Let’s Talk About Shifts
Yesterday, I unveiled the newest incarnation of the xBABIP equation, this time tacking on a shift-related component. The defensive shift has been all the rage these past couple of years as teams are utilizing data more and more for any incremental advantage they could find. Finally with the Splits Leaderboard, we have all the data at our disposal to dive into. Let’s jump in, shall we?
We’ll begin with a table of BABIP marks by season for pulled grounders with no shift and a traditional shift:
| Season | Pulled GB No Shift | Pulled GB Traditional Shift | Diff |
|---|---|---|---|
| 2012 | 0.198 | 0.110 | 0.088 |
| 2013 | 0.197 | 0.099 | 0.098 |
| 2014 | 0.196 | 0.117 | 0.079 |
| 2015 | 0.195 | 0.124 | 0.071 |
| 2016 | 0.197 | 0.135 | 0.062 |
Wellllll, that’s interesting. The BABIP on pulled grounders without a shift has remained remarkably consistent, stuck in a range between .195 and .198. That hasn’t been the case for pulled grounders hit into a traditional shift, though. BABIPs for that batted ball type situation have risen nearly every single season, suggesting continually decreased effectiveness of the defensive positioning strategy. Perhaps batters have somehow figured out how to “hit ’em where they ain’t”, hitting seeing-eye grounders through the shifted defenders? Or maybe Bill Buckner is manning all shifts (sorry Bill)?
Now let’s check in on the frequency of the traditional shift:
| Season | No Shift BIP | Trad Shift BIP | % BIP Shifted |
|---|---|---|---|
| 2012 | 119033 | 4572 | 3.7% |
| 2013 | 117405 | 6878 | 5.5% |
| 2014 | 110339 | 13290 | 10.7% |
| 2015 | 100931 | 17723 | 14.9% |
| 2016 | 88296 | 28065 | 24.1% |
My xBABIP spreadsheet contains historical xBABIP marks for all hitters going back to 2012 and it helps guide my Pod Projection (available now!) BABIP forecasts. As I had been perusing it each day when projecting players, it seemed as if 2014 was the beginning of the shift era, and the above data backs it up. That season represented the first real jump, and defensive shift usage has continued to increase dramatically since.
But this brings me back to the first table — how is it that teams are shifting more than ever, yet it’s actually becoming less effective? The knee-jerk answer is that hitters are learning to go the other way, but remember, the BABIP in the first table is only from pulled grounders. So even if hitters have adjusted, that would simply reduce the number of pulled grounders in the BABIP calculation.
While this won’t provide any sort of explanation, let’s look at another calculation required for my xBABIP equation — Pull% on GB While Shifted. How has the league trended when being shifted against, and, when shifted against more and more? Are they pulling their grounders less frequently and going the opposite way or up the middle instead? Let’s find out:
| Season | Pull% | Cent% | Oppo% |
|---|---|---|---|
| 2012 | 56.6% | 31.7% | 11.8% |
| 2013 | 57.2% | 30.7% | 12.1% |
| 2014 | 58.8% | 29.8% | 11.5% |
| 2015 | 56.4% | 32.2% | 11.4% |
| 2016 | 55.1% | 32.7% | 12.2% |
Ehhhh, yes? But barely. 2016 did see the lowest rate of pulled grounders and the highest rates of Center% and Oppo% when facing a shift, and after peaking in 2014, the Pull% has trickled down slightly each year, inverse to the rate at which teams employed the shift. So that’s the trend we would expect to see, but probably not the magnitude. You had to figure batters would have adjusted to a much greater degree than they apparently have.
Lastly, we’re going to finish the full calculation and check out the trend for Pull GB While Shifted%, which is the official newest component of the xBABIP equation:
| Season | GB% | % BIP Shifted | Pull% on GB While Shifted | Pull GB While Shifted% |
|---|---|---|---|---|
| 2012 | 45.1% | 3.7% | 56.6% | 1.0% |
| 2013 | 44.5% | 5.5% | 57.2% | 1.5% |
| 2014 | 44.8% | 10.7% | 58.8% | 3.2% |
| 2015 | 45.3% | 14.9% | 56.4% | 4.5% |
| 2016 | 44.7% | 24.1% | 55.1% | 7.8% |
As you may have expected, Pull GB While Shifted% has risen as % BIP Shifted has surged. Batters are hitting the same rate of overall ground balls, and a very similar rate of pulled grounders against the shift. The only difference is the rate at which teams are employing the shift, which has spiked more than eight-fold since 2012.
