Who You Face Matters
The beauty behind the philosophy of advanced analysis is that it seeks to eliminate as much variance as possible. As simple of a thought as this may be, it is one that still eludes the majority of the baseball world, and most of society in a variety of other areas. Our metrics here at Fangraphs do seek to base value on much more that raw numbers. We can not only adjust for league, but also park and era, among other variables.
Unfortunately, there seems to be a limit to this. As much as we know Dan Haren gets unlucky for pitching at Chase Field (or the opposite for Dante Bichette at Coors), or that pitchers in the late 60s were helped by a higher mound (or the opposite thanks to steroids in the late 90s), we still haven’t found a way to, somewhat literally, level the playing field in terms of whom a pitcher or hitter faces. Here’s an example:
Tim Lincecum: 6.4%
Bronson Arroyo: 7.4%
Jonathan Sanchez: 5.3%
Wade LeBlanc: 6.0%
Mike Leake: 8.2%
Those are the HR/FB rates of the pitchers in baseball who have faced the worst opponents in baseball sorted by OPS. Here are some more:
Josh Beckett: 13.0%
Joel Pineiro: 10.9%
Jeremy Guthrie: 8.6%
David Huff: 10.1%
Mitch Talbot: 7.8%
Ben Sheets: 12.0%
Those are the starting pitchers who have faced the best hitters based on OPS this year. I didn’t run a full study, but I would think that the correlation between HR/FB and the OPS of opposing batters is decently high. This is logical and intuitive: better hitters in baseball have better HR/FB rates, so if you face more of them you’re likely to feel the effects (and vice versa). When you see a pitcher have a bunch of years of giving up HR/FB rates either above or below average, you may want to believe it is more of a “skill” than a “trend.” But said pitcher may have just been facing competition the whole time that would dictate the results, and with a little bit of luck added in, it looks like a trend.
But what does it mean? It means we shouldn’t just think of things like HR/FB and BABIP as a pitcher getting “lucky” or “unlucky” based on the quality of the balls in play, but also by the quality of the opponents. Tim Lincecum‘s opponents have an OPS of .675 this year. For reference, that’s about 2009 Randy Winn, who had a wOBA of .302. Josh Beckett’s opponents have an OPS of .767 this year. That’s roughly 2010 Chipper Jones, who has a .349 wOBA.
Luckily, that’s as big of a difference as you’ll generally find. However, sorting out even the most minor differences has some significant value. I don’t have a panacea, but it’s something we should keep in the back of our minds when analyzing players. It often goes overlooked.
Pat Andriola is an Analyst at Bloomberg Sports who formerly worked in Major League Baseball's Labor Relations Department. You can contact him at Patrick.Andriola@tufts.edu or follow him on Twitter @tuftspat
But how much can you influence the hitters that you face? You expect good pitchers to have low OPS against them, deflating the OPS of hitters they face. In a division like the NL west (where 3 of those pitchers come from), with good pitching all around, I wonder if there is a dampening effect on the OPS of hitters the pitchers face.
Sure, there’s a certain level of chicken and egg involved, of course. But Guthrie is in the AL East, where the pitching is great, but it doesn’t stop the hitters from being great also. Pineiro is in the AL West, where the hitting stinks, but he’s faced some good hitters over the course of the season thanks to schedule variance.
Pitchers don’t just face their own divison/league as well. The majority, but not an extreme.
One more reason to do away with the imbalanced schedule.
I thought that the reason that pitchers in the 60s were better was due in part to a higher mound not a smaller one. Reason being that they were pitching at a better angle relative to the batters.
Yeah, this is why Fangraphs needs to hire interns for editing purposes 🙂 Fixed, thanks.
Very good post. This is something that definitely does not get enough consideration. I would love to see a thorough study of this sorta thing.
This is one of the reasons it’s always dangerous to rely on small sample sizes. Everyone of those players will regress to the mean before the season is over (or, bet on it at least.)
Has anyone ever tried to develop a “league difficulty factor”?
Agreed, this is something obvious that is seldom discussed. I would think it’d perhaps be more accurate to use OPS of hitters faced subtracting the starts of said pitcher you are evaluating and compare that way. Pain in the ass granted, but may be a better picture of the quality of opponents.
Although it’s certainly interesting to consider while looking at mid-season statistics, isn’t it something that generally evens out over the course of the entire year? Sure, Lincecum will end up facing some batters more often than Halladay due to factors like unbalanced division play schedules and luck (e.g., rotation, rainouts), and will face some unique batters due to interleague play or possibly even injuries. But it seems that the biggest variances found in the midseason statistics would become less and less over the course of 162 games.
In 2009, Todd Wellemeyer faced an average hitter of a .707 OPS. Ubaldo Jimenez was at .717. Joel Pineiro was at .718.
in 2009, David Price faced an average hitter of .776. Roy Halladay was at .770. Felix Hernandez was at .759.
These have to do with differences in league, and most people fall near league average, but if we can throw a grain of salt on those near the extremes we can be doing a bit of good for proper analysis.
great piece. at the risk of sounding stupid, whats the easiest way to access a players’ OPS-against numbers? is it hidden somewhere on the fangraphs’ player page? sorry if this is abundantly obvious, i just cant figure it out.
I’m pretty sure that BPro’s stats are available without subscription and they have quality of opponents faced for every hitter and pitcher. The interface is a little old and clunky. I could be wrong about those being pro bono info.
Derek Carty at The Hardball Times introduced Context Adjusted Pitching Statistics (CAPS) last year to adjust for quality of opponent. He then did a few player profiles and I believe he mentioned that he was going to make the all of the numbers available at some point but I haven’t heard a thing about them since the middle of last summer. Here are the articles that I could find.
http://www.hardballtimes.com/main/fantasy/article/introducing-quality-of-opponent-adjustments-and-caps-for-pitchers/
http://www.hardballtimes.com/main/fantasy/article/caps-profile-aaron-harang/
http://www.hardballtimes.com/main/fantasy/article/caps-profile-gil-meche/
http://www.hardballtimes.com/main/fantasy/article/the-best-pitcher-of-2009-is/
Pitchers in the 60’s didn’t have ballparks made for homeruns, higher mounds, the DH, umps who thought they were bigger than the game(but gave guys like Koufax and Seaver most calls.)
Where you pitch matters as well… look at Vasquez last season with the Braves compared to this year. Or guys like Darryl Kile and Mike Hampton who went to Coors Field when it made guys like Dante Bichette and Ellis Burks look like MVP caliber players.
“pitchers in the late 60s were helped by a higher mound (or the opposite thanks to steroids in the late 90s),”
The latter part of this sentence is ill-considered. It assumes one of numerous potential reasons for the offensive explosion of the period, and it also assumes pitchers were not using PEDs, when we have plenty of evidence to indicate otherwise.
I don’t mean to sidetrack from an otherwise excellent post, and obviously the point is the same, but these kinds of things stick in my craw a little. 🙂
When writing it, I was thinking, “Is this 100% true? Eh, the readers will get the point nonetheless.” Thanks for keeping me honest. I still think it’s probably true, but hey, that may be another post!
Well, we know Bonds, McGwire, Matt Williams, Ken Caminiti, Jason Giambi, Miguel Tejada, Sosa, Ortiz, Man-Ram, A-Rod, Bret Boone all used… the only pitchers we know for sure are Clemens, Pettitte, Kevin Brown, and Eric Gagne(the others are fringe pitchers for the most part.)
I’d say blaming steroids to an extent is completely fair. They have made it far easier for hitters with game footage(hitters couldn’t research pitchers like Koufax back in the day like it is now, lowered mounds, sandbox ballparks, crappy umpires who squeeze the best pitchers in the game and have taken away the pitcher’s right to throw inside…
I don’t think pitchers using roids could ever get the edge back… Maddux and Trevor Hoffman didn’t need them.
” crappy umpires who squeeze the best pitchers in the game and have taken away the pitcher’s right to throw inside…”
For every umpire like that, there’s one calling curveballs 6″ off the plate strikes.
Sounds like OPS Against should be added to the stats/leaderboards?
Or for it to count in the formulas measuring how valuable an individual player has been.
Rich, I know… the umpires are horrible. But I’ve actually seen Halladay and Verlander have about 4 or 5 easy strike calls that would have been strike 3 not called.
It didn’t cost Halladay, but Verlander ended up losing.
Halladay also benefitted from 3 rather obvious calls in his perfect game (2 of them woul have been immediate walks).
Star pitchers get far more leeway than they do squeezed.
I have been an advocate of this for a long time. I thought it was obvious even when Voros first theorized this stuff. the true out comes and especially HR/FB rates are based on the idea that they even out as competition varies. But if a pitcher is facing an unusually low amount of competition it makes sense that his BABIP or HR/FB rates may get higher. Its not the only reason but especially when a player is outperforming this is an explanation that usually gets factored into luck(luck is essentially circumstance when you get down to it). Bottomline what FIP, DIPS and the like have been saying is these are things pitchers control the most but there are also factors hitters have some control as well. They often times get close to evening out but sometimes they don’t.
Its something to look at. Its also one step toward helping people understand FIP and XFIP what the pitcher can control and what they can’t. The hitters, ballpark and the like do influence the game they just usually balance out a bit over a long season. If the pitcher doesn’t control these outcomes then something must either ballpark, hitter, defense. We need to recognize that more, I know a lot of people do but sometimes even the best of us forget the assumptions that build the foundation of stats like FIP.
I think that research has shown that a juiced ball is MUCH more likely to have been the cause for the offensive explosion than steroids. The explosion occurred over two years and has lasted consistently since then, for steroids to have been the main factor, 25% of players would have had to have taken steroids, and their performance because of the steroids would have needed to improve by over 50% if I remember correctly from using the steroids. That’s a mite ridiculous I’d say.
I think the problem we’re running into here is that this is one of those times when it’s easy to get bogged down in minutia. As far as I understand these pitching stats really exist for three reasons, they’re predictive, they can be used to observe the difference between “actual performance” and “skill”, and they inform our calculations of value.
The problem with getting this detailed is that you can’t stop at OPS. One guy is going to do very well in a good hitting division because of his skill set, if he doesn’t walk guys and gives up few fly-balls the effect of a power hitting league will be less than that of a better pitcher who gives up more walks but strikes alot more guys out while giving up more fly-balls. So you’d have to make the adjustments at a skill by skill basis for each pitcher. With the amount of turnover year to year you’d lose all predictive value by doing this. In addition, better hitters will affect ERA and xFIP more or less equally, so it won’t really change our analysis of how much he over or under performed. As for value, it won’t tell us any more about how much he actually did to improve his teams chance to win. It’s interesting, but not something that really invalidates current theory or really will even improve how we evaluate except at a very detailed level, where we’re making that analysis anyway.