At WAR with Pedro
Somebody asked me this question on Twitter: If I could have any pitcher from any time pitch one game (say a seventh game of the World Series or the ubiquitous “pitching for your soul” scenario”) who would I choose?
I immediately typed: Pedro. 1999.
This is always my fallback position. Back in the Trivial Pursuit days, my mother would guess “Babe Ruth” on pretty much every sports question. She has actually become much more knowledgeable about sports, in part because of this mess of a blog, but back then it was always “Babe Ruth,” even on, you know, billiards or horse racing questions.
And that’s how I am with Pedro Martinez’s 1999 season. Any baseball pitching question can be answered, somehow, by: Pedro, 1999. I would actually like to answer ALL questions that way. When I go fill up gas, and the little pump screen asks: “Cash or Credit” I’d love to be able to type in: Pedro, 1999.
Anyway, the choice lit up the Twitter lines with the expected objections — what about Bob Gibson in ’68 or Randy Johnson in 2001 or Walter Johnson in 1913 or Sandy Koufax in 1965.* You could make an argument for those and a couple dozen more — Carlton in ’72, Gooden in ’85, Grove in ’31, Hershiser in ’88, Mathewson in ’08, on and on.
*Am I the only one who gets kind of annoyed when people put some sort of finality stamp at the end of their opinions? You know what I mean by finality stamp — someone will not just say “Sandy Koufax in 1965 was quite sprightly.” No, they will say something like “Koufax. 1965. End of story.” Or: “Gibson. 1968. The end.” Or: “Carlton. 1972. Period.” Or: “Old Hoss. 1884. Goodbye.”
What are these emphatic termination words supposed to achieve? I mean YOU put those words there, right? I didn’t miss some mediator coming in and ending declaring your viewpoint supreme, did I? It’s not like you pulled Marshall McLuhan out of nowhere to confirm your opinion … YOU confirmed your opinion. How does that mean anything? Is this like the Internet equivalent of taking off your shoe and clomping it on the table like a gavel? Stop doing that. It’s stupid. Period. End of story. Goodbye.
Anyway there was one alternative to Pedro 1999 suggestion that I found interesting for a completely different reason.
The suggestion: Pedro in 2000.
This post is not actually about Pedro Martinez, not specifically, but about WAR. As I assume everyone reading this blog knows, there are two prominent variations of the statistic “Wins Above Replacement.” There is Baseball Reference WAR. And there is Fangraphs WAR. Best I can tell when it comes to everyday players, the two systems are fairly similar — any real variations on players’ totals probably comes down to how defense was calculated.
But the two calculate pitcher’s WAR differently and this might be seen mostly clearly in Pedro Martinez’s 1999 and 2000 seasons.
Martinez made the same number of starts and threw roughly the same number of innings both seasons, which is helpful comparison purposes. In 1999, Martinez threw 213.3 innings. In 2000, he threw 217 innings.
The other numbers, though, are quite different:
1999: 23-4, 2.07 ERA, 5 complete games, 1 shutout, 160 hits, 313 Ks, 37 walks, 9 homers.
2000: 18-6cfi, 1.74 ERA, 7 complete games, 4 shutouts, 128 hits, 285 Ks, 32 walks, 17 homers.
OK, before diving in, here is what Baseball Reference WAR says:
1999: 9.7 WAR
2000: 11.7 WAR
So Baseball Reference has Pedro’s 2000 season worth two extra wins.
Here’s what Fangraphs WAR says:
1999: 11.9 WAR
2000: 9.9 WAR
And it’s almost precisely reversed — Fangraphs has Pedro’s 1999 season worth two extra wins.
Obviously both seasons are all variations of awesome and we’re just picking between Reese’s Peanut Butter Cups and Mini Reese’s Peanut Butter Cups. And you probably know exactly why the two Pedro Seasons are calculated differently but let’s go step-by-step here.
Baseball Reference WAR values the 2000 season more because Pedro Martinez gave up fewer runs and fewer hits. It’s a bit more complicated than that, but that’s at the crux of things. Martinez’s ERA+ in 2000 was 291, which is the record for a season. In 1999 his ERA+ was merely an otherworldly 243.
