Comments on economics, mystery fiction, drama, and art.

Saturday, June 25, 2011

A Baseball Post

This week, Jim Riggleman, the now former manager of the Washington Nationals, resigned because the team refused even to discuss a contract extension. Over his managerial career, Riggleman's teams have had a combined .445 winning percentage (in 1,486 games). At the very least, this record does not leap out at me saying that this is a man who should have a long-term contract. [For comparison purposes, though, it's worth noting that at a similar point in his career, Joe Torre had managed his teams to a cumulative .466 winning percentage (in 1,574 games).]

Can we get any kind of a "handle" on how Riggleman has done as a manager, compared with what we might call our expectations? That is, how well has he done in his full seasons as a manager, compared to how we might expect a team with the talent available to him to have done? And how does he stack up is we compare him, on this scale, to Torre?

Bill James created a simple measure of a team's expected winning percentage, which he calls the "Pythagorean" winning percentage, equal to [(RS)^2]/[(RS)^2 + (RA)^2], where RS is a team's runs scored and RA is a team's runs allowed, and the operator (^2) indicates we are squaring that value. A team, then, scoring and allowing the same number of runs has an expected winning percentage of .500; a team scoring 600 and allowing 700 has an expected winning percentage of about .425. A team scoring 700 and allowing 600 has an expected wining percentage of about .575.

How did Riggleman do with his full-season teams? In six full-year managerial seasons, his teams exceeded expectations twice (by a total of 7.7 games) and were below expectations 4 times (by a total of 20.8 games; a lot of that came in his first full season, with San Diego, when the team finished 11.7 games below expectations). In those six full seasons, then, he managed four teams with an expected winning percentage of .458 to an actual performance of .423. In the two years his teams exceeded expectations, he got a .485 wining percentage out of teams whose RS and RA would have predicted a .461 record. Overall, his teams won at a .443 clip against an expected .459. This is, in fact, not a stellar record.

What about Torre, through the same part of his career? Torre managed 8 full seasons, and his teams finished below expectations 5 times (367-443, .453) and above expectations 3 times (253-233, .521). Overall, Torre's teams finished 620-676, .478 against an expected 630-666, .486. Riggleman's record was a little, but not a lot, worse than Torre's at that point in their managerial careers. Had his winning percentage been only .008 below expectations, his teams would have won 7 more games in 6 years. The real difference is that Torre managed teams with an (average) expected record of 79-83, while Riggleman's had an (average) expected record of 74-88, five games per year worse.

But there's an additional point. Really bad teams--teams with an expected winning percentage below .450--tend to underperform their expectations (between 2006 and 2010, 19 teams had expected records of 73-89 or worse; 15 of them won fewer games than their RS/RA nnumbers would lead us to expect). Two of Riggleman's full-season jobs were with teams expected to finish below .450 (two right at.450). With those two teams, one finished 2 wins worse than expected and the other finished 2.6 wins better than expected. Torre's first three teams (the Mets, 1978 - 1980) all had expected records worse than .450, and all under-performed that, by 11.5 games (3.8 per year).

Or maybe two additional points, With teams expected to have losing records, but to finish .450 or better (Rigglleman has had 3 such teams, Torre only 1), Riggleman's teams finished an average of 8 wins below expectations and Torre's 1 such team was 2 wins above expectations. Riggleman had only one team that more RS than RA, and he brought it home 5 games above expectations. Torre had 3 such teams; 2 did an average of nearly 3 games better than expectations and 1 did 5 games worse than expectations.

So Riggleman did better than expected, and Torre worse, with really bad teams. Riggleman did worse than expected with moderately bad teams, and, the one time he had a good team, considerably better than expected. Torre about met expectations, overall, when he had moderately bad or good teams. (All of this comes with very small samples and is almost certainly not statistically significant.)

Two managers, both of whom had not very good records with not very good teams through about 1,500 games. One is now almost a lock to make the Hall of Fame as a manager. The other? Getting another managerial job may be a really, really unlikely event.


Post a Comment

<< Home