I started out planning to project the AL East, a first step in my preseason predictions. My methodology is not truly sabermetric: I plan to take each position, estimate NOT by the official projections but by my own gut feeling, how much better/worse the team will be at each position (Yankees lose 6 wins at second base, pick up 4 at catcher, etc.) and then apply the result to last year’s wins to get this year’s projection. Simple, no. Simplistic, yes. Fun, YES.
I got sidetracked by a simple problem: three teams in the AL East out- or under-performed their Pythagorean W/L by 3 or more games, which really messes up the stats. The team I root for (Yankees, if you are reading this blog for the first time, however unlikely) missed by the most: they won SIX more games (85) than the RS/RA suggests (79). So should I add the improvements (they were voted at ESPN the most improved team in MLB this offseason) to 79 or 85?
That question is answerable only if you can answer this one: why do teams outperform or underperform their Pythagorean projection? There are four answers put forth to explain this: 1) Pure luck, 2) Clutch performance, 3) Excellent managers, 4) Strong bullpens. The MARKET (as opposed to the statheads) believes the fourth one: when FanGraphs calculates $/WAR, they have to exclude relief pitchers, and especially closers, because they get WAY more $ than their WAR would suggest. The market thinks that closers in particular and relievers in general, are worth more than the WAR calculation suggests.
If, then, teams that outperform their RS/RA projection ALSO have strong bullpens/closers, it would suggest that WAR is mistaken, and relief pitchers ARE worth more than WAR gives them credit for. If the differences are due to luck (some would say clutch performance IS luck) then we overpay relievers, and my analysis should add to the Pythagorean numbers. If the result is the manager, we are never going to be able to prove it, but we should add to the previous year’s total.
I thought to myself: well, the Yankees over-performed, AND had the best closer in history, AND he had a good year. So maybe there is a correlation between the quality of the year and the performance of the closer. Unfortunately, the Red Sox UNDER-performed (they projected to win 100 games instead of the mere 97 they actually won), AND they had for much of the season the most unhittable closer in recent memory. So the correlation is hardly obvious. A good thought, but no.
I then looked at 1-run games. The Red Sox, despite winning 97 games AND having Koji Uehara, were 21-21 in 1-run games. The Yankees were 30-16 in 1-run games. AHA I thought: a team’s record in 1-run games is a prime determinant in their real to projected W/L. Then I looked at the Orioles, 5 games over-performed and exactly the same actual record as the Yankees: they project at 80-82, actually went 85-77, and their closer (Johnson) led the league with 50 saves. Excuse me? No correlation there, either.
What about the manager, then? IS there any way to see if the manager has a measurable effect? Not really, no. What I finally thought of was this: I could look at the real versus projected record for each manager that I wanted to do (it is work, so I don’t plan to do an exhaustive study). IF the difference tends to be in large part the manager (I remember Bill James claiming that the Weaver Orioles out-performed their projection every year), then it should show up in a year-by-year comparison. So I looked at a couple of managers:
Girardi (with NYY): 6, -1, -6, -3, 6, 1 Net 3 in 6 years
Torre (with NYY): -5, 1, 5, 12, 4, 3, 5, 1, -5, 3 Net 24 in 10 years
Showalter (with NYY): 1, 1, 1, -4 Net -1 in 4 years
Showalter (with BAL) : 0, 11, 4 Net 15 in 3 years
So, no, I can’t find any indication that the difference is in the manager. And in the case of the Yankee managers, I always thought that Torre’s brilliance was in keeping the clubhouse happy and productive, but that he was a poor in-game manager. So I didn’t expect that HIS luck would be better than Girardi’s, at all.
So what I really did was waste my time so I didn’t get to the analysis I wanted to do. maybe next time.
For future expectations: I am home on Monday and Thursday, and hope to post consistently on those days. I am away on Wednesday and expect to post quite infrequently on those days. Tuesday, Friday and Saturday are wild cards: I am home but so is my wife, so we may have plans or we may not. I will post sporadically on those days. And I have to be at church on Sunday at 7:30 am, so I doubt I will do much posting on THOSE days.
But games that count will be played in 12 (!!) days, and the real season starts in 20. Life is good!