clock menu more-arrow no yes mobile

Filed under:

Shift Happens: Measuring effectiveness with wBABIP

While still imperfect because of data set limitations, we should be using wBABIP to get a better picture of the efficacy of the shift.

David Butler II-USA TODAY Sports

Author's Note: This article originally referred to wBABIP as wOBA throughout, but has been updated as of 11:08 am, on 4/19/16 to more accurately reflect the nature of the statistic used. This very good amendment was suggested by Matt Jackson.

Last week, our very own Matt Jackson took a look at the newly publicly available shift data on FanGraphs to try to determine whether or not the ever-prevalent shift has been successful in suppressing offense. His results were somewhat inconclusive, but in general he found that approximately 33 hits were saved by the shift between 2010 and 2015. The shift was more effective against certain players, like Ryan Howard and David Ortiz, and less so against others.

Matt’s research falls short in his methodology. Matt chooses BABIP as his metric of choice to evaluate the efficacy of the shift, but that only gives us part of the picture. If the shift saves singles but allows more doubles and triples when hitters beat the shift, was it really that effective? If being shift on changes a hitter’s or pitcher’s approach even marginally, strikeout and walk rates are affected as well.

Matt makes the observation in his article that a great deal of noise exists in shift data, as groundballs (and to an extent line drives) are the type of hit most affected by the shift. Since we cannot isolate BABIP on groundballs only for the purposes of finding out the effectiveness of the shift, Matt notes, broad strokes of all batted ball types will have to do. This seems like another reason to move away from BABIP as the sole metric for evaluating shift efficacy.

Additionally, I’m not entirely convinced of the importance of solely looking to groundballs (when it becomes possible to do) for the determination of the effectiveness of shifting. That seems to build in a bias in the data in favor of the shift working. If the alignment of infielders is done in regard to a hitter’s tendency to cluster groundballs to one portion of the field, it would logically follow that on groundballs the batter’s BABIP would be lower. Instead, I would contend that it would be more practically useful to look at the overall effect that shifting has.

After all, if the purpose of shifting is to have a net-negative impact on offense, should the whole of the results not be taken into consideration? Yes, the shift only affects a small number of batted balls, but that’s the point. If the shift is put on for the purpose of suppressing offense, its effectiveness should be measured based off of how well offense was suppressed.

A better metric to measure this exists in wOBA, and so that is what I will use to evaluate the shift. It is far from perfect, since the publicly-available data doesn’t factor in key components to wOBA such as strikeouts and walks, but the primary reason for using this statistic is for the weights on singles, doubles, etc.  Even with the shortcoming of the ultimate number coming from only balls in play data, the weights are not arbitrarily derived as with slugging percentage. The result of what we get is not so much wOBA, but wBABIP - BABIP where singles, doubles and triples are given their equivalent wOBA weights. It is far from perfect, but it is the best that we have to work with, so let’s dig in.

TABLE 1. Non-pitcher wBABIP non-shifting vs. shifting, 2010-2015

No Shift wBABIP Shift wBABIP
2010 .297 .307
2011 .295 .285
2012 .294 .303
2013 .296 .290
2014 .299 .295
2015 .296 .296

Here we see that in broad strokes there is no difference in wBABIP in situations where defense was and was not shifted. This means that from 2010-2015, shifting has not suppressed offense at all. But of course we are actually able to break shift data down further and separate shifts into two bins: traditional (i.e. – Ted Williams), and non-traditional shifts. So let’s have a look at what that yields us, starting first with the more common traditional shifts:

TABLE 2. Non-pitcher wBABIP non-shifting vs. traditional shifting, 2010-2015

No Shift wBABIP Shift (Trad.) wBABIP
2010 .297 .302
2011 .295 .278
2012 .294 .289
2013 .296 .280
2014 .299 .288
2015 .296 .286

Over the five years, we see the traditional shifts have indeed suppressed offense by an average of nine wBABIP points. So in broad strokes, the traditional Ted Williams shift does its job in suppressing offense. Given that the traditional shift is effective, the cumulative effect is in the range of .5-1.5 wins gained by a team, depending on how much they use the shift throughout the season.

TABLE 3. Non-pitcher wBABIP non-shifting vs. non-traditional shifting, 2010-2015

No Shift wBABIP Shift (Non-Trad.) wBABIP
2010 .297 .324
2011 .295 .313
2012 .294 .345
2013 .296 .334
2014 .299 .360
2015 .296 .328

Talk about backfiring. It is a much smaller sample size with only 13,262 plate appearances (against 47,305 traditional shifts), but non-traditional shifts have failed spectacularly on teams, with wBABIP increasing an average of 38 points! But the thing to remember is that we are talking in generalities and the moniker "non-traditional shift" is vague in its description. A non-traditional shift may be more effective against certain players, but as a whole it is not effective.

The man who has been non-traditionally shifted against the most, Billy Hamilton, has a .288 wBABIP against non-traditional shifts and a .284 wBABIP when no shift is employed. It’s certainly not a 38-point disaster, but not effective in suppressing offense as teams would presumably hope.

We see similar offense-boosting effects on the rest of the top-five most non-traditionally shifted players: Dee Gordon (3 points), Billy Burns (103 points), Erick Aybar (5 points), and Elvis Andrus (5 points).  It is worth noting that none of these players have been shifted against enough to have a statistically significant sample size, so there is in fact a great deal of randomness involved.

What we’ve learned is that broadly speaking, the traditional shift is in fact effective in suppressing offense. While there will obviously be variation from player to player, there is on average a nine-point decrease in a hitter’s wBABIP. The overall benefit of shifting, as with many managerial strategies, is small, but real. Non-traditional shifts are somewhat more risky and have largely failed. These should be used with caution, if at all.

It is important to note that with hitters getting more used to being shifted against, all of this will likely change. Some organizations make a point to have their minor league clubs utilize the shift to get the infielders comfortable with shifting. But with shifting on the rise in both the minors and majors, at some point hitters are going to adjust and become more proficient in beating the shift, and when they do, it will be time to revisit the efficacy of the shift.

Joe Vasile is the Broadcasting and Media Relations assistant for the Salem Red Sox. Follow him on Twitter at @JoeVasilePBP.