We’ve Reached Peak Shift
There’s a growing number of windmills sprouting around the country. One can see fields of them from plane windows at 30,000 feet. There’s even a giant one on the east side of Cleveland that’s visible, on a clear day, across Lake Erie from the shores of Bay Village, Ohio, a small municipality on the west side of the city, where this author now resides. The windmills are harnessing the power of the wind to create clean energy, which can be delivered anywhere, I presume, including the car-charging stations I’ve noticed pop up in places like the nearby Whole Foods parking lot. Despite very conspicuous advances and installations, we’re apparently decades away — perhaps not until the 2040s — from reaching Peak Oil, according to Reuters. Clean and renewable energy is taking some time to gain market share.
Some innovations take time to grab hold, others advance quickly. It can be difficult to project when a trend will reach its high-water mark. But we might have seen a prominent 21st trend and strategy reach that point in major-league baseball. In 2016, after a dramatic rise this decade, baseball might have reached Peak Shift.
This year, shifts declined for the first time since their rapid rise earlier in the decade. Ten years ago, shifts were being used against left-handed power hitters exclusively. There wasn’t a shift employed against a right-handed hitter, according to the Baseball Info Solutions database, until June 11, 2009, when the Phillies moved three infielders to the left side of second base against Gary Sheffield. At that point, in the late 2000s, the Rays and Brewers were at the vanguard of wholesale shifting. In 2012, league-wide shift usage doubled, though it was still modest in terms of raw numbers. Gradually more teams bought in — teams like the Pirates, who increased their shift usage by 400% from 2012 to 2013. In 2014, usage doubled again, as the 10,000-shift mark was breached for the first time. The practice has since proliferated across the sport.
From 2012 to -16, shift usage grew by at least 34.8% each season.
But in 2017? This past season, usage dropped by five percentage points, representing the first year-over-year decline in the shift era. Consider the following table:
| Season | Total shifts | Pct. Change |
|---|---|---|
| 2011 | 2,350 | n/a |
| 2012 | 4,577 | 94.8 |
| 2013 | 6,882 | 50.4 |
| 2014 | 13,229 | 92.2 |
| 2015 | 17,826 | 34.8 |
| 2016 | 28,130 | 57.8 |
| 2017 | 26,705 | (-5.0) |
Shifting was still quite popular, it was still a standard practice in 2017, but what we saw was a slight contraction, a hint that some teams are beginning to lessen their dependence on the foreign alignments.
Even some of the top shifting teams, the most devout believers, have begun to dial back shift usage. Consider the following chart ranked by 2017 shifting usage:
| Rank | Team | 2016 | 2017 | % Diff |
|---|---|---|---|---|
| 1 | Brewers | 1,489 | 1,578 | 6.0 |
| 2 | Astros | 1,869 | 1,522 | (-18.6) |
| 3 | Rays | 1,588 | 1,515 | (-4.6) |
| 4 | White Sox | 783 | 1,490 | 90.3 |
| 5 | Mariners | 1,482 | 1,479 | (-1.0) |
| 6 | Pirates | 1,489 | 1,386 | (-6.9) |
So what’s going on here?
What’s happened is that the value of the shifts has possibly decreased, a point explored recently for Baseball Prospectus by the excellent Russell Carleton.
Let’s begin with the ground ball, which the shift is designed to suffocate. The distribution of ground balls hasn’t changed. For most batters, it’s difficult to hit a ground ball to the opposite field.
Consider the distribution of grounders in 2007: 53.2% pulled, 33.6% center, 13.3% opposite.
And in 2017: 53.7% pulled, 33.6% center, 12.7% opposite.
As you might expect, however, the total number of ground balls to the opposite field that have become base hits has increased as shifts have increased. Compared to 2011, MLB hitters have produced about 500 more base hits on ground balls to the opposite field.
| Season | AB | H | AVG1 | wRC+ |
|---|---|---|---|---|
| 2011 | 7036 | 2067 | 0.293 | 67 |
| 2012 | 7536 | 2357 | 0.313 | 79 |
| 2013 | 7326 | 2411 | 0.329 | 90 |
| 2014 | 7694 | 2500 | 0.325 | 91 |
| 2015 | 7728 | 2699 | 0.349 | 102 |
| 2016 | 7487 | 2790 | 0.372 | 111 |
| 2017 | 7046 | 2594 | 0.368 | 106 |
Hitters’ average on pull-side grounders has declined from .202 in 2010 to .190 last season and .186 in 2016.
