Friday, November 4, 2011

Did The 2002 A's Of "Moneyball" Fame Win More Games Than Their Stats Might Predict?

Maybe. I plugged their OPS differential into the following equation for winning percentage:

Pct = .5 + 1.26*OPSDifferential

The A's had an OPS of .771 while their pitchers allowed an OPS of .699. So their differential was .072. The equation predicts a pct of .591 or about 95.7 wins. They actually won 103. So they won about 7.3 more games than expected. The standard error of the regression that generated the above equation was 5.04, so the A's were 1.44 standard deviations above their expectation. Not huge, but not small either.

They did have a 32-14 record in 1-run games for a pct. of .696. They had a .612 pct. in all other games. If they had that for all games, they would have won 99.16 games, alot closer to what their OPS differential predicts.

I don't see anything in particular that explains why they outperformed their projection. The table below shows how both their hitters and pitchers performed in various situations followed by their differentials in those situations (data from Retrosheet).

Nothing really jumps out. Their differentials with runners in scoring position (RISP) are a bit higher than with none on and to a lesser extent with bases loaded. Their OBP differential looks good in close and late situations but their SLG differential is much lower than normal. Their overall OBP differential was .024 while for SLG it was .048.

The A's grounded into 128 DPs, just one more than their opponents, whom they out OBPed .339 to .315. Their GIDP rate was 10% while the league average was 11%, the rate the A's allowed (some data also from Baseball Reference).

The A's only had 20 sacrifice hits while their opponents had 50. So they saved 30 outs that way. The A's were46-20 stealing while their opponents were 68-46. So they saved 26 outs there. All of that is about two games worth of outs.

The A's out homered their opponents 205-135, by 70. They only hit 10 more 2B's than their opponents and had the same number of 3B's. So their advantage in SLG was almost entirely determined by this big HR advantage and HRs have one additional edge over other hits in that they guarantee at least one run.

No comments: