The Case Against Sabermetrics
Sabermetrics are an incredible tool for managers, GMs, and owners.
For anyone who’s trying to put together a winning team, pulling together all the stats you can is obviously a good thing to do. My beef is not with them. My beef is with an overreliance on sabermetrics as a replacement for analysis.
There are good stats and good ways to use stats. Basketball advanced stats make a lot of sense. Turnover rate, adjusted tempo, and rebound percentage are all easy to explain and easy to wrap your head around. Taking points per game as the most important stat for team offense would have you believing that Wisconsin had the 67th best offense nationally last season, rather than the top ranked offense when you take into account that they had the second slowest offense in the country.
There are stats of questionable use in basketball too. Offensive rating on an individual level is a pretty good example of one. Am I really to believe that Tyson Chandler (1st in the NBA in offensive rating) and Ed Davis (5th) are better offensive pieces than James Harden (16th) or Stephen Curry (11th)? Then you have a stat like win shares, which takes more than 2500 words to actually explain. If it takes a college dissertation to explain your statistic, odds are it’s too complicated. Maybe DeAndre Jordan did contribute 12.8 wins to his team this year. There is literally no way to prove it. It’s like Sigmund Freud doing statistics: it’s pseudo-science.
In basketball, we’re beginning to see more and more sabermetrics make their way into everyday talk about the game. It’s hard to talk about the Atlanta Hawks’ season without talking about Kyle Korver’s unreal true shooting percentage (a stat the incorporates free throw percentage, two point field goal percentage and three point field goal percentage) of .699. But for the most part, basketball talk is still very much rooted in stats like points per game, rebounds per game, assists per game, and shooting percentages of different types (FG%, FT% and 3FG%). Only a couple of people had beef with Andrew Wiggins taking home Rookie of the Year while being a pretty inefficient scorer.
Baseball, on the other hand, is now a statistician’s game.
Turn on Baseball Tonight, or whatever other baseball show tickles your fancy, and time how long it takes before you hear a reference to some stat like WAR or OPS+. I promise you it’s not long. A game that became a national pastime as a brief rural escape from 20th century urbanity is now almost 100% about what your calculator tells you about a team.
I’ve read Moneyball three times now, and every time I love it. The movie wasn’t great, but who cares. Building a winning team on a shoestring budget by believing in players and stats that people overlook is a really cool story and clearly other front offices needed to catch up to what the Oakland A’s were doing.
But what has spawned after it is a rash of stats that are esoteric in nature and a way of using them that is shortsighted at best and lazy at worst. Advanced stats are so omnipresent in baseball, that if you really wanted to, you could say that just about any player in the league should be an all star. Maybe Daniel Murphy is the toughest out of any infielder in baseball. After all, no infielder in the league who’s played half of his team’s games has a higher “contact rate” than he does (92.8%). Maybe Rene Rivera, catcher for the Tampa Bay Rays, is actually a really good hitting catcher. Yeah he’s hitting just .143, but his average on balls put in play is an abnormally low .176. Surely those numbers will even out. Maybe the Cleveland Indians have a miserable defense so they should call up a defensive specialist. After all, they’re 26th ranked ultimate zone rating is really low.
Just looking at stats can lead you into some really off-base conclusions. It is really easy to try to break down a game by just looking at box scores (or whatever medium one has to dig up to find someone’s Weighted Runs Created+ stat). It is also almost always a mistake. Maybe Daniel Murphy is just an incredibly unlucky player who has been hitting at-him balls all season. More likely, he’s just not a great hitter. Maybe the Indians’ defense is atrocious, and they need new players to fix it. Or maybe their defensive shifts are wacked out, and players just aren’t in the right place to make the plays. Maybe DeAndre Jordan is a great offensive player who shoots an unreasonably high clip from the floor. Or maybe he’s an incredibly limited offensive player who can jump high, messes up offensive spacing, and can’t hit the broadside of a barn from the foul line.
Real analysis is not 100% about stats, and it’s not 100% about the eye-test. There needs to be a kind of hybrid. With the continued proliferation of advanced metrics, it’s now really easy to just look up a stat that agrees with your argument, throw it down, and say that you are objectively right. That’s not analysis. There’s as much useful information in things that “don’t show up in the box score” as things that advanced stats will show.
Good sports writing will cover both.