Sowell on Baseball

It's a little scary to disagree with Thomas Sowell about anything. In the economic realm, I'd never dare. But in a couple of columns recently written for National Review Online, he's tread into an area that I fear he hasn't done the research: baseball.

Sowell tries to determine who are the greatest hitters and pitchers of all time. In the hitting department, he doesn't give a single answer, but narrows the list to five: Ty Cobb, Rogers Hornsby, Babe Ruth, Ted Williams, and Lou Gehrig. All worthy choices, certainly, and I'd be willing to agree that he's ended up with sound selections. But his methodology is seriously flawed. The problem is when he writes this: "But it would be hard to consider someone for the title of the all-time greatest hitter if someone else had both a higher lifetime batting average and a higher lifetime slugging average."

That's just not true. Suppose player X hit .300, slugged .600, and rarely walked (a factor Sowell entirely ignores), so that his on-base percentage (OBP) is .330. Meanwhile player Y hit .298, slugged .595 and walked often for an OBP of .400. Before you reply that, walks notwithstanding, X is clearly, if narrowly, better than Y, let's add one more piece of information: X's career took place mostly from 1990-2005, and Y's from 1960-1975. It's pretty well established that X would have been playing at a time when, for whatever reason, hitters flourished, while Y played when pitchers did. Can you clearly say that X is better than Y?

Ignoring walks is another problem. You might say it's the job of a hitter to hit, not to walk, but that doesn't really make sense. The game is won by scoring more runs than your opponent. Since only hitters (and, subsequently, runners) can score, ultimately games are won by hitters somehow scoring. If you walk and score, that counts exactly the same as if you'd singled and scored. (Singles are still more valuable since they advance other runners more reliably than walks.) Furthermore, it's clear that hitters have a lot to do with drawing walks: they aren't just randomly handed out by opposing pitchers. There are hitters who regularly draw 100 walks a season, and others who struggle to draw 20. These numbers are fairly stable season to season, indicating that walking has something to do with the hitter's ability.

People who've really studied baseball statistics, a field known as "sabermetrics" (so named because of the Society for American Baseball Research, or SABR), generally start by looking at a combined hitting metric called "OPS", which stands for on-base percentage plus slugging. To calculate it you simply add the OBP to the slugging. In the previous example, player X has an OPS of .330 + .600 = .930 and player Y .400 + .595 = .995. This number correlates pretty well (better than batting average or slugging average taken separately) with the run-generating ability that we really care about.

Of course, if we really care about run-generating ability, and we have some way to tease that out, why not use it directly? We can do that: it's called "runs above replacement" and measures how many runs the hitter generated above some generic "replacement" player. But even that doesn't really help us compare players across generations, because fewer runs were score in some eras than others. So it's really "wins above replacement" (WAR) that gives us the best single metric of a hitter's value.

Here are the top five players in career WAR according to Baseball Reference: Babe Ruth, Ty Cobb, Barry Bonds, Willie Mays, Hank Aaron. Sowell's other three are farther down the list: Williams (6th), Hornsby (8th) and Gehrig (13th).

There's plenty to argue with here. That's part of the fun of this sort of thing. But Sowell wants us to think that he's narrowed the list of hitting greats to a definitive five, and that's just not supportable. It's particularly galling to see him buy into the use of the RBI as a meaningful hitting stat (which is how he squeezes Gehrig into his top five). Sowell says the RBI measures clutch hitting, which may be true, but it also measures how often you were put in clutch situations. Gehrig may have hit a lot of homers with the bases loaded, but he also came to bat a lot with the bases loaded: he player on a team packed with other great hitters. Put Gehrig on the '61 Mets and his RBI total drops considerably.

I suspect, cliched as it is, that the fact that Babe Ruth tops the career list for WAR, OPS+, OPS, and just about any other comprehensive hitting statistic you wish to measure tells us something: that he's simply the single greatest hitter ever. It may be boring, but there's just no other player who dominated like he did. So while I think Sowell's not accurately narrowing the field to five, I do think it's possible to essentially narrow it to one: the Babe.