OPS+ and ERA+ are on completely different scales

ERA+ has issues. This was brought to my attention awhile ago by Sky Kalkman (see the Storify for lots of gruesome details). I’ve written about some of the issues with ERA+ as they relate to Rick Reuschel. But today I want to bring up something different.

I was looking at the ERA+ career leaders (you know, because Mariano Rivera). I found this:

That’s interesting, no? I started to wonder:

Further:

Kind of crazy, no? This trickles down, too. David Cone received no Hall of Fame support. His ERA+ of 121 was actually tied for 46th among pitchers with 2500+ innings. If that was his OPS+, how would it rate among position players with similarly long careers (7500+ PA)? How about tied for 142nd with the likes of Harold Baines, Chili Davis, Al Oliver, and Chet Lemon?

So, we need to remember that while an OPS+ of 120 over a long career is impressive, an ERA+ of 120 over the same career length is downright legendary.

Leave a Reply

42 Comments on "OPS+ and ERA+ are on completely different scales"

Notify of
avatar
Sort by:   newest | oldest | most voted
Dr. Remulak
Guest

This bolsters the case for Andy Pettitte, Hall of Famer. Aside from 256 career wins, >100 games over .500, plus 19 postseason victories, Andy’s ERA+ ranks ahead of numerous HOFers, including, among many others: Eckersley, Bunning, Carlton, Jenkins, Niekro, Roberts, Bender, Hoyt, Ryan, Ruffing, Sutton, Galvin, and Wynn.

Lawrence Azrin
Guest
All of the pitchers you named except for Eckersley and Hoyt pitched considerably more innings than Pettitte (in many cases a LOT more), and all of them besides Sutton and Hoyt had greater peaks than Pettitte,in some cases MUCH greater peaks. Eck is in the HOF for brilliance in his role as the first one-inning-only “closer”, and Sutton for his career totals (324 wins; 3rd in games started; 7th in innings pitched and strikeouts). Hoyt is a borderline choice who probably doesn’t belong, and likely got in via the Veteran’s because of his Yankees association (sound familiar?). Pettitte has a… Read more »
baltimorechop
Guest

Sutton, Ruffing, Hoyt, Ryan, Galvin and Wynn all still have a worse ERA+ around 3300 IP. Bunning, Carlton, Jenkins, Niekro, Roberts. Bender & Eck didn’t hit 3300 IP.

Just fact checking, no opinion on this one.

Lawrence Azrin
Guest

@9/baltimorechop,

Thanks; serves me right for not checking. I still think Pettitte has a worse HOF case than all the other pitchers named in #1, except for Hoyt and maybe Bender. Anyone disagree?

Bender gets extra points from me for being the name of the amoral robot in Futurama. No connection to baseball but it’s my system.

Ed
Guest

Lawrence – I think you could put Pettitte ahead of Ruffing and Wynn as well. And about even with Bunning (same WAR, Bunning has higher peak, Pettitte did it in fewer innings).

But I agree with your broader point (i.e, that Pettitte mostly looks good in comparison to guys who probably shouldn’t be in the HOF).

Lawrence Azrin
Guest

Also,the BBWAA hasn’t been choosing a lot of pitchers lately. In the past twenty years, there were (not counting Negro Leaguers) seven pitchers elected by the writers.

Five were starters, and all but one had 318 to 329 wins. The only non-300 game winner, Burt Blyleven is probably considered by most people here more qualified than Pettitte.

Dr. Remulak
Guest

ERA+: Bert 118, Andy 117. Pretty similar.

mosc
Guest
I think most of the writers would chose Pettitte over Blyleven actually. I think Blyleven got lucky because he got in before Clemens/Johnson/Maddux/Glavine/Martinez/Smoltz/Schilling all descend on the ballot. Pettite will be after all those guys and because of that has little chance. Neither does Mussina, same explanation. They’ll probably squeeze in Jr, Thomas, and Biggio in there as well, with lots of “steroid-this” and “steroid-that” for the next 10 years. Pettitte admitted HGH use too. I think he’ll get >5% for 15 years don’t get me wrong and may or may not have a shot at the 2033 ballot a… Read more »
Mike L
Guest
Interesting. It may tell you something about the way we value players. If a pitcher can’t get outs he’s not going to get innings for any extended period of time unless there’s something very unusual about his perceived potential (Luke Hochevar) or his contract (Barry Zito). They will wash out of the system and their poor performance will not be as much of a drag on the averages. But a position player is different. He can be a Brendan Ryan, or he can be great with a pitching staff, or be a versatile utility player, or he can be fast.… Read more »
Doug
Guest

Mike,

Definitely think this is part of the answer. To your point, there are only 60 pitchers since 1901 with ERA+ of 90 or less and 1000 IP. But, there are 402 hitters with OPS+ of 90 or less and 3000 PA.

