Andy Sonnanstine and Situational BABIP
On the list of things I thought I’d be writing about today, Andy Sonnanstine ranked somewhere just south of nuclear physics. He was among the worst pitchers in baseball last year, running a 7.09 FIP in 36 innings at the big league level, leading the Rays to non-tender him yesterday and put his Major League career in jeopardy. He’ll probably be able to land a minor league deal from some team that needs Triple-A innings, but there’s a pretty decent chance that he’s seen his last days in the big leagues.
So, why am I writing about him? Well, thanks to the rabbit hole that David Appelman has created with his custom reports features on the leaderboards (seriously, if you’re not using these, you’re really missing out), I ended up spending a decent amount of time this morning looking at some of the differences in how pitchers perform in various situations. What led me down this path?
Well, when looking at Sonnanstine’s player card, I noticed that his career LOB% was an incredibly low 65.5%, the kind of number that is almost always driven by a relatively terrible performance with men on base. Some pitchers just do not perform well pitching from the stretch, so the gap between their bases empty performance and their men-on-base performance can be stark. Indeed, when looking at Sonnanstine’s splits, this is exactly what we find.
With the bases empty, opposing batters have hit him for just a .257/.301/.439 mark, but with men on base, the slash line jumps to .329/.375/.528, thanks mainly to a dramatic spike in the BABIP he allows. It’s hard to keep runners from scoring when you give up a .900 OPS to opposing batters when there are runners to drive in, and so Sonnanstine’s results have been a lot worse than looking at his numbers in a context-neutral environment would suggest. Given that he’s a guy with extremely fringey stuff anyway, it’s not hard to believe that Sonnanstine just loses enough velocity and movement without the benefit of the wind-up that he turns into a pumpkin in those situations, but I felt like we needed a baseline to compare his situational splits to.
So, I pulled up the league averages for performances with the bases empty, men on base, and runners in scoring position. Here are the relevant numbers over the last 10 years (note that BB% here excludes intentional walks):
| BB% | Empty | Men On | RISP |
|---|---|---|---|
| 2002 | 7.7% | 7.5% | 8.5% |
| 2003 | 7.6% | 7.4% | 9.1% |
| 2004 | 7.6% | 8.1% | 9.4% |
| 2005 | 7.3% | 8.1% | 8.8% |
| 2006 | 7.5% | 8.7% | 8.6% |
| 2007 | 7.6% | 7.7% | 9.2% |
| 2008 | 7.7% | 8.4% | 9.1% |
| 2009 | 8.0% | 7.9% | 8.8% |
| 2010 | 7.6% | 8.1% | 8.7% |
| 2011 | 7.4% | 8.2% | 8.9% |
| Avg | 7.6% | 8.0% | 8.9% |
Walk rate is lowest with men on base, then goes up as the situation gets more threatening. This makes intuitive sense, as pitchers are far more willing to pound the zone and risk the chance of a home run if it will only cost them one run. Once you put them in a situation where a hit could drive in runs, they begin to nibble, and thus, walk rate goes up.
| K% | Empty | Men On | RISP |
|---|---|---|---|
| 2002 | 17.4% | 16.1% | 16.6% |
| 2003 | 17.1% | 15.6% | 16.1% |
| 2004 | 17.4% | 16.3% | 16.5% |
| 2005 | 17.1% | 15.7% | 16.3% |
| 2006 | 17.3% | 16.2% | 16.7% |
| 2007 | 17.6% | 16.4% | 16.9% |
| 2008 | 18.2% | 16.7% | 17.1% |
| 2009 | 18.6% | 17.2% | 17.6% |
| 2010 | 19.1% | 17.7% | 18.3% |
| 2011 | 19.3% | 17.7% | 18.2% |
| Avg | 17.9% | 16.6% | 17.0% |
Strikeout rate is highest with the bases empty, which goes along with pitchers pounding the zone when the cost for doing so is lowest. We’ll see the fruits of that trade-off in the next table.
