Fire Gardy

In 2008, the Twins went 88-75 and finished second in the division. Jerks from Boston leaped eagerly from the woodwork to claim that the Twins were, in fact, terrible; that the only reason they won 88 games was because they were lucky enough to score runs with men on base, whereas “good” teams score most of their runs with nobody on base.

(That was their point, right?)

The issue, of course, was that the Twins hit very well with runners in scoring position. Considerably better than they did without runners in scoring position. The logic goes that since this isn’t a repeatable skill and is just random, the Twins are sure to regress and therefore score many fewer runs next year.

On the other hand, the best predictor of actual record is Pythagorean record (which is based on runs scored versus runs against). The Twins scored 5.1 runs per game while giving up 4.6 runs per game; this corresponds to an 89-74 team, which is remarkably close to reality*. Despite the raucous calls of those self-proclaimed mathematical geniuses, Baseball Reference points out that the Twins were actually UNLUCKY in 2008, when it came to their record.

Amusingly, the White Sox scored 5.0 runs per game and gave up 4.5, and thus had an identical 89-74 Pythagorean record. It really should come as no surprise that these two teams went down to a 163rd game, they were about exactly as good. And if the Twins had come out on top of that 1-0 game, we’d have a Luck of 0 while the White Sox have our Luck of -1. Although given that we scored 99 runs against the White Sox, and they scored 101 against us, a reversal of that 1-0 score would have put us each at 100 runs, adding to the bafflingly poetic symmetry.

That said, though, the runs-scored vs runs-against nature of Pythag doesn’t account for the crux of the issue, which is that the Twins scored runs when it counted most, and that ability has not correlated from year to year (ie, when you do it one year, that doesn’t mean you’ll do it the next year — which means it isn’t a “skill”).

So … thankfully we have access to FanGraphs’s new context neutral Win Values system, and the way they do it at a team level is just to add every player’s values up (they do this for you, thankfully) and get the entire teams “Wins above replacement level.” For the purposes of their system, the replacement level is set as a .289 team*. That’s 46.8 wins. The Twins got 17.6 wins from their position players, and 16.2 wins from their pitchers. Which puts them at 80.6 wins overall. Rather than being an 88-89 win team (based on reality and Pythag), the Twins instead were just an 80-81 win team (based on context neutral win values).

I wonder what a team would actually be like if EVERY player were actually a replacement level player. It’s a fun though experiment; say the entire 40 man roster gets pissed at you and goes on strike, or flees the country, or are arrested for not paying their taxes. What would the team do? Shut down? Or would they run around grabbing AA players and minor league free agents and scrounging the waiver wire and holding tryouts at the stadium? Just how bad would that team be? That’s what a replacement level player is supposed to be — someone who’s freely available in AA or in the scrap heap. Would such a team really go 46-116?

If the Twins had taken some of their clutch hits and moved them to non-clutch situations, of course they’d have scored fewer runs. That’s basically all this says. But it indicates that the Twins were, in fact, a little lucky in that their run-scoring was highly leveraged, if not that they won an inordinate amount of close games. (Using the same method, the White Sox were a 92 win team in 2008.)

The one thing to give us hope, though? The correlation of team Win Values to reality and Pythag.

Win Values to Pythag: .90 Win Values to Wins: .85

So the actual record is more closely correlated with the Pythagorean record than it is with the Win Values record. For what that’s worth.

Frankly, I think the Twins need to get on base more and get more doubles and home runs.

[Update: Apparently there was a bug on the pitching team values page at FanGraphs, where it failed to correctly count values for pitchers that switched teams midseason. The Twins went from 16.2 to 17.7 for their pitching, which puts us at 82.1 expected wins in 2008.]

No comments