Other part is that since we’re basically only talking about starting pitchers, you would expect about 3 times as many hitters as pitchers at any point on the scale. The additional bit is as you point out: bad hitters can have long careers if their defense and/or speed are sufficiently valued, but much less likely for pitchers.

CursedClevelander
Guest
And of course, we value batters and pitchers differently. OPS doesn’t tell you anything about a player’s baserunning, or his defense, or his positional value. Two very different styles of player can end up with an identical OPS (and OPS+, if ballpark context is similar enough), with one being an OBP-guy without much power (say, something like a .280/.400/.420 line) and another being an all-or-nothing power guy (maybe more like .240/.300/.520), and those two identical OPS numbers mean very different things. With ERA, we don’t really care how you arrive there, we just care about the IP and the results.… Read more »
CursedClevelander
Guest
I think even without the adjustment there’s a pretty big difference between OPS and ERA. I mean, OPS has a definite minimum and maximum (range of .000 to 5.000), and nobody has ever put up anywhere near a 5.000 in more than a couple PAs. Meanwhile, ERA has a defined minimum but no maximum, and you get quite a few guys who have super low ERAs in a decent amount of IP and wind up with insane ERA+ numbers (Joba in 2007, Kimbrel in 2010, Coleman and Crain this year…). And of course, when a guy does have a 0.00… Read more »
mosc
Guest
This should be analyzed statistically by looking at the standard deviation from the mean. Obviously 100 OPS+ and ERA+ are the mean so simply give the standard deviation for both and you give the difference in the variation you are talking about. Basically you’re saying the difference between a great and terrible pitcher is less than between a great and terrible hitter. This is very true, obvious, and expected. We get to see plenty of pitchers hit and even the occasional hitter pitch. As much as we like to say otherwise, good hitting beats good pitching. Good pitching, at least… Read more »
Brendan Bingham
Guest
I too thought that looking at Standard Deviations would lend some insight… However, I have just calculated SDs for “qualifying” hitters and pitchers for 2013. OPS+: Mean = 113.1, SD = 23.18, n = 140 ERA+: Mean = 111.1, SD = 23.15, n = 81 A few thoughts: 1) That the means do not equal 100 is probably an indication that qualifiers are as a group above average players. Qualifying hitters are everyday starters, all with 120 or more games played; qualifying pitchers are exclusively starters, all with 25 or more games. 2) The SD values are stunningly similar –… Read more »
mosc
Guest

You have to do OPS+ and ERA+ of a batter or pitcher on a per ab/bf situation. Your average hitter may have an above average OPS+ but the average OPS+ of a batter in an average at bat should be 100. Guys with higher OPS+’s typically get more AB’s so that needs to be factored.

baltimorechop
Guest

I’ve noticed this too, when people will dump on a pitcher for having like a 115 or so ERA+. Of all pitchers with 3000 IP (only 10 HOF starters with fewer IP) only 54 have 115 ERA+.

Of all position players with 7500 PAs, 54 gets you to 137 OPS+. Eppa Rixey is tied for 51st for pitchers, and Reggie Jackson is tied for 48th for hitters. There’s a lot more to value than ERA+ and OPS+, I do realize, just making a connection.

Only 13 pitchers have 130+, 13th for OPS+ is 157 Tris Speaker.

Dr. Doom
Guest

Another issue is that, the more PAs, the closer you move to average. Take Harmon Killebrew and Bert Blyleven. Both played 20 MLB seasons. Killer finished with 9833 PAs. Bert finished with 20491 BF – more than DOUBLE Killebrew’s total. Killer’s 143 OPS+ may be 21% higher than Bert’s ERA+ (though we should really use ERA- and say Bert was 15 percent better than league, rather than 18, meaning Killebrew was more like 25% more effective) but in 48% as many opportunities. Also something to chew on.

mosc
Guest

Almost perfect except clearly there is significant contribution from the other 8 defensive players where OPS is nearly entirely a batter’s own personal stat. I agree with it from a WAR type argument though. WAR for ERA+ and OPS+ per ba/bf would be more similar than ERA+ and OPS+ without compensation, as you suggested.

mosc
Guest

WAR should be WAA there since OPS+ and ERA+ are against league average rather than replacement level.