| HR/9 | Empty | Men On | RISP |
|---|---|---|---|
| 2002 | 1.15 | 0.94 | 0.95 |
| 2003 | 1.19 | 0.96 | 0.97 |
| 2004 | 1.23 | 1.02 | 1.01 |
| 2005 | 1.10 | 0.98 | 0.97 |
| 2006 | 1.21 | 1.02 | 0.96 |
| 2007 | 1.10 | 0.94 | 0.92 |
| 2008 | 1.08 | 0.94 | 0.89 |
| 2009 | 1.16 | 0.93 | 0.91 |
| 2010 | 1.02 | 0.88 | 0.86 |
| 2011 | 1.02 | 0.85 | 0.83 |
| Avg | 1.13 | 0.95 | 0.93 |
And here’s the cost of throwing all those strikes with the bases empty – home run rate is dramatically higher when there’s no one on base, even though pitching from the wind-up is theoretically an advantage for the pitcher. Pitchers are clearly okay giving up an occasional solo home run to keep out of rally situations.
And now, the interesting table, and the one that relates most to Sonnanstine.
| BABIP | Empty | Men On | RISP |
|---|---|---|---|
| 2002 | 0.290 | 0.288 | 0.286 |
| 2003 | 0.289 | 0.292 | 0.287 |
| 2004 | 0.295 | 0.291 | 0.285 |
| 2005 | 0.290 | 0.293 | 0.287 |
| 2006 | 0.298 | 0.298 | 0.293 |
| 2007 | 0.298 | 0.301 | 0.298 |
| 2008 | 0.297 | 0.295 | 0.293 |
| 2009 | 0.297 | 0.294 | 0.290 |
| 2010 | 0.293 | 0.294 | 0.290 |
| 2011 | 0.293 | 0.289 | 0.284 |
| Avg | 0.294 | 0.294 | 0.289 |
As you can see, BABIP is essentially even with the bases empty and with men on base, and then actually goes down with runners in scoring position, which is not what I expected to see. Given that the bases empty situation is almost always is a from-the-wind-up pitch, that the defense gets to align itself optimally with no runner on first to hold, and that selection bias means that a greater proportion of the sample pitched by high quality pitchers, I would have expected BABIP to be lowest with no one on.
Instead, we see the opposite, as it is historically and consistently lower in the highest leverage situations. My guess is that this has to do with more relief pitchers (who run lower than average BABIPs as a group) pitching in RISP situations and potentially a selection bias effect on the hitters due to intentional walks, but those are just untested theories at the moment. Still, regardless of the reason, the data doesn’t lie – there is no league-wide evidence that pitcher BABIP rises significantly when a man reaches base.
Theoretically, a pitcher who posts tremendously different BABIPs in the various baserunner situations is likely due for some regression. That isn’t to say that there is no variance in ability to pitch out of the stretch and that every pitcher is going to perform at league averages in those situations, but it does offer some hope for pitchers like Sonnanstine – his abysmal performances with men on base might not be quite as predictive as just looking at his slash line suggests. You may not look at Sonnastine’s career .306 BABIP and see much room for him to improve, but the timing of when he’s given up those hits may not say too terribly much about when he’s going to give up hits going forward. Simply redistributing some of his hits on balls in play from men on base to bases empty situations could have a pretty significant effect on his results, and make him a potentially decent back-of-the-rotation guy for a team that plays in a big park.
By the way, here are 20 pitchers whose BABIP splits diverged greatly last year. If you’re looking for a list of candidates whose 2012 results might be quite a bit different than their 2011 results, this is a pretty good place to start.
Overachievers:
| Empty | Empty | Men On | RISP |
|---|---|---|---|
| Ian Kennedy | 0.279 | 0.254 | 0.174 |
| Jeremy Hellickson | 0.214 | 0.241 | 0.183 |
| Ricky Romero | 0.264 | 0.211 | 0.191 |
| Max Scherzer | 0.353 | 0.261 | 0.202 |
| Jhoulys Chacin | 0.252 | 0.276 | 0.208 |
| Joe Saunders | 0.280 | 0.258 | 0.211 |
| Johnny Cueto | 0.254 | 0.243 | 0.212 |
| Hiroki Kuroda | 0.307 | 0.252 | 0.218 |
| Tim Stauffer | 0.285 | 0.276 | 0.220 |
Underachievers:
| Empty | Empty | Men On | RISP |
|---|---|---|---|
| Livan Hernandez | 0.297 | 0.335 | 0.364 |
| Madison Bumgarner | 0.307 | 0.343 | 0.363 |
| Brandon Morrow | 0.273 | 0.335 | 0.355 |
| Jake Westbrook | 0.284 | 0.350 | 0.355 |
| Chris Volstad | 0.299 | 0.325 | 0.341 |
| Cliff Lee | 0.312 | 0.255 | 0.340 |
| Zack Greinke | 0.330 | 0.302 | 0.337 |
| David Price | 0.281 | 0.279 | 0.331 |
| Jason Vargas | 0.270 | 0.306 | 0.327 |
I’m pretty surprised to find Westbrook and Lackey here, maybe because I wasn’t paying enough attention. Anyway, this is an incredible work again. Can we rename this article as “the effect of stress on pitchers” ?
Can’t you run the same BABIP with RISP test excluding relievers? That would be much more compelling.
Yeah, there is a lot of junk in here that disrupts the narrative. This would get rid of some of it, but worse pitchers throw more often with men on base and with fewer outs when there are men on base.
There are many other issues:
* sacrifice plays (bunts, SF) only happen with men on base, so the options to the batter expand;
* pitchers should probably be exulted from the batting side because they are so different at the plate than most other players;
* the #4 batter is much more likely to come up with RISP than the #8 batter. This should improve batting (and decrease BABIP due to HRs) with RISP.
Compiling the data would probably be markedly harder
This begs the obvious question: skill or luck? We know what the traditional narrative says, as Tom Glavine can attest.
Not a big deal, but actually it raises the question “skill or luck”, it doesn’t beg the question. Begging or begs the question refers to a logical fallacy.
It doesn’t beg the question. It raises it without answering it (and even acknowledges it in passing). Begging the question means asking a question and then presupposing in the answer what was controversial in the question. You beg the question by giving an answer that only compels a repetition of the same question. To take an example, Euclid asks whether side-angle-side congruency holds and then says, more or less, that if you transpose the triangles on top of one another they are congruent. This begs the question because it gives no independent principle for affirming SAS other than SAS itself. This is why David Hilbert later takes SAS as an axiom of his geometry. Rather than beg the question he just admits that we have no answer for it but we, nevertheless, would like an axiomatic system of Euclidean geometry.
Let the word go forth that begging the question is not the same as raising it without addressing it.
I suppose “ain’t” isn’t really a word either, huh?
I’m chomping at the bit to hear more about this.
I’m fine with “ain’t;” there’s no question about what it means and use can establish new words. Reasons for not using it have nothing to do with whether or not it is a word but rather how you will be perceived by others in various contexts for using it. In general, common use of words and phrases establishes their existence and meaning within a language. This is a different consideration than the one against using “begs the question” as if it meant “raises the question.” This latter consideration concerns preserving important distinctions in meaning among words or phrases within a language. In this case, it is very relevant to distinguish between which criticism is actually being made of the article. Begging the question is a much worse failure in an essay, article, argument, etc., than (implicitly) raising a question without addressing it, which is perhaps not at all a flaw and even inevitable.
You can stop chomping now.
The list of pitchers whose BABIPs were low or high with men on base could be misleading. If those BABIPs do not diverge significantly from their season BABIPs then they might not be due for regression. Of course, the list of pitchers who fared well with men on base have BABIPs so low that even if they are similar to their season BABIP, it is still reasonable to expect a regression. But take Derek Lowe as an example. His season BABIP was .327, only 9 points lower than with men on base. Over his three years in Atlanta he averaged around a .320 BABIP, which gives some evidence that his elevated BABIP is not just due to luck. Further evidence for this is that he averaged a .285 BABIP in four years in LA. The difference between LA and ATL might just be their defense. In which case, perhaps Derek Lowe won’t regress at all. Or he will only appear to regress if Cleveland’s defense is an improvement on ATL’s, but I don’t think it is. I didn’t check the other pitchers on the list, but this possibility seems worth taking into consideration when asking whether these pitchers are due to regress.
I amended the last two tables to show BABIP by all three baserunner states to provide more context. I would suggest, however, that just because a guy had high or low BABIP in all three categories does not mean that his overall number is any less subject to future regression.
I agree and never denied that last bit. My only point was that more analysis needs to be done on each individual pitcher before concluding that he is due for regression in his men on base BABIP.
Have you considered hitter’s approaches with runners on base as a reason for lower BABIP? For instance with a runner on second and no one out, a hitter might hit a routine groundball to second to advance the runner, this would depress BABIP because the hitter sacrificed an out. Same with runners on third in a tag situation, etcetera. It shouldn’t come as a surprise that BABIP with runners on is lower because the number of positive outs (If there is such a thing) increase with the number of runners on base.
More generally, outs on balls in play should be easier to get when there are more runners on base. You have force out situations and fielder’s choice (shorter throw, surer out) to contend with.
Another possible hitter approach factor:fewer strikes thrown in RISP situations and undisciplined hitters making weaker contact.
I haven’t run the numbers recently, but when I studied fastball speed from the stretch vs. in the windup for 2008-2009 data, Sonnanstine’s average fastball was 87.1 mph in the windup and 86.9 mph in the stretch.
http://www.hardballtimes.com/main/blog_article/does-the-stretch-cost-a-pitcher-fastball-speed/
The surprise on this list for me was Bumgarner.
Cliff Lee is an underachiever. Ha, who would have guessed.
Nice article Dave. Thanks.
Kennedy does not surprise me here.
Interesting article. Before I looked at the two lists of pitchers I already knew Brandon Morrow was going to be on the underachievers list. Reason being he has serious mechanical problems out of the stretch. I think it’s dangerous to draw conclusions as to skill/luck from this article because you have some pitchers like Morrow with mechanical problems and others for which the statistics may be an anomaly. Morrow is a really frustrating player to watch because he has some of the best stuff in the game and I have been waiting for him to make the jump into the elite for a couple seasons, however he just seems to lose it once runners get on. Hope he finds a way to get comfortable from the stretch
Another interesting thing about this article is that every pitcher on the underachievers list had a 2011 FIP under their ERA (with the exception of Cliff Lee and it was very close). Every pitcher on the overachievers list had a 2011 FIP over their ERA (with the exception of Max Scherzer). It seems like this analysis may have some relevance for predicting future performance. Good stuff Dave
I agree on Morrow, completely unsurprising to see him there. He just unravels when runners get on. He could be pitching a 1-hit gem into the 6th innings, but as soon as a guy or two gets on, bets are off. Frustrating.
Nice to see the data back up the observations though.
“home run rate is dramatically higher when there’s no one on base, even though pitching from the wind-up is theoretically an advantage for the pitcher.”
In that case, should a power hitter hit leadoff? Probably.
MadBum had the second worst BABIP with RISP, yet he still posted a great year. says volumes for him.
Why wasn’t this dismissed like situational hitting is in this community? Are we saying that only pitchers can be clutch?
Excellent job of being completely off topic. Sonnastine’s faliures this year had almost nothing to do with his performance with runners on vs. no runners on. A quick look at Sonnanstine’s stats he actually had hie LOWEST BABIP (.256 vs career .306) and HIGHEST LOB% (79.5% vs career 65.5%). The reason his FIP was so poor was because he gave up 2.52 HR/9!!!! He had a freakin 22% HR/FB ratio which will regress to his average of 11%. Sonnanstine is a crummy player who just puts too many balls in play to ever be effective. Just wanted to clarify (don’t mean to hate, just I belive the reaserch is unrelated to Sonnanstine’s failure). Thanks
PS (you wrote “Walk rate is lowest with men on base, then goes up as the situation gets more threatening” even though BB rate is lowest when the are no men on, but I assume it is a typo, so I’ll give you a free pass)
Yea I was considering nitpicking on that typo as well, but the article is very well written aside from that. I just had to do a double take and correct it myself.
It would be interesting to see the same table (10 best/10 worst situational babip) for 2010, with 2011 results next to it.