CSS Button No Image Css3Menu.com

Baseball Prospectus home
Click here to log in Click here for forgotten password Click here to subscribe
<< Previous Article
Premium Article Wait 'Til Next Year: F... (12/11)
Next Article >>
Premium Article Player Profile: Mark T... (12/12)

December 12, 2007

Predictatron Post-Hoc

Looking Back at 2007

by Jason Paré

The Wrapup

With the closing of the Winter Meetings and the igniting of the yearly hot stove, it's once again time to examine the results from BP's Predictatron contest. Roughly equivalent to an overgrown fantasy football pick'em, Predictatron asks BP readers and staff alike to forecast the individual records of each MLB team, along with a bracket-like prediction of playoff results and the eventual World Series winner. First prize in the contest is $500 and a framed picture of Commissioner Bud Selig. With swag like that, who needs the money? Full rules can be found here.

The 2007 season was replete with Cinderella stories, collapses, renaissances, and disappointments. How did these events jive with predictions? Let's go to the divisional standings and look at how the contest participants saw 2007 shaking out.

The Results

NL East
Average difference between wins and predicted wins: 4.16

Team    Wins   Pred. Wins  Mortal Lock%  Playoff%   Div%     WC%
PHI     89        87.14       5.2%        64.3%    28.2%   36.1%
NYN     88        89.94      15.1%        87.4%    64.4%   22.9%
ATL     84        83.05       1.1%        17.0%     7.0%   10.0%
WAS     73        62.16      53.6%         0.0%     0.0%    0.0%
FLO     71        76.22       1.3%         1.8%     0.4%    1.4%

As one of the worst collapses in baseball history, the Mets' failure to make the postseason accounted for a minus on more ballots than any other team. Even worse, 15 percent of readers placed them as their mortal lock, and were burned doubly by points subtracted. The East was also home to another ballot-beater, the Washington Nationals, as an overwhelming 53.6% of participants picked them as their NL mortal lock to win fewer than 72 games.

AL West
Average difference between wins and predicted wins: 8.45

Team    Wins    Pred. Wins  Mortal Lock%  Playoff%  Div%    WC%
ANA     94         87.03       2.7%        66.6%   66.1%   0.5%
SEA     88         74.08       3.2%         0.7%    0.7%   0.0%
OAK     76         84.62       1.4%        28.9%   28.3%   0.6%
TEX     75         79.29       0.6%         4.9%    4.9%   0.0%

By average difference in win totals, the AL West was easily the wackiest division in 2007. The Angels provided their usual high level of competition, but few readers predicted them to dominate the west so fully. An interesting switch occurred between the (supposedly) perennially contending A's and perennially underachieving Mariners. While the A's were out of the hunt with injuries, the Mariners remained competitive until the final few weeks of the season to cap a respectable 88-win campaign. This represented the biggest difference between predicted and actual wins for any individual team.

NL Central
Average difference between wins and predicted wins: 3.51

Team    Wins    Pred. Wins  Mortal Lock%  Playoff%  Div%    WC%
CHN     85         83.10       1.2%        32.5%   31.2%   1.3%
MIL     83         83.74       0.6%        37.5%   35.4%   2.0%
SLN     78         83.65       0.7%        30.3%   27.2%   3.1%
HOU     73         78.42       0.8%         4.5%    3.8%   0.7%
CIN     72         75.70       1.8%         2.1%    2.0%   0.1%
PIT     68         71.67       9.5%         0.4%    0.4%   0.0%

Readers were torn about this division, and predicted about 83 wins for each of the top three teams. Though the order was not perfect, no one team made a clear break from the others. St. Louis represented the largest difference between reality and what was predicted, largely because of a bubblegum-and-paperclips starting rotation in the wake of Chris Carpenter's injury and Anthony Reyes's ineffectiveness.

NL West
Average difference between wins and predicted wins: 5.99

Team    Wins    Pred. Wins  Mortal Lock%  Playoff%  Div%    WC%
ARI     90         85.65       2.9%        55.4%   49.2%   6.2%
COL     90         76.32       1.0%         1.3%    1.2%   0.1%
SDN     89         85.55       1.9%        36.1%   25.8%  10.3%
LAN     82         84.54       1.9%        27.3%   22.1%   5.2%
SFN     71         76.95       1.4%         2.0%    1.7%   0.4%

The NL West played host to two of the season's great stories: the late-season emergence of the Colorado Rockies and the consistent winning of the Diamondbacks in spite of their well-publicized negative run differential. While 55 percent of readers picked the Diamondbacks to make the playoffs, the average win prediction was still over four wins below their actual record. Conversely, only 1.3 percent of you picked the Rox to make the postseason, reflected in their 13.6-win difference between projection and reality.

AL East
Average difference between wins and predicted wins: 3.30

Team    Wins    Pred. Wins  Mortal Lock%  Playoff%  Div%    WC%
BOS     96         91.69       18.3%       87.3%   35.7%  51.6%
NYA     94         93.38       37.7%       93.1%   63.1%  30.0%
TOR     83         82.28        0.4%        4.2%    1.0%   3.2%
BAL     69         73.41        4.4%        0.6%    0.2%   0.4%
TBA     66         72.44        4.0%        0.0%    0.0%   0.0%

With the exception of the Red Sox capturing the divisional title, it was business as usual in the AL East. With the lowest average difference between wins and projected wins, it was evident that readers saw the Red Sox/Yankees battle as the probabilistic crapshoot it turned out to be. Both the Orioles and Devil Rays slightly underperformed their predicted wins, but overall the participants nailed this one. It's worth noting that the Yankees had the lowest individual difference between prediction and reality out of all 30 teams.

AL Central
Average difference between wins and predicted wins: 3.51

Team    Wins    Pred. Wins  Mortal Lock%  Playoff%  Div%    WC%
CLE     96         88.84        5.9%       63.7%   60.6%   3.1%
DET     88         86.24        1.2%       23.4%   18.0%   5.5%
MIN     79         85.71        1.7%       23.2%   18.7%   4.5%
CHA     72         79.84        0.6%        3.4%    2.7%   0.7%
KCA     69         67.36       17.8%        0.0%    0.0%   0.0%

It seems that, for our readers, the AL Central was a foregone conclusion. This was the only division in which Predictatron participants predicted the team standings in perfect order. Many BP readers wisely remembered Cleveland underperforming their Pythagorean record in 2006, and banked on their bounce back to contention. The Central also boasted the only non-Sox or Yankees team to have double digit "Mortal Lock" potential in the AL (the Royals), and they met these dim expectations by scratching out 69 wins.

Inside the Ballot of a Winner

This year's cash and Bud Selig photo go to Noel Langlois, a BP reader since 2005. Though he correctly picked the Red Sox to win it all, the real power of Noel's ballot was in his regular-season predictions, as you can see below:

TeamYour GuessActualDifferencePrediction NotesScorePlayoff Points
ANA86948Made Playoffs8+2
ARI88902League Champion2+2 +4
BOS95961World Series Champion1+2 +4 +8 +16
CLE92964Division Series Winner4+2 +4
MIL88835Division Series Winner5
NYA91943Mortal Lock Made Playoffs6+2
NYN94886Made Playoffs6
PHI86893Made Playoffs3+2
WAS65738Mortal Lock 16
TOTAL SCORE: +1000128+48

He picked an outstanding 22 teams within five wins of their actual total, 14 teams within three wins, and was within one win for an astounding 10 teams. The only discrepancies of 10 games or more were the overachieving Mariners and the miraculous Rockies. Perhaps his most impressive insight was his nearly spot-on prediction for the Diamondbacks, which he pegged because of the D'Backs' wealth of young players and the "March Madness-like upset factor." He's going to use the cash to fund his annual BP subscription, and maybe even put a down payment on half of a Green Monster seat. Congratulations, Noel.

Comparison and Conclusion

As with any projection system, it's illuminating to compare the accuracy between different systems. In this case, I've compared Predictatron with preseason PECOTA and ESPN Expert predictions, which are plotted on the graph below:

prediction comparison

R-squared values for the methods' correlation with actual wins are as follows:

Predictatron: 0.568
PECOTA: 0.479
ESPN: 0.462

For the second year in a row, Predictatron's prognostications surpassed both PECOTA's and ESPN's in accuracy. Many thanks to everyone who participated for another successful year. May your team's offseason be fulfilling, and we look forward to seeing your predictions again next April.

Related Content:  Preseason Predictions,  Wins

0 comments have been left for this article.

<< Previous Article
Premium Article Wait 'Til Next Year: F... (12/11)
Next Article >>
Premium Article Player Profile: Mark T... (12/12)

BP Wrigleyville
BP Mets
Premium Article Minor League Update: Games of August 22
What You Need to Know: The Business
Short Relief: A Broadcaster Out of Step with...
Premium Article Rubbing Mud: The Mental Side of Shifting
Baseball Therapy: Positional Anarchy!

Premium Article The Class of 2008
Premium Article Player Profile: Mark Teahen

2007-12-12 - Predictatron Post-Hoc
2007-11-11 - Premium Article Dropping In