Dr. Doom
Guest
Yes, there is a significant defensive contribution. But we could use FIP- instead, and assume an average team behind the pitcher in order to account for that, to some extent. I don’t disagree with what you’re saying; in reality the batter has roughly 50% of the credit/blame for the result of a PA, the defense the other half. And credit goes, what, maybe 3:1 pitching to defense? So maybe you just remove a quarter of Blyleven’s BF from the above example… nonetheless, it’s still a much greater number than what Killebrew got in terms of PA (only like 70% more… Read more »
mosc
Guest

Plus, that only tells you about Killebrew’s bat and not his defense, base running, etc. Not trying to argue value in totality a la WAR but to equate OPS+ to ERA+, I think you are on the right track. BF/BA need to be considered.

bstar
Guest

I’m with you on ERA-.

But if we’re going to switch pitching metrics, I’d prefer we go all the way:

-take out the defense (RAdef)
-count the unearned runs (RA9)
-use a strength of opposition measure (RA9opp)
-use RA9role to put reliever/starter RA9 on a more level plane

This is what rWAR does to judge pitching performance. It would be great if someone crunched all these numbers into one metric.

bstar
Guest
Adam, can’t all of this be explained by the fact that OPS+ = (OPS+) + (SLG+) – 1? Suppose a hitter is 20% better than league in OBP, SLG, and OPS. What’s his OPS+? It’s not 120, it’s 140. 20 points for 20% better than league in OBP and 20 points for 20% better than league in SLG. This effectively doubles the difference between league average and player. It rarely works out that perfectly; sometimes a 120 OPS+ player will have 17 points for SLG and 3 points for OBP, or whatever. This is why OPS+ has higher scores, because… Read more »
mosc
Guest

I don’t agree wit this one bstar. OPS at least tries reflects all possible outcomes of a PA. Separating slugging and OBP kind of breaks the stat. OPS gives credit for a walk, single, double, triple, HR, an out, and a sacrifice in rough line with the run probability added for each of those events. Pulling OBP apart from the value added slugging gives you weird results IMHO. For example, slugging is the more valuable part of most of our all time greats than their obp.

bstar
Guest
mosc, thanks for responding, but you’ve got it backwards. Let me be clearer: my first sentence IS the way OPS+ is currently computed. The second method is the way a lot of us think OPS+ is computed. Go to Babe Ruth’s page and click on More Stats and go down to Advanced Batting. Compute the numbers for yourself. Divide RuthOPS by his lgavgOPS, and you’ll get a 1.54 ratio. Multiply by 100, and you’ve got 154. But Ruth’s career OPS+ is 206: OPS+ = OBP+ + SLG+ – 100 OPS+ = {(OBP/lgOBP)x100} + {(SLG/lgSLG)x100} – 100 OPS+ = {(.474/.354)x100} +… Read more »
bstar
Guest

I hit “submit” too early: what that last list proves is that if OPS+ only had one component, the scale of it would look far more similar to the scale of ERA+ than it currently does with two components.

mosc
Guest

I see what you’re saying now. I agree! I just assumed OPS+ was calculated as a total rather than in parts like that. The formula of OPS+=SLG+ + OBP+ -100 is mathematically incorrect, as you said. I’m not sure you can calculate OPS+ from SLG+ and OBP+. It should really be calculated as OPS relative to that year’s league average OPS and park factor correction. Since slugging and on base percentage are not equally weightered in OPS, it does seem to add it up incorrectly.

Dr. Doom
Guest

I guess the question for me is, “which one corresponds to actual run-scoring?” We know that ERA- does. We also know that wRC+ does… and since OPS+ (as it currently is) is nearly identical to wRC+, shouldn’t we assume that the current version, though flawed, is preferable, since it actually corresponds to what we’re TRYING to measure (which is the impact on run-scoring, rather than the relative OPS of the player)?

bstar
Guest

I wasn’t arguing that the computation of OPS+ should be changed. I was just pointing out why ERA+ and OPS+ appear to be on different scales. It’s because OPS+ has two components, added together, and ERA+ only one.

So, yeah, my explanation doesn’t really get into the ramifications of this, which is I think Adam’s main point. How does it affect pitchers on the cusp of the Hall of Fame? Are we undervaluing them because a 115 ERA+ doesn’t look dazzling enough?