In employing the FanGraphs splits leaderboards, we can also see performance against ground balls when shifts are on:
| Season | AB | H | AVG | wRC+ |
|---|---|---|---|---|
| 2010 | 175 | 89 | 0.509 | 196 |
| 2011 | 164 | 70 | 0.427 | 150 |
| 2012 | 338 | 155 | 0.459 | 177 |
| 2013 | 452 | 196 | 0.432 | 161 |
| 2014 | 744 | 367 | 0.493 | 199 |
| 2015 | 1297 | 614 | 0.473 | 180 |
| 2016 | 1892 | 924 | 0.488 | 183 |
| 2017 | 1789 | 838 | 0.468 | 169 |
While the volume of hits has increased — perhaps irking pitchers and coaches — the overall rate of success is relatively steady. Batters are not able to guide ground balls, generally, through vacated portions of the infield any better than they were in 2011.
On pull-side grounders, shifts remain effective:
| Season | AB | H | AVG | wRC+ |
|---|---|---|---|---|
| 2010 | 934 | 149 | 0.159 | -12 |
| 2011 | 922 | 138 | 0.149 | -20 |
| 2012 | 1541 | 227 | 0.147 | -19 |
| 2013 | 2112 | 282 | 0.133 | -28 |
| 2014 | 3856 | 524 | 0.135 | -27 |
| 2015 | 5856 | 907 | 0.154 | -17 |
| 2016 | 8168 | 1294 | 0.158 | -18 |
| 2017 | 7535 | 1199 | 0.159 | -18 |
The shift decline appears somewhat contradictory. If the shift seems to be roughly as effective as in the past, why would teams employ them less often?
To better understand that, it’s necessary to return to the beginning, to the man who helped usher in the era of shifts in Tampa Bay: Joe Maddon. It is Maddon, now in Chicago, who is in part responsible for the shift’s decline in 2017.
What was interesting about 2016 is that, in the peak year of shifting to date, the top defensive team was the Cubs. They produced a historic defensive campaign and converted a remarkable percentage of batted balls (74.5%) into outs. It was the best rate since 1950, as Sam Miller noted in an ESPN piece last year, and it was during that season when the Cubs shifted an MLB-low 399 times.
In 2017? The Cubs were again the least-shifting club in the majors, with just 302 of them.
This is remarkable in context. It was Maddon, for example, who had his defenders account for 10% of all shifts in 2010. Now Maddon has his defenders playing straight up — or, at least not in a way that one would classify as a shift. While the Cubs aligned their defenders in more subtle ways, while Maddon said he was mostly concerned with defending the low line drive, Chicago is essentially playing straight up with some premium defenders.
Cubs president Theo Epstein explained the move away from shifts:
“When you shift you risk turning hitters into better hitters than they otherwise would be. You’re opening up holes and encouraging good hitters to use the whole field.”
Perhaps some of the answers reside there, but we have to look up for answers.
The overall distribution of fly balls and line drives hasn’t changed.
2007 line drives and fly balls: 30.6% pulled, 34.4% center, 35.0% opposite
2017 line drives and fly balls: 29.1% pulled, 36.1% center, 34.8% opposite.
But against shifts? Batters are evolving to pull fewer balls that are put in the air. While it’s tough to hit a ground ball to the opposite field, it’s easier to drive one there in the air.
Consider the distribution of line drives and batted balls vs shifts:
In 2010: 31.5% pulled, 37.9% center, 30.6% opposite.
In 2017: 25.8% pulled, 40.8% center, 33.4% opposite.
And in visual form:
Hitters operate a little differently in the air versus shifts, more often getting balls in the air when the defense is realigned. It’s one motivation behind the Air-Ball Revolution. Consider that, in 2011, hitters posted a 1.9 GB/FB ratio against shifts. This past season? A 1.3 mark.
