Putting Pedro Martinez’s Minus Stats in Context
Although Pedro Martinez may not have had the longevity or durability of some of baseball’s other pitching greats, there is little doubt that his peak years were some of the best, if not the best, that any pitcher has ever produced. With the introduction of the “minus stats,” ERA-, FIP-, and xFIP-, we have yet another tool with which to put these fantastic years in context. Here’s a look at Pedro’s 1999-2003, with the reminder that 100 is average and unlike the “plus stats,” (OPS+, wRC+, etc.), lower is better.
| Year | ERA- | FIP- |
|---|---|---|
| 1999 | 42 | 30 |
| 2000 | 35 | 46 |
| 2001 | 52 | 36 |
| 2002 | 50 | 51 |
| 2003 | 48 | 49 |
Of course, without context, those are just numbers. Beautiful numbers, but still just numbers nonetheless. To truly understand how legendary Pedro’s season was, we need to compare it to a number of other storied pitching performances and see how they compare.
Let’s start in Pedro’s own era. Martinez certainly wasn’t alone in pitching greatness, as Roger Clemens, Greg Maddux, and Randy Johnson – all easy first-ballot Hall of Famers – often rivaled him for the Cy Young award. Johnson’s total of six sub-60 ERA- seasons and two sub-50 FIP- seasons are impressive, but don’t compare to Pedro’s ’99-’01. Clemens never posted a FIP- below 50 or an ERA- below 45 (a number he achieved twice, ’97 and ’05). Maddux may have had a chance to best Pedro’s 2000 ERA- of 35 in either 1994 or 1995, both strike-shortened seasons in which Maddux posted a 37 and 39 ERA- respectively. However, Maddux never struck out enough batters to approach Pedro’s FIP- marks, never bettering a 53 in 1994.
Although the whole careers of the above pitchers compare very well to Pedro’s, at least by ERA- and FIP-, their peaks come up short. With that in mind, let’s look back into the annals of baseball history at some of the more legendary pitching seasons of all time.
As far as pitchers without longevity but fantastic peaks go, it’s hard to top Sandy Koufax. Koufax put up four of the best consecutive years in MLB history from 1963 to 1966, compiling at least 7.8 WAR (according to Baseball-Reference) each season. His best years bookend this stretch: 1963 and 1966 both resulted in 10.8 WAR. However, Koufax “only” managed ERA- marks of 62 and 53 and FIP- marks of 62 and 67 respectively. Of course, although Koufax didn’t quite dominate the league like Pedro did, Koufax’s season spanned over 300 innings, while Pedro’s hovered around 200, another difference in the two eras.
Finally, here’s a look at some of the best individual pitching seasons of all time.
Bob Gibson, 1968: 1.12 ERA, 1.78 FIP, 38 ERA-, 65 FIP-
Steve Carlton, 1972: 1.97 ERA, 2.01 FIP, 55 ERA-, 60 FIP-
Dwight Gooden, 1985: 1.53 ERA, 2.13 FIP, 44 ERA-, 59 FIP-
Walter Johnson, 1913: 1.14 ERA, 1.89 FIP, 39 ERA-, 62 FIP-
Gaylord Perry, 1972: 1.92 ERA, 2.50 FIP, 60 ERA-, 74 FIP-
Bob Feller, 1940: 2.61 ERA, 2.88 FIP, 63 ERA-, 72 FIP-
And just for kicks, a couple of recent ones.
Zack Greinke, 2009: 2.16 ERA, 2.33 FIP, 47 ERA-, 52 FIP-
CC Sabathia, 2008 (just with Milwaukee): 1.65 ERA, 2.44 FIP, 39 ERA-, 57 FIP-
Roy Halladay, 2010: 2.44 ERA, 3.01 FIP, 60 ERA-, 74 FIP-
Although with the older pitchers, their performances were coming in many more innings pitched (usually around 100 more, to be specific), nobody was able to outclass the rest of their league in the way that Pedro did from 1999 to 2003. Even matching one of Pedro’s three seasons from 1999 to 2001 is nearly impossible. Honestly, given how fungible pitching is and how ridiculously talented Pedro Martinez was, I would be quite surprised if I see such a tremendous peak again in my lifetime.