So that’s at the heart of Baseball Reference’s process — Pedro Martinez gave up 11 fewer runs in 2000 (largely because the league hit an almost unbelievable .167 against him, 38 points less than the year before) and that meant it was a clearly better season.
Fangraphs WAR, meanwhile, doesn’t deal with ERA. It deals with the three things that Fangraphs believes a pitcher can control: Strikeouts, walks and home runs. In 1999, Martinez struck out an obscene 13.2 batters per nine inning (just behind Randy Johnson’s 2001 record) and he hardly walked anybody and, perhaps most overlooked, he gave up NINE HOME RUNS all season. Both of these seasons were smack in the middle of the Selig Era, when home runs flew like confetti, and to give up nine homers all year …
Well, let’s look at the top five in the AL that year in homers per nine innings:
1. Pedro, .380
2. Mike Mussina, .708
3. Freddy Garcia, .805
4. Omar Olivares, .831
5. Jamie Moyer, .908
That doesn’t look very close, does it?
Anyway, of the three things at the heart of the Fangraphs process he did two of them (strikeouts, home runs allowed) better in 1999 and the third (walks) was more or less a wash. So that’s why Fangraphs thinks 1999 was a clearly better season.
What makes this cool, though, is that it’s a great way to decide exactly which kind WAR speaks loudest to you. Which season do YOU THINK is better? If you think the 2000 season was better, then you are probably a Baseball Reference person. If you think 1999 — you’re Fangraphs.
I asked Tom Tango what he thinks and, as usual, he came up with an interesting way of looking at things. Looking at it another way, the question in play is this: How much control do you think a pitcher has on balls hit in play — yes, we’re crossing back to the famous BABIP (Batting Average on Balls In Play).
If you think a pitcher has COMPLETE CONTROL over balls in play then you will naturally think that Pedro was a better pitcher in 2000 when his ERA and hits allowed were much lower. The BABIP numbers could not be more stark.
— In 1999, despite his dominance, Pedro allowed a .325 batting average on balls in play — which was actually the FIFTH HIGHEST in the AL.
— In 2000, Pedro allowed a .237 BABP — which was the LOWEST in the AL.
So, if you believe a pitcher has complete control over balls put in play then you will believe that Pedro Martinez learned a whole lot between the end of the 1999 season and the beginning of 2000.*
*I believe it is this Pedro gap, by the way, that helped inspire Voros McCracken to come up his theory about pitchers not having control of balls hit in play.
OK, but if you think a pitcher has ZERO CONTROL over balls in play, then you will definitely believe that Pedro was a better pitcher in 1999 and was just a whole luckier in 2000 (or had a team that played much better and smarter defense, which is in a way the same thing for a pitcher).
What Tango says — and I concur — is that it’s likely neither absolute is true. It’s likely that pitchers do not have complete control on balls hit in play, and it’s likely that pitchers are not entirely powerless.
“Since reality is somewhere between the two … we get into our conundrum: must we take a 0/100 approach to everything we track?” Tango asks. “Or, can we start to give partial credit? … No one likes the idea of partial credit, because it implies a level of precision that we can’t possibly know.”
Tango comes down closer to the side that a pitcher has limited control over balls in play. I again agree. I think there will still be studies and thought experiments that get us closer to that relationship between pitching and defense, but right now I lean just a touch more to the Fangraphs side. I think Pedro pitched a little bit better in 1999 than he did in 2000. That 313-37 strikeout to walk ratio is just absurd. Those nine home runs allowed, wow. I don’t think he was a full two wins better. But one game — we’re talking one game — I’m taking that Pedro Martinez in the middle of the Selig Era who didn’t let the ball in play much, who always kept it in the ballpark and who was good for 13 outs a game on his own.
And Tango? Well he says Baseball Reference and Fangraphs give us the extremes … and the answer, almost certainly, lies in the middle. And this is why Tango, when looking at Baseball Reference WAR, at Fangraphs WAR will split the difference.
This would make Pedro’s 1999 and 2000 seasons almost EXACTLY EVEN.
Which, if you think about it, is a good way to end this. Period.