Answer: Not sure.
I’m trying to figure out this BABIP (batting average on balls in play) puzzle. Orioles first baseman Chris Davis, who notched a BABIP south of .324 just once in his career before last year (.275 in 2010), saw said statistic drop by almost 100 points in 2014. It’s easy to point to the defensive shift as a cause — when defenses shift on you 83 percent of the time, you almost have to — but I’m reluctant to buy in on this just yet.
Unfortunately, there is not much, if any that I know of, publicly-available defensive shift data. Prior to the 2014 season, Jeff Zimmerman published 2013 data courtesy of The 2014 Bill James Almanac. A haphazard calculation yields a 2013 “shift BABIP” about 18 points, or about 6 percent, lower than the MLB aggregate BABIP of .297. During the 2014 season, an ESPN feature projected defensive shifts to reach an all-time high, and by quite a margin, too. In light of this, one could hypothesize that more overall shifts would cause a lower aggregate BABIP. However, MLB’s aggregate BABIP in 2014 was .298.
None of this really tells us a whole lot. The shift BABIP would be awesome if it could be broken down by location of the ball in play — accordingly, I would strictly focus on a pull-side shift BABIP — but, alas, it does not. FanGraphs also breaks down a hitter’s spray chart numerically — you can view Davis’ pull-side splits here — but it does not indicate how many times defenses shifted against him when he pulled the ball. Until this gap in the data can be both a) plugged and b) made publicly available, the answers we seek regarding the true effectiveness of the shift may evade us.
No matter, because I still want to try to figure some things out. Let’s talk a little bit of theory. Like a hitter’s BABIP, I think his shift BABIP is also likely to be volatile. No matter where you place your fielders, you cannot predict where a batter will hit the ball. If you study the spray charts and play the probabilities just right, you’ll surely turn a few more would-be hits into outs. But just like regular BABIP, there will still be an element of luck involved.
Thus, when I look at this table, reproduced from Mike Podhorzer’s FanGraphs post…
|Season||At-Bats||Balls in Play||Shift Count||% Shifted||Shift BABIP||No Shift BABIP|
… I see all sorts of luck. I think the mistake is made when one relates shift percentage with shift BABIP. I expect more shifts to correlate with greater effectiveness — results that would be reflected in the hitter’s depressed batting average. But more shifts does not equate to greater effectiveness on a per-play basis, which is essentially what shift BABIP measures. In short: given that a player’s batted ball profile is identical year to year, his shift BABIP should have some semblance of consistency. We know that BABIP is pretty volatile, but there is a small element of consistency to it (for example, Edwin Encarnacion‘s BABIP is perennially stuck in the mid-.200s while Mike Trout‘s is typically buoyed in the upper-.300s). Thus, I would expect shift BABIP to exhibit at least a little bit of consistency, and for that consistency to produce consistently lower marks than that of the regular BABIP.
Speaking of batted ball profiles, Davis’ pull-side profile was consistent between 2013 and 2014:
Yet Davis’ pull-side BABIP dropped from .338 to .185. The decrease makes sense intuitively, but he saw the fewest shifts in 2012 and actually had a worse pull-side BABIP than he did in 2013. I don’t have to run a regression to show there’s no correlation to be found there (albeit in a minuscule sample size). Now, his increasing tendency to pull the ball (43.3% in 2012, 46.2% in 2013, 50.9% in 2014): that is something that should correlate well with shift BABIP. Because the shift BABIP doesn’t differentiate among ball placement, where the player hits the ball ought to affect his shift BABIP, especially if he predominantly pulls the ball. Thus, an increase in balls in play to the pull side should correlate with a decrease in shift BABIP. Despite all this, Davis recorded his highest shift BABIP during the year he pulled the ball with the least amount of authority.
Now, forgive me, but I have to try to make something of all of this. Let’s take the 6-percent decrease in aggregate BABIP when accounting for shifts (from earlier), and let’s say that teams shift on Davis 100 percent of the time. (It’s not unfathomable, given defenses shifted against him five times out of six, and it appears — it appears — to have succeeded with flying colors.) Given an identical batted ball profile from year to year, maybe I could expect his BABIP, which sat at .335 and .336 the two years prior to 2014, to fall to around .315 permanently. Even if his “true” BABIP benchmark is closer to .300, then maybe his overall shift BABIP is in the .280 ballpark. As he hits more and more balls to his pull side, his shift BABIP will decrease, as will his batting average. That I can fathom.
But I cannot bring myself to accept that a 10-percent increase in pull-side balls-in-play from 2013 to 2014 correlates with a 24-percent decrease in shift BABIP. I don’t think the latter can reasonably be larger than the former without a significant luck element involved. Then again, the 7-percent increase in pull-side balls from 2012 to 2013 resulted in a 17-percent decrease in shift BABIP produces an almost identical ratio (24/10 = 2.4, 17/7 = 2.429), so maybe there’s something I’m missing. But allow me to speak hypothetically: Let’s say Davis puts 100 balls in play, consisting of 50 to his pull side and 50 everywhere else. This silly 2.4-to-1 ratio demonstrates that one more ball hit to the pull side — that is, now he hits 51 balls to the pull side and 49 everywhere else — means not only is that one extra pulled ball an automatic out but also almost one-and-a-half more balls not to the pull side become outs. It’s simply incomprehensible, and I maintain that a percentage increase in balls hit to the pull side would correlate with at most a percentage decrease in shift BABIP.
Wrapping things up: I think it goes without saying that Davis got unlucky in the BABIP department in 2014 — it’s more a matter of determining how unlucky and why. I think his shift BABIPs betray Davis; I think he got especially lucky against the shift in 2012 and especially unlucky in 2014. In general, more shifts should suppress a hitter’s batting average but not his shift BABIP, and it’s Davis’ shift and pull-side BABIPs that absolutely tanked in 2014. Considering he still managed to hit a home run in 5 percent of his plate appearances, I a full 600 from Davis to yield at least 30 bombs, and I think that’s a modest projection. Couple that with a batting average rebound — which I fully expect at this point, strikeout rate disclaimers withstanding — and the down-and-out Davis could be a nice draft day bargain.
Apologies for the lull between posts. I’ve been entertaining friends and family in my adopted city of Portland, Ore. for the past week — while mourning the fact I will likely miss the playoffs in my head-to-head league because of a tiebreaker. More like a heartbreaker.
I’ll get back to more quantitative analysis in the coming days. For now, here’s more quick stuff.
Brandon Beachy, ATL
For playoff contenders, abandon ship (unless you’ve got space on the DL or you’re in a dynasty league). The guy has been filthy throughout his professional career, so some offseason rest will likely do him some good. Potential top-30 pitcher next year, and that’s being modest.
Marco Estrada, MIL
Is this the same Marco Estrada who humiliated me earlier this year? Part of me wonders if he only likes to turn it on after the All-Star Break. Take a look at his post-ASB numbers the past two years:
2012: 3.40 ERA, 1.21 WHIP, 88 K (9.1 K/9), only 7 HR allowed in 15 starts
2013: 1.88 ERA, 0.71 WHIP, 21 K (7.9 K/9), only 3 BB allowed in 4 starts
I’m being a bit facetious, because Estrada was quietly good for the entirety of 2012, but he was plagued by the long ball and poor control in the first half of this year. Aside from the flashy ratios, the three walks across 24 innings is particularly pleasing, reassuring, what-have-you. As Papa Roach once eloquently sang, “The scars remind us that the past is real” — and the scars Estrada gave me this year (further deepened every time I remember I watched Hisashi Iwakuma sit in free agency for three starts before getting signed) make it hard for me to trust him immediately. But, again, if I’m a contender, I’m on board. If his amazing post-Break K/BB ratio continues into 2014, I’m buying again.
Dan Haren, WAS
I’m sold on the bounceback… but I’m not, ya know? Haren has been very hittable this year, serving up a ton of home runs, and that trend has continued through the All-Star Break. However, since the Break, he has posted a 2.74 ERA and 0.87 WHIP in seven starts — certainly hard to ignore. But his BAbip is also .225, more than 100 points lower than his first-half mark, meaning his sudden turnaround is kind of a fluke.
Ultimately, fluctuations in HR/FB rates are largely a product of good or bad luck, and Haren’s 2013 rate is the highest of his career, as was his BAbip heading into the Break. His K/BB rate is one of the highest of his career, comparable to his All-Star/Cy Young contender days, and his strikeout rate is the best it has been since 2010. If the Washington Nationals can put 2013 in the past next year, I could see Haren bouncing back quite nicely if he can maintain his progress.
Carlos Martinez, STL
The scouts love him, but he was sent down again by the Cardinals. He may not help much this year, so don’t count on it. I’m wary of his walk rate becoming something unmanageable at the major league level, but his ability to induce outs as well as his high strikeout rate should help suppress any issues his walk rate may cause.
Danny Salazar, CLE
Salazar has become a rather underwhelming option after taking the league by storm in his first handful of starts. As Chris Towers of CBSSports.com noted, the Indians have been very strict with Salazar’s innings. Unless he is incredibly efficient, he won’t eat enough innings to be truly effective — he won’t strike out as many guys, and he may not even reach the five-inning mark needed to qualify for a win more frequently than not, just like has he has done twice in his last three starts. He’s a fashionable option now, but his leash is very short.
Oh yeah, and…
Matt Harvey, NYM
Yikes. Rarely have I muttered an expletive out loud while reading a text message — and I don’t even own him. This has surely freaked out a lot of owners, and I don’t have much solace to offer. He’ll be back next year? The Mets may actually be a force to be reckoned with in 2014?
Let’s look at the big picture, though. If you’re in a standard rotisserie league, you have about 320 innings (of 1,600) left to throw. You’re a contender with a 3.502 ERA and 1.180 WHIP with 1,200 strikeouts. So let’s say Harvey would have thrown another seven or eight starts — say, 48 innings — before season’s end. Here’s how Harvey would affect your numbers:
Before Harvey’s injury (1280 IP): 3.502 ERA, 1.180 WHIP, 1200 K (8.44 K/9)
If Harvey was healthy (1328 IP): 3.456 ERA, 1.171 WHIP, 1251 K (8.48 K/9)
See, we’re so deep into the season that Harvey’s rest-of-season projected impact (based on his current stats) is greatly diluted — only for certain teams in certain leagues will an improvement of half a run earn you multiple points in the standings. And given how few starts pitchers have left, someone who lost Harvey may even have something to gain by playing the hot hand of someone with a 0.708 WHIP over his last seven starts (Haren) or a 7.88 K/9 since coming off the DL (Estrada).
In head-to-head leagues, the story is a little different, but no so much. It is less about the big picture, like in rotisserie, as it is about the current week. It relies much more heavily on small sample sizes, and that’s what the end of the regular season truly is. Quit crying and ride a hot hand. You’ll be OK, trust me!