What’s interesting is that even though we know teams are shifting more than ever before, and pulled grounders hit into the shift is a BABIP and batting average killer, albeit less so with each passing season, overall league BABIP has actually increased! Without even looking, I bet I could guess why. But here’s the table of all the xBABIP components from each year:
| Season | BABIP | LD% | True FB% | True IFFB% | Spd | Hard% | Pull GB While Shifted% |
|---|---|---|---|---|---|---|---|
| 2012 | 0.297 | 20.9% | 30.6% | 3.4% | 4.6 | 28.5% | 1.0% |
| 2013 | 0.297 | 21.2% | 31.0% | 3.3% | 4.4 | 29.9% | 1.5% |
| 2014 | 0.299 | 20.8% | 31.1% | 3.3% | 4.4 | 29.1% | 3.2% |
| 2015 | 0.299 | 20.9% | 30.6% | 3.2% | 4.3 | 28.8% | 4.5% |
| 2016 | 0.300 | 20.7% | 31.2% | 3.4% | 4.4 | 31.4% | 7.8% |
The league isn’t hitting more line drives, in fact, it is hitting fewer. The league isn’t hitting fewer fly balls, in fact, it is hitting more. The league isn’t hitting fewer pop-ups, in fact, it matched its 2012 peak. And the league isn’t getting any faster.
But what the league is doing is hitting it harder, as it remained in a relatively stable range from 2012 to 2015 before exploding in 2016, pushing that Hard% above 30%. It ranked as the second highest Hard% since we have data going back to 2002 (2007 was highest at 32.0%, and guess what, that was the highest BABIP recorded since 2002 as well at .303).
And this gradual rise in BABIP is all coming when that Pull GB While Shifted% just keeps going up, and up, and up. Perhaps a defensive shift beyond the field’s fences is the next big idea?
Mike Podhorzer is the 2015 Fantasy Sports Writers Association Baseball Writer of the Year and three-time Tout Wars champion. He is the author of the eBook Projecting X 2.0: How to Forecast Baseball Player Performance, which teaches you how to project players yourself. Follow Mike on X@MikePodhorzer and contact him via email.
This all looks like survivor bias. Shifts exploded because many batters could be exploited by them. The players who were disproportionately affected adapted or filtered out of the league.
Also if more shifts are being employed than ever presumably the average expected value of each shift goes down as the bar is lowered for what batters are shifted in which situations.
This was my first thought, too. When the league is shifting 17.6% of the time, that might be encompass 95% of hitters who are most adversely affected by the shift (MAABTS). So when the league shifts 31.8% of the time, those 95% MAABTS are already included, and the players added to the “worth shifting, but not MAABTS” are only marginally worse vs the shift than they are vs traditional alignment. So you gain a little bit of an advantage (maybe), even though the overall effectiveness goes down.
In other words, 95% of the low-hanging fruit were picked in 2015 whiel any fruit you had to jump for was left in the tree. And in 2016, 95% of the low-hanging fruit are still picked, but we’re also trying to pick the fruit that we have to jump a bit for, which takes longer and is less “fruitful” (pun intended). The marginal “fruit picked per hour” has gone down, even if the fruits themselves are not distributed differently in 2016 vs 2015. In 2015 we paid 4.8 cents per apple, and now in 2016 we’re paying 5 cents per apple to the pickers because it takes longer for the ones they have to jump for. Our marginal cost of labour paid per apple for those we have to jump for might be 5.2 cents, but if we can sell apples for 6 cents apiece then it still makes sense to pick those apples. So, efficiency appears to be going down overall, but profitability goes up. Thus ends my way-too-long apple picking analogy.
This was a fantastic analogy and makes perfect sense to me. Thanks for this!