do with a bizarre season than with any particular flaw in their methodology.

  6. Kerry Says:

    Hossrex:”You aren’t a particularly knowledgeable baseball fan if you predicted that San Diego, Cincinnati, and Texas would win 90 games… yet if you didn’t… you’re going to be penalized for making a “really bad individual prediction.””

    Chuck, who predicted Cin would win 86 and Tex 94, might disagree with you on that one. Besides, if everyone misses by a lot, they are all penalized (although certainly a 2-game difference when everybody misses by a lot counts more than a 2-game difference when everybody is close).

    I think people tend to guess extremes too often. There are more 100-win or 100-loss teams than there “should be” from the talent levels alone, due to upward and downward fluctuations (“luck”) — that’s just the way the world (i.e., statistics) works. So why should people necessarily get credit for that when it happens?

    The AWD method encourages taking chances, which is fine (and it’s why I followed it as well as RMSE), but I don’t think it’s inherently better — the differences between AWD and RMSE are subjective. And for the most part people who do well in one do well in the other, even though the exact order is not the same.

  7. Kerry Says:

    Hossrex:”But clearly the reason Baseball Prospectus did so poorly had more to do with a bizarre season than with any particular flaw in their methodology.”

    I’m not sure about that. As I understand it, two years ago they switched all their analysis software from Nate Silver’s spreadsheet to another computer, after which Nate Silver was not involved (I think that’s right, maybe Sky can correct me if I’m wrong). And the BP predictions have not been good ever since.

    That doesn’t prove anything, but it is suggestive. After all, other computer-based methods still do OK for the most part, or at least much better than BP. BP was worse than 81 wins in RMSE, and almost worse in AWD. That’s bad.

    So I think there is a problem with their methodology. I start with PECOTA myself and add a bunch of tweaks to it, with completely different results.

  8. Kerry Says:

    Whoops, I had a problem with italics there, they didn’t turn off when they should have.

  9. Chuck Says:

    Hey, Kerry, how about posting the original chart showing everybody’s picks?

  10. Kerry Says:

    Chuck, they are in the AL and NL links above. (The “AL” and “NL” in the second sentence are clickable.)

  11. Cameron Says:

    Wow, in between Toronto, Cincy, Seattle, and San Diego, no one would’ve had a good year this year.

  12. Mike Felber Says:

    Here is a very good article about the likely change in HR production in the last several decades. The conclusions are not intuitive, & the relative (lack of) weight of expansion & parks surprising. A good case is made without invoking an implausible conspiracy theory.

    http://www.hardballtimes.com/main/article/changes-in-home-run-rates-during-the-retrosheet-years/

  13. brautigan Says:

    It shows how often I’m on the ESPN website. I didn’t know until today that Rob Neyer left.

    (Good for you Rob)

Leave a Reply

Advertisement