Since 2011, hitters have reduced their ground-ball rate by seven percentage points against shifts and raised their fly-ball rate by five points. One last chart:
Overall, ground-ball percentage has remained steadier.
John Dewan estimated in the 2016 that the upper limit of shifts could reach 40,000 per season. Maybe the sport will not approach that ceiling.
To understand why we might have seen Peak Shift, and why teams are perhaps seeing diminishing returns from the practice, we must not look to the ground, but rather, the sky.
A Cleveland native, FanGraphs writer Travis Sawchik is the author of the New York Times bestselling book, Big Data Baseball. He also contributes to The Athletic Cleveland, and has written for the Pittsburgh Tribune-Review, among other outlets. Follow him on Twitter @Travis_Sawchik.
I’m not sure the argumentation is valid. If a ball is elevated over the shift the shift neither hurts, nor helps, those are hits regardless (if not caught by the outfield).
More elevated balls might reduce the absolute number of outs by the shift but on the balls not elevated the shift is still effective. Bautista was a flyball machine but he was still shifted because IF he hit one on the ground it was pulled (his fbs were pulled too).
The shift never was worried about GB quantity, just with were you hit them if they hit them.
I mean even if you are able to elevate the other way the conservative alignment wouldn’t help the defense either.
And lastly is hitting flyballs to oppo good? It probably is better than grounders but generally fly balls the other way are less often going out
https://www.crawfishboxes.com/2013/9/9/4677304/outcomes-of-flyballs-line-drives-and-groundballs-by-direction-2013
So if a power hitter looks middle away in order to avoid rolling over might hit a couple more flyballs but less bombs and that is a win for the defense too.
That is the tough thing about elevating well, pulled fbs are more effective but usually pulling means more rolled over grounders. Elevating the other way is easier but also less productive. That is also why bautista was so good he was able to selectively pull in the air.
Now of course there are a few hitters with true oppo power but while the new ball migh help with that those are still rare, not a lot of peak ryan howards out there (and even howard as a great oppo flyball hitter was a victim of the shift).
most damage on fly balls is done pull to pull center gap and if hitters try to avoid that I think that is good for defenses.
Opposite field air balls last season: .297 avg/.502 SLG/ 102 wRC+ …. All grounders: .241 avg./.262 SLG/ 29 wRC+.
http://www.fangraphs.com/leaderssplits.aspx?splitArr=12,13,80&strgroup=season&statgroup=2&startDate=2017-03-01&endDate=2017-11-01&filter=&position=B&statType=mlb&autoPt=false&sort=15,1&pg=0
Do you also have pulled flyball wRC+. Sure oppo fly balls are better than grounders but I’m sure pulled FBs are a lot better.
Sure oppo fbs are better than grounders (but way worse than pulled fly balls) but still the baseline for the shift is the standard alignment. Of course the shift becomes less effective with more FBs but so does any other if alignment. If anything lifting oppo fly balls is more effective vs straight up alignment because oppo grounders are more likely to go through the shift than standard alignment.
So if the shift incentivizes players to lift oppo balls they should even moreso do that against standard alignment because oppo grounders do worse against it.
If a defense wanted to adjust to this they would need to move an infielder out of the infield completely.
Sorry for being a PITA today:)
I don’t think it’s good to include line drives in the same category as fly balls for this analysis. Line drives are very consistently good for the batter regardless of whether they’re hit to the pull side (363 wRC+) or the opposite side (305 wRC+) and regardless of whether the defense has shifted or not (347 wRC+ vs 321 wRC+).
As Dominikk hypothesizes, opposite field fly balls are typically not good outcomes for the batter regardless of if there was a shift or not (-22 wRC+ for shift and -33wRC+ for no shift). These statistics indicate that teams shouldn’t shift less simply because some batters will attempt to elevate the ball more when shifted on. I think what is much more likely, is that the potential value of the shift has gone down with the flyball revolution. Not that it doesn’t work, but that it just doesn’t matter as much now.
The reason I lumped line drives and fly balls together is because they are distributed more evenly as air balls, whereas most ground balls are pulled. That’s just the nature of the ball on the ground vs. the ball in the air. LIne drives are generally the good air-ball outcome, fly balls less so. Hitters can use the whole field when launching balls in the air, much more difficult to do with ground ball.
Stay at home mom Kelly Richards from New York after resigning from her full time job managed to average from $6000-$8000 a month from freelancing at home… This is how she done it
…….
???USA~JOB-START
Shift usage decreases after Joe Girardi whines and complains about it for years?
I’ve always wanted to start a conspiracy theory.
Is there any trends towards shifting an infielder deeper into the outfield?
If the answer has been to increase balls over the infielders, it seems to make sense to start covering more of the outfield. Ie: Shallow center-right or left and let the two flanking true outfielders play deeper. You could reduce the value of balls kept in the yard.
I think that is also limited by the speed.slow sluggers are often played very deep on the shift but against a fast guy that can get infield hits that is limited.
That might be one of the reason why speed is correlated with xWOBA over performance (I wrote an article in the community section about that last week so I have the data still in my head).
You first think about infield hits first but if a guy is slow the defense also might prevent more grounders from going through and even lower line drives might suffer as they now don’t all go over the infield but a couple more might be caught.
Agree with this. I know Correa has thrown out Pujols (I think it was him) from shallow left field. Also, Altuve sometimes plays incredibly deep on slow LHBs. Astros typically do vary their depth on speed of the hitter.
It is interesting that pull/oppo% against the shift has only changed in the air and not on the ground, especially since on the ground is likely where it would be most beneficial to the hitters. I would also like to see that further broken down into separate measures of line drives and fly balls. I wonder if the pull/oppo effect is more related to line drives than fly balls, since line drives tend to better represent a “successful” outcome for a hitter (i.e. he did what he was trying to do).
I think the main takeaway is that one reason the shift has lost effectiveness is that hitters are adapting by pulling the ball less and hitting the ball in the air more while the shift is on. But it is also essential to realize that we never should have expected the amount of shifts to keep rising indefinitely.
You call two of your tables “weird” or “contradictory.” The first is is the one entitled “MLB Performance on Opposite-Field Grounders with Shift.” The data show steady performance per batted ball over time. Why is this not exactly what one should expect? The second is the one entitled “MLB Performance on Pull-Side Grounders with Shift.” Again, the rate performance is steady. And again, this is exactly what we should expect! The volume of shifts should depend on the total volume of production allowed, NOT (directly) on the rate of production per batted ball. More shifts means less overall production on pulled hits and more overall production on opposite field hits (assuming primarily pull hitters are shifted). It’s not hard to see that shifting is likely to have diminishing marginal returns, and it seems that we have reached the point where it is no longer beneficial to shift any more frequently (with adaptation by hitters playing an important role as well, as you mentioned).
You’re correct that both charts were rate-related. My apologies. Fixed! And I perhaps should have broken down air balls by LD and FB rates, but I was concerned I’d be over charting in this post.
Elevating is a good adjustment vs the shift but it doesn’t really effect the effectiveness of the shift vs playing straight up.
As long grounders are still pulled the shift makes sense even if there are less grounders.
Unless the shift consciously or subconsciously compels betters to more often go over it
Elevating WILL influence the effectiveness of the shift if for a given batter it is true that every air ball is more desireable than a groundball given the evidence that hitters just can’t hit oppo grounders at free will to counter overshifting. Getting the ball into the air seems a pretty good counter strategy then and seems much more plausible and doable.
Now when facing a weak hitter, why give him incentive to lift the ball more by shifting him?
On the weirs/contradictory, I think he is thinking that hitters would be hitting the ball harder to the opposite field if they were trying to hit it to the opposite field. I would not expect to much of a change in performance per batted ball. What surprises me is that as shifts have increased, hitters aren’t hitting a higher rate of grounders the other way. Shifts are up ~10 fold and batted balls to the opposite field are up ~10 fold over the period in his tables. I would have guessed hitters would get better at going opposite field the more they are shifted against.
Not quibbling with the shift argument, but “peak oil” refers to a theoretical point where maximum extraction is reached. “Peak oil demand” is not a term of art, and the article to which you linked misuses the term in the headline. Normally wouldn’t post a nitpick like this, but this is part of the most pressing issue facing our species and it behooves us to be informed about it, at least as much as the ebbs and flows of where men stand while playing a child’s game.
