Is the LeMahieu Shift the Boldest One Ever?
In the age of the shift, it takes a lot for a particular defensive alignment to merit real attention.
But over the weekend, the Diamondbacks managed to do just that, utilizing what appears to be the most dramatic shift in recent history — notable even more so because the park at which they did it, Coors Field, features one of the game’s largest outfields.
We’ve seen about every variety of infield shift over the last four years, but we’ve never witnessed anything quite like what the Diamondbacks employed against DJ LeMahieu.
That gets your attention. That is bold.
Diamondbacks manager Torey Lovullo was aware of the implications. He relayed his thoughts to Nick Piecoro of the Arizona Republic.
“It’s a super-shift times 10,” Lovullo said. “You’ve never really seen anything like it.”
“With nobody in left field, it could be an inside-the-park home run… You’re kind of holding your breath.”
And in the fifth inning Friday night, on its third attempt, the shift worked.
That’s a line drive to “center fielder” A.J. Pollock, catching the ball after just a few quick steps to his left.
A.J. Pollock's catch on DJ LeMahieu's lineout in 5th inning last night. Almost certainly wouldn't have made w/o shift. pic.twitter.com/BoxpAeUUqC
— Daniel Kramer (@DKramer_) September 2, 2017
The idea for the extreme shift was pitched to Luvollo by Diamondbacks coach Dave McKay, who’d studied the data provided by Arizona’s front office. Mike Fitzgerald, who leads the club’s analytics department, had helped bring the infield shift to Pittsburgh in 2013.
In this case, the data clearly shows that LeMahieu has an extreme batted-ball profile. While most hitters pull the majority of ground balls, fly balls tend to be more evenly distributed. LeMahieu is an outlier, though. Consider his fly-ball distribution compared to the league since 2015:
| Location | MLB Flies, 2017 | LeMahieu Flies, 2017 | LeMahieu Flies, Career |
|---|---|---|---|
| Pull | 24.2% | 12.4% | 11.1% |
| Center | 36.3% | 28.4% | 38.0% |
| Opposite | 39.4% | 59.3% | 50.9% |
For those who prefer visual presentations, here’s a heat map of all LeMahieu’s fly balls against right-handed pitchers since 2015:
Said LeMahieu to reporters after the game:
“I mean, I don’t blame them. I hit a lot of balls that way, but at the same time, if I hit a ground ball, it’s a triple to the left side. I’m not really going to change my approach. I hit plenty of balls to the left, enough balls to the left that if they keep playing me like that, I’ll burn them a couple times, at least.”
Said McKay to Piecoro:
“Tony told me one time, ‘No guts, no glory … If you really feel like it’s the right thing to do, go for it. No guts, no glory. He’ll probably beat us on it tomorrow. He’s that good of a hitter.”
Diamondbacks starting pitcher Taijuan Walker was surprised to see what was going on behind him. “I was so confused… I’m definitely glad I didn’t know,” Walker told reporters. (Maybe the fewer parties that know of such a plan, the better.)
After Friday’s game, Lovullo told Piecoro he planned to tweak the alignment. Rather than have left fielder David Peralta and center fielder A.J. Pollock shift to the right, they would swap the pair, keeping Pollock furthest to the left, to keep the team’s fleetest defender in center in the event that damage damage control was necessary in left field.
“(We’ll have) two natural right fielders and (in center) your best fielder by speed and range is going to cover the most ground,” Lovullo said. “We think about it as much we can, and we probably didn’t think about it probably as thoroughly as we (did) after the game last night.”
Lovullo also said the Diamondbacks would be unlikely to feature the same alignment with a left-handed pitcher on the mound, though LeMahieu goes the other way often against lefties, too:
“Against left-handed pitching, the habits and trends aren’t showing the same things they were last night,” Lovullo said. “It’ll be a case-by-case situation.”
You could argue that teams ought to consider doing this against other hitters, too.
While only Joe Mauer (70.5%) and Domingo Santana (62.2%), for example, both hit more fly balls to the opposite field than LeMahieu. Of the league’s qualified hitters, 27 have hit at least 50% of their fly balls to the opposite field this season.
In fact, 137 of 313 qualified hitters hit 40% or more of their fly balls to the opposite field.
So there’s almost certainly some benefit to experimenting with a more pronounced shift. Of course, the strategy carries significant risk, too.
LeMahieu is an extreme ground-ball hitter, producing 2.6 grounders for every fly ball over the course of his career. The ratio is 3-to-1 this season. And LeMahieu has more evenly distributed his ground balls (34.4% pull vs. 43.6% center vs. 22.0% opposite) over the course of his career.
Here are all of LeMahieu’s batted balls this season:
A hard ground ball or line drive to left would provide significant damage against such an outfield shift.
But when a right-handed starter returned to the mound Sunday, the Diamondbacks returned to the LeMahieu shift. LeMahieu pounded his first three batted balls into the turf on the left side of the infield, perhaps a product of attempting to burn the alignment.
He did fly out to right to end the game Sunday.
Even if this was nothing more than a curiosity, it was at least a lot of fun for one weekend. But teams have become more interested in outfield alignments.
I wrote earlier this year about how the Astros have been at the forefront of outfield shifting. In such cases, however, the possible negative cost of an outfield shift gone awry (an extra-base hit) is greater than the same thing for an infield shift (usually just a single).
When Fitzgerald was with the Pirates last season, the club moved their outfielders to dramatically shallower positions. The logic? That the team’s ground-ball staff produced more grounders to the outfield and that a shallower outfield could better prevent extra bases taken on the base paths. The plan backfired, however, when the staff didn’t execute as well off the mound. The outfielders were often running with their backs turned toward the infield. All shifts start with execution off the pitching mound.
Earlier this year against Joey Votto, the Cubs employed a four-man outfield. You could argue playing a five-man infield against LeMahieu might make more sense than a dramatically shifted outfield.
Teams are becoming more interested in outfield alignment, but it’s also a more challenging enterprise, as there’s more space to cover and a greater number of batted-ball types for which to account. It’s like going from chess to three-dimensional chess.
Maybe the Diamondbacks are onto something, or maybe they were lