THE BOOK--Playing The Percentages In Baseball

Monday, May 26, 2008

Evaluating the team forecasts so far

Here is a chart of various forecasters’ pre-season team w/l projections, along with the Vegas lines (rounded to the nearest even whole number), along with each team’s current record prorated to a whole season, along with (finally) each team’s end-of-season expected w/l totals given their schedule, current record, and my estimate of their true talent win percentage.  The second to last column is if the average forecaster had to bet over or under the Vegas line, whether they would have won or lost if the season ended today (prorating each team’s current record to 162 games).  The last column is each team’s chances of winning the World Series given their schedule, current record, and true talent WP, as estimated by me.

Of course, comparing pre-season projections to current ones is a little unfair, as forecasters presumably consider who is on the DL at the beginning of the season, who is going to get called up mid-season, which teams might improve or get worse through trades, etc.

Team Pecota MGL Neyer Chone Vegas Current wins prorated Forecasters versus Vegas Projected end-of-season (EOS) wins Chance of winning WS (in %)

TBA 85 86 81 89 76 97 W 89 7
BOS 91 90 89 92 94 95 L 94 16
TOR 80 82 85 83 85.5 84 W 81 .7
NYA 93 92 94 92 94 81 W 86 4
BAL 73 70 67 65 65.5 79 W 74 0
CHA 79 77 74 76 78 89 L 83 3
MIN 78 76 75 76 74 81 W 78 .5
CLE 87 91 91 92 90 75 L 85 7
KCA 70 70 76 70 73.5 68 W 71 0
DET 83 88 90 91 93 68 W 82 3
ALA 87 87 88 91 91 93 L 89 9
OAK 85 80 82 75 73.5 89 W 85 4
TEX 73 75 74 72 75 81 L 75 .1
SEA 77 77 77 83 83.5 57 W 71 0
FLO 77 73 68 70 68.5 96 W 81 .9
ATL 85 87 85 83 86 87 L 85 4
PHI 81 82 84 87 88 87 W 86 5
NYN 90 91 95 92 93 78 W 86 5
WAS 70 74 71 70 71.5 70 L 70 0
CHN 90 86 91 87 87.5 94 W 89 8
SLN 81 78 77 75 76 93 W 85 3
HOU 72 72 74 75 75 90 L 79 .4
PIT 74 74 71 75 70 78 W 75 0
MIL 89 89 85 84 84 75 L 81 1
CIN 77 77 76 78 78 73 W 77 .2
ARI 88 86 85 85 88 97 L 92 12
LAN 85 85 88 84 87 86 W 86 4
COL 80 84 81 77 83 65 W 75 .1
SFN 67 66 70 72 71 64 W 65 0
SDN 83 86 86 84 84.5 59 L 72 0

The forecasters are beating Vegas so far, 19-11, which is no surprise, to me at least.

If we take the square root of the average squared error (root mean squared) for each forecaster as compared to the current records prorated to 162 games, we get 10.8 for Pecota, 12.6 for me, 13.3 for Neyer, and 13.8 for Chone.

Notice how some crappy teams with bad records still have some chance to win the WS, through luck alone.  Only BAL, KC, SEA, WAS, PIT, SD, and SF never won a WS in 10,000 seasons.  Also notice how small some teams’ chances are despite good records so far (because they are really crappy teams), like FLO and HOU.  Notice also how despite bad records so far, the Mets, Yankees, and Cleveland have a decent chance to win the WS!  These numbers are based on running 10,000 seasons through a sim which pits each team against another, according to the schedule, using a static WP for each team (and a log 5 matchup method).