The Fan:
Levels of Analysis

Benjamin Gregory Ramm

Does it ever make sense to make an out intentionally? Does a team need situational hitting?

Many educated baseball fans answer no because teams with a high OPS (On-base Plus Slugging percentage) tend to win many games over the course of a season. But those who play the game - and win the game - sometimes answer yes.

The argument goes hopelessly wrong both when a media commentator declares that a team's key off-season move was the signing a thirty year-old journeyman who seems to be able to bunt really well and when a statistics professor complains that the journeyman pinch hit for a free-swinging power hitter in some close and late situation.

So who is right? The sabermetric adherents who argue for keeping the possibility of a rally alive for as long as possible, or the small-ball advocates who will sacrifice the possibility of many runs for an increased probability of one run?

This is a classic "levels of analysis" problem where different observers view different parts of the same problem and argue past each other. Putting together a successful team involves the season-level philosophy of putting together a team for 162 games, the game-level strategy of approaching a particular pitcher, and the tactics used in a particular game situation.

If you ever want to assign blame, give credit, or understand causation, don't conflate different parts of a team's success. Each level of an organization has responsibility for a certain set of decisions. A general manager implements management's team philosophy, while a manager devises a strategy for a particular game, and then pays enough attention to the players and to the tactical situation of a particular game to know when deviating either from a general philosophy or a particular strategy improves his chances of winning.

My philosophy says to build a team with a high OPS. Try to include Mike Piazza, Mark McGwire, Craig Biggio, Ken Caminiti, Alex Rodriguez, Albert Belle, Ken Griffey Jr., and Juan Gonzalez, with Frank Thomas at DH in your lineup. And if you can support that payroll, forget about a "true lead-off hitter", "team speed", "chemistry", or "situational hitting". To win at least 130 games, would you need them?

But what if this team had to face Randy Johnson, Pedro Martinez, or Greg Maddux? They would only score eight runs. In other words, what makes for a great example of a high team OPS philosophy makes for an absurd test case for a game-level strategy. Even the most potent offense rarely has more than two All-Stars. For teams with normal budgets, the front-line power-hitting lineup that works very well against most pitchers might not work against certain sorts of pitchers. In September of 1996, I saw Jamie Moyer shut out a very good power-hitting Texas Ranger team. Moyer never challenged them. He threw off-speed junk and it was comedy. What if they had fielded a line-up that was speedier, less prone to strike-outs, but hit for less power? Over the course of 162 games, they would have won fewer games, but for that one game, they might have had some success.

The success that the Baltimore Orioles had against Randy Johnson in 1997 might support my assertion. Admittedly, Orioles' pitching had more to do with the wins than anything else, but Davey Johnson's strategy bears examination because it worked. Against the most intimidating pitcher in baseball, Davey Johnson played his second-tier, righthanded hitters, encouraging them to be patient and make contact. Eventually, they broke down Randy Johnson and they broke open two of the games, winning all four.

Does this have any impact on how a general manager should build a team? No. Don't pursue bench players, but play them. These guys may not contribute 90% of the time, but make sure that the bench has something to offer when the game-level strategy needs to deviate from the season-level philosophy.

Deviating from sound philosophy comes under fire when it fails, and it usually does. Most teams who try Davey Johnson's strategy against Randy Johnson lose both because they don't have the pitching that the Orioles did, and because they don't have the hitters that the Orioles did. Plus, measuring the success of risk-averse strategies against dominant pitchers is almost impossible because dominant pitchers do just that - dominate.

Even when the highest total OPS line-up will do well for much of the time, there will be a few games that end up tied one out into the bottom of the ninth, with a man on third. In those games, drawing a walk, starting rallies, and hitting three-run homers would not be any. OPS measures performance that would be irrelevant to that situations, like walks and extra-base hits while discounting bloop singles and long flies.

There is one more consideration that anyone with the most basic understanding of game theory knows: if you let the other player see a predictable pattern in your behavior, your philosophy will gradually lose effectiveness. In other words, if the other team knows where you will hit the ball, they can put their players there. Don't be predictable. Use the whole field.

Statistical correlations or rules such as "team OPS correlates with wins" will show a manager the best options most of the time, just like putting your money into an indexed mutual fund will do pretty well. When playing baseball or investing money, however, you should look to do better than expected. Genius is knowing when there will be exceptions to the rules: a good mutual fund manager will do better than an index fund just as a good baseball manager will do better than the best fantasy-league couch potato.

Benjamin Gregory Ramm often reminisces about working the Wrigley Field scoreboard in the '30s, and has a couple digits kicking around the house to prove it. Honest, they're there somewhere. Give him some time to find them, then send for verification at bgr@strikethree.com.