Sabermetrics vs. (Uh) Sabermetrics?
This Pat Caputo column about the Hall of Fame has, well, a lot of problems. A lot. For one thing, he left off a bunch of players (28 to be exact) and then wrote, “You can speculate all you want about those I left off my ballot and performance enhancing drugs.” That seems, I don’t know, kind of irresponsible to me. He left off, among others, Curt Schilling, Mike Mussina, Craig Biggio, Jeff Bagwell, Mike Piazza, Edgar Martinez, Jeff Kent, Larry Walker, Luis Gonzalez, Sean Casey, Hideo Nomo, Kenny Rogers and Mike Timlin. Apparently we’re just supposed to speculate about all these people in addition to the ones who were actually connected to steroid abuse?
Say it isn’t so Hideo. I’m sure this column will be covered at length by the folks over at Baseball Think Factory. Fish. Barrel. That sort of thing. But there is a paragraph in there that is worth pulling out for discussion because I think it gets to the heart of a sabermetrics-oldschool argument that, frankly, should not be happening.
Pat was making his argument for Jack Morris as a Hall of Famer, which is fine, but then he writes that sabermetrics misses a key point about Morris. The point is that Morris’ 3.90 ERA -- which, everyone knows, would be the highest in the Hall of Fame -- was much better than it looks because much of his career took place “during the height of the so-called steroids era.”
He writes this:
Sabermetrics has its flaws. One of them, for evaluation purposes regarding the Hall, is not accounting enough for statistics era to era. A 3.00 ERA in 1968 didn’t mean nearly as much as a 3.00 ERA in 1995, for example.
OK, there are a half a million problems with all this. Morris did not pitch in 1995. In fact, he really didn’t pitch almost at all in the steroid era -- he made only 50 starts in 1993 and 1994, which might count as a early steroid era.* He certainly didn’t get anywhere close to the “height” of the era. He actually pitched in a very LOW scoring era historically. His neutralized ERA (neutralized to an average run-scoring season) is actually 4.28, which you will note is HIGHER than his actual ERA.
*By “Steroid Era,” I think we are all talking not about when players used steroids but when scoring runs got out of control. Players might have used steroids back in the 1970s, who knows? In fact, it’s all but certain that at least a couple of players used steroids in the 1970s. What we’re talking about here is high-scoring years. You will note the spike in runs per game -- THAT is the Steroid Era:
1985: 4.33 1986: 4.41 1987: 4.72 1988: 4.14 1989: 4.13
1990: 4.26 1991: 4.31 1992: 4.12 1993: 4.60 1994: 4.92
1995: 4.85 1996: 5.06 1997: 4.77 1998: 4.79 1999: 5.08
2000: 5.14 2001: 4.78 2002: 4.62 2003: 4.73 2004: 4.81
This past year, it was 4.17. I would say, other than the blip in 1987 (when the ball was thought to be juiced) the steroid ERA probably began in 1993 and peaked from 1996-2001 or so.
Morris also never had a 3.00 ERA in his entire career. The closest he ever came was a 3.05 ERA during the 1981 strike season.
But my point is not Morris -- Pat makes some fair points about Morris’ durability and postseason success. My point is: Of course sabermetrics has its flaws, but accounting for statistics from era to era is ABSOLUTELY not one of them. This, in many ways, is at the very heart of what sabermetrics try to do. This is at the very heart of Bill James’ philosophy about baseball. For countless years, most people judged baseball players in a vacuum. A .300 batting average at Fenway Park was viewed exactly the same way as a .300 batting average at old Memorial Stadium in Baltimore.. A 2.30 ERA at Dodger Stadium was viewed as being better than a 2.60 ERA at Wrigley Field. Chuck Klein’s .368 batting average at the absurd Baker Bowl in 1933 was obviously better than Yaz’s .301 batting average in 1968.
How do we put all this in context? Right: Sabermetrics. This is why we have such things as OPS+ and ERA+ and a hundred other context-driven baseball statistics. They try to remove layers of nonsense and get closer to the heart of things. Every viable advanced baseball statistic adjusts for era and ballpark and the value of a run and what it takes to win games in that time.
Chuck Dobson had a 3.00 ERA in 1968. That was a 93 ERA+ -- well below average.
Greg Maddux had a 3.00 ERA in 2000. That was a 153 ERA+ -- way, way above average.
There. Accounted for.
There is nothing like a good baseball argument. But to have a good baseball argument, you need both sides to bring with them at least a beginner’s idea of what the argument is about. I remember when I was doing all sorts of work comparing Dan Quisenberry and Bruce Sutter -- I mean, I broke them down game-by-game, inning by inning, it took me weeks and weeks, dozens of spreadsheets -- and at the end people would email me by saying: Did you happen to notice that Sutter had more saves than Quiz? ... Really? Saves? Why didn’t I think of that?
That’s what this is like. Sabermetricians are spending countless hours breaking down the game, developing theories, testing those theories, improving their stats, finding boxscores from every game for 100 years, going through millions of play-by-plays, probing every premise ... and then someone comes along and says, “Hey, you know, they scored more runs in 1995 than in 1968, you might want to consider that.” That doesn’t help.