Jack Moore's work can be seen at VICE Sports and anywhere else you're willing to pay him to write. Buy his e-book.
Am I the only person slightly thrown off by the use of minus instead of plus here? I understand that it’s trying to capture the “lower is better” nature of ERA and FIP, but I think most all of us are at least passingly familiar with ERA+, and I know I’ve certainly become accustomed to comparing players on indexed stats where more is better, regardless of how the raw stat behind it works. Maybe it’s because the positive extremes are bounded here instead of the negative extremes. Pedro’s ERA- of 35 in 2000 just doesn’t have the same sort of oomph that his ERA+ of 291 does.
I’m not thrown off at all. I can see what you mean, but it seems just as intuitive to me to have lower be better.
The part that is annoying is that Pedros FIP- of 30 in 1999 means he was 70% better. 30 to 70 is easy enough to do on the fly, but his FIP- of 36 in in 2001 requires a little more thought to translate over to 64% better.
You can call me a simpleton for not wanting to have to do simple math in my head, but it just feels awkward to even need to. A stat should say what it wants to tell you, not bring you 98% of the way there and make you finish the rest.
Except I don’t think that’s the case. According to ERA+, Pedro was 191% better than league average in 2000. His ERA- isn’t saying he’s only 65% (100-35) better than league average.
@ Kevin
The reason that I like ERA- better is that the denominator does tell us how much better the player was than the league. ERA+ tells us how much worse (or better) the league was than the player. Pedro’s ERA+ tells us that the league was 191% worse than him, which is just a weird way to think about it. It’s not wrong, just weird.
@Kevin
ERA+ doesn’t work like that. When you have a player ERA of 1.74 and a lgERA of 5.06, then the player is (5.06-1.74)/5.06 (65%) better than average. The key is that the 5.06 is in the denominator. When making that calculation with ERA+, the player’s ERA is in the denominator.
This is why a 291 ERA+ does not imply an ERA 191% better than average, but a 35 ERA- implies an ERA 65% better than average.
I’m not sure if there’s some sort of other advantage that we’re not seeing here, but I generally agree with you.
I agree with Matt, interesting new tool, but really isn’t as practical to me.
I think what’s cool about ERA- or FIP- is that it should be on the same scale as wRC+ for rating batters. So… in 2000 Mike Lansing was the worst hitter in MLB with a wRC+ of 48. Pedro made the average hitter he faced worse than Mike Lansing. That’s pretty cool.
Except it’s not really the same thing. You’d need to know Pedro’s wRC+ against… which is actually something I would like to know.
No, it isn’t the same thing, but I think together ERA- and FIP- are equivalent and better than wRC+ against. Equivalent because it’s at the same scale. Better because it’s better suited to measuring pitchers.
wRC+ is an estimate of how much a hitter contributed to offense, per-PA. It’s needed because hitters don’t score runs all by themselves. But starting pitchers finish most of the innings they start, so we don’t need to figure out how many runs they gave up from the hitting events, we can take it straight from the box score, and then make ERA- from it. Why look at an estimation when you have the real thing?
If you’re not interested so much in how effective the offense was against a pitcher, but rather how well the pitcher pitched, FIP- measures just the things the pitcher has the most control over, combines them and adjusts them to the same scale. There are probably better pitcher evaluation numbers out there than FIP, but FIP is pretty good. Why use a hitter’s estimation when you have a pitcher’s estimation?
It’s how your phrased the question. If you want to know what he made the hitters look like, you need to use the hitters’ stats. I wasn’t saying wRC+ against was a better way of evaluating a pitcher, I just thought it would be a cool thing to know.
I’d probably prefer it to be on a negative and positive scale, with 0 the average and the distance from zero being the percentage from average (with negative being worse, and positive being better).
I’m not sure why scaling around 100 (+ or -) is better than 0. It could be aERA or @ERA or |ERA| or ERA& or whatever notation to represent “absolute” ERA, as the number is your “absolute” percentage from average, with direction applied by – or +.
That makes good sense