Understanding Clutch

We've gone over how the SPI is calculated. I've also explained that you can take a team's SPI rating and extrapolate how many games you would expect them to win or lose. Sometimes, however, you find a team who is either winning or losing far more games than the numbers would lead you to expect. How can we account for this?

Ballplayers and fans typically explain this with something called "clutch." Teams that have clutch are those who can pull out close wins late in games. They usually win by small margins and in dramatic fashion. Those without clutch tend to fold in big games or at the end of close games. So how does one measure clutch? It's difficult, since clutch is more of a feeling than anything concretely measurable, but there are some statistics one can use. Let's compare an overachieving team and an underachieving team as an example. Our overachievers are the Los Angeles Angels (actual win-loss of 61-39, with an expected win-loss of 54-46), and our underachievers are the Atlanta Braves (actual win-loss of 47-53, with an expected win-loss of 52-48). They're off by an average of six games, which is enough to dramatically change a pennant race. If Atlanta were playing six games better, they'd be contending for the NL East title. If Los Angeles were playing six games worse, they'd still be in the AL West lead, but with a much smaller margin.

To start, I looked at each team's record in one-run games. If a team consistently does well in close games, it's likely that they're a team that's good in the clutch. (They could also just be lucky. More on that later.) The Angels are 19-13 in one-run games this year, while the Braves are an abysmal 6-22. Ouch. Those are games that could have swung either way, but they tend to consistently swing toward the Angels and away from the Braves. But not all close games are decided by one run. I looked for two-run games to see if we could see the same trends. In both cases, the team's records improved, but the difference is still clear. After factoring in two-run games, the Braves land at 13-24 while the Angels rise to a staggering 39-18. That's pretty significant.

Except none of this accounts for defense. It's one thing to see a team's offense narrowly outscoring another, but it's something else to see a team's defense protect a slight lead. That's a little trickier to manage, since the defense is part good pitching and part good fielding, but there's a statistic that will do the job. Pitchers can earn what is called a "save" if they enter the game with a lead of at most three runs, pitch at least an inning, and preserve the lead. Pitchers (and teams) with lots of saves tend to be involved in close games, since no one is credited with a save for preserving a lead of four runs or more. Team pitching stats are just as revealing as their close game records. The Angels have 43 team saves, 41 of which belong to Francisco "K-Rod" Rodriguez. The Braves, by comparison, have just 15. That could be partially due to poor pitching, but it could also be due to poor fielding. (A look at play-by-play transcripts show a dearth of late-game errors, so it's more likely poor relief pitching.) A high number of blown saves is a good explanation for a team's poor record in close games.

So why, then, is clutch not factored into the SPI? Because while I feel that there is some measure of skill involved in clutch, most of it has to do with luck. A close game can be decided by something as small as the way a ball bounces. A good team will be able to pull through in a close game most of the time, but they won't be able to consistently beat the odds. Sooner or later, luck is bound to catch up with them (for better or for worse) and bring them back toward the center. And then there's the simple fact that I trust a high-scoring team more than a low one. If you had to put money on a team to win one game, would you bet on the one who consistently won by a high margin or the one who tended to narrowly pull out games? Personally, I'd feel safer betting on the higher-scoring team, but maybe that's just me.