More to the point: that opening paragraph was unhelpful. Why make the readers wade through the first 10% of a 1,200+ word article on an analogy that adds no value whatsoever? “We’ve reached peak shift” is a pretty clear headline. As a reader, I’m looking for a tightly written explanation of why.
I think that what’s actually going on is that the large ‘traditional shift’ is being replaced by some teams with batter/count/pitch specific, player-specific movement of individual players. The Dodgers do a ton of this, and you also see some of the earmarks of it (like the little positional cards) on other teams, like Jose Altuve had in his hat. There are still the ‘pulls almost all their ground balls’ guys who get the traditional 3 man shift, but a lot of other positioning tweaks are too small to be obvious and tracked, and no one is looking for them. If we had more searchable statcast data, I bet we could find the players with the most variance in their standard positioning and find out whose using this type of defense most often.
The 3 men to one side shift is large, blunt instrument. It’s good against exaggerated tendencies. What we’re almost certainly seeing is that it’s being employed less against people who have more marginal tendencies and being replaced with the batter specific data.
You do see a standard outfield shift these days, which recognizes that for the same reasons players are likely to pull most of their ground balls, they’re likely to hit most of their fly balls to the opposite field. You see a lot of reverse-shifted outfields when shift-type batters are at the plate (for LHB, with the CF shaded strongly towards LF and the LF pretty close to the line).
We don’t have a way to analyze this since it isn’t being strung and we can’t search statcast data, but it’s almost certainly replacing standard shifts for the teams with the most advanced analytics.
I believe those BIS shift totals include traditional (three infielders to one side) and non-traditional shifts … but it is not capturing all defensive movement and alignment as you note as teams are shifting by count, against balls not put in play, etc
Yeah I know BIS tracks non-traditional shifts, but its not going to capture adjustments that fall within the ranges of normal positioning.
I think this is going to become more and more of a problem for the sabermetric community over the next several years. Originally, teams weren’t doing any advanced analysis and it all came externally. You could call this the early BP era.
Slowly teams started to adopt these strategies and then you had teams and third party publishers had very similar information (minus information from the players themselves, which only teams had). I think BP’s pitch framing was the pinnacle of this: BP researched something that some teams may have had in a proprietary manner, independently, and brought the information to everyone. I don’t think MLB liked this phase very much; teams were now spending money to win via analytics, with the risk of their work being duplicated by media parties and becoming leaguewide.
Now we’re in the Statcast era, defined by a lot of really cool information that is proprietary to MLB. Teams have MORE access to this data than we do. The information we get is released primarily as an entertainment product (catch probabilities, hit probabilities). Some data that teams are receiving is not publicized (batted ball spin data, for example), which means that not only are their analytics departments now often larger and better funded than anything we can do, they simply have better information than we do.
I don’t think MLB will ever let this status quo change at this point. Most teams are now invested in analytics: the rich teams are trying to do everything, and the poorer ones are trying to find A Thing that the rich ones didn’t that can give them an advantage. None of those parties are going to want to duplicate the experience of BP dropping catcher framing metrics into the market and evaporating any advantage they’d created by figuring it out themselves.
As teams are doing more and more with Statcast, I think we’re going to get more and more frustrated by the fact that we simply don’t have access to enough data to keep up with what teams are doing. We’ll always be behind the curve. Teams already have 2 years of fielder positional data on every pitch that we have no good way to sort but that we know is being tracked on every play. We’re getting outfield catch probabilities in star ratings and teams are using this stuff to determine the range in each direction of their infielders so they can position guys according to their skills and the dispostions of the batter.
I think the era of media analytics being ahead of, or even on par with, what teams are doing is over and probably over forever.
I personally saw the Reds shift against Mark McGwire in 1998. In fact, I thought it was so interesting that I took a picture, which hangs on my wall at home (amongst some other baseball stuff). I was sitting in the upper deck right behind home plate, so the picture is very clear. Baseball Info Solutions can suck it! 🙂 (They probably don’t have data that far back.)