I think it’s a good point.

jeff hill
Guest
I think Andy Pettitte is a very good pitcher but not a HOF pitcher. His post season record is nice, but nothing he did there was exceptional. 19-11, 3.81, 1.305 WHIP, 9.3 H/9. We all know how overrated W/L records are and how completely stacked those Yankee teams were. He only had 3 seasons where his WAR reached over 3.8 and his 1.35 WHIP is pretty average, I’m being generous here(he’s tied for 708th…Aaron Harang is 719). He finished with 9 years having an ERA over 4 and only 3 years with an ERA under 3. Yes, his 100 games… Read more »
Mike L
Guest
On stats alone, I agree with Jeff; Pettitte is not a Hall of Fame player–irrespective of the PEDs. But he is an archetype of the player GM would love to pencil into his roster for the next 10-15 years; durable, consistently above average, a gamer, someone who isn’t a #1 but you would be comfortable putting out there in a big spot. We have been talking about ERA+, and Pettitte has only one season in 18 below 100 (ERA+ of 97 when he was 36 years old) and positive WAR every year. That’s real value, superstar or not. I agree… Read more »
no statistician but
Guest
The discussion has kind of gotten away from Adam’s point—or points—I would say. As to his last comment about ERA+ being, in essence, a much more rigorous measure for pitchers that OPS+ is for batters, my own feeling is that WE are the ones who make the mistake of regarding the two as equivalent. ( I’d go further and say that it is a mistake to equate WAR for pitchers and WAR for batters, but that’s a subject for another time.) Here’s my view—I think. Maybe. What we’re really getting back to here is the enticing but futile search for… Read more »
mosc
Guest
Well, I think WAR is a holy grail stat. Nobody should disagree with the value of WAR but I question anybody who thinks we calculate it accurately. WAR hopefully gets refined and improved as our understanding does. Now, ERA+ and OPS+ on the other hand are very much no WAR and do not imply WAR. Pitching WAR is very proportional to ERA+ and IP which makes it closer to what we think of as WAR than, say, OPS+ but neither is WAR. But maybe your point is that OPS+ is further correlated from WAR than ERA+ from a statistical point… Read more »
birtelcom
Editor
Top 10 “OPS+ Against” in MLB in 2013, min. 162 IP (ERA+ is in parens): 1. Jose Fernandez 48 (176 ERA+) 2. Clayton Kershaw 51 (194) 3. Matt Harvey 55 (157) 4. Max Scherzer 57 (145) T5. Stephen Strasburg 66 (126), Anibal Sanchez 66 (163), Yu Darvish 66 (145) T8. Chris Sale 71 (140), Madison Bumgarner 71 (120) 10. Hisashi Iwakuma 78 (138) Bartolo Colon is in the top 10 in ERA+ (his ERA+ of 141 is 7th best) but his OPS+ Against of 85 falls out of the top 10 in that category. Ditto for Zach Greinke (10th in… Read more »
mosc
Guest

I don’t even understand how you could calculate OPS+ against…

Paul E
Guest

Adam:
Our link here on this site directly to baseball-reference is certainly convenient; however, it sure sounds like your making a case for wRC or wOBA on Fangraphs or Baseball Prospectus or one of those other sites 🙁

Jim Bouldin
Guest

I have no idea who “Sky Kalkman” is but it sure would be nice if he labeled his graph axes and explained whatever it is he’s trying to illustrate.

Brendan Bingham
Guest

Jim,
I agree; SK’s graph is confusing. As far as I can tell, the unlabeled X-axis is ERA. If so, then when SK says ERA+ is not linear, what (s)he means is that ERA+ is not a linear function of ERA.

Richard Chester
Guest

The x-axis is ERA and the assumption is that the league ERA is 5.00. If a pitcher has an ERA of 2.50 the y-value for the blue (ERA+) line is 200.

Jim Bouldin
Guest

OK, thanks guys.

Jim Bouldin
Guest

No hold on, the issue here is much simpler, and yeah, there really is a problem here. It’s because ERA+ reverses the directionality of the ERA axis (low=bad, high=good), and (apparently) they do so by inversion, because that’s the only way to get a hyperbolic shape like that. ERA- must be an ERA tranformation in which the directionality is preserved–that’s why it’s perfectly linear.

Another adventure in ad-hockery in the sabermetric world. Add it to the pile.

Jim Bouldin
Guest
The reason that “OPS+ and ERA+ are on completely different scales” is captured in different fragments of comments 3,4,6, and 18. One issue is that the underlying stats (ERA and OPS) are fundamentally different in their distribution; ERA is based on runs scored, and therefore will have a long tail to the right, a skew distribution, the underlying reason being that all games start 0 to 0 and scoring runs is inherently a rare event relative to the length of a game. OPS conversely is a sum of averages (a questionable thing, but that’s another story), of a process in… Read more »
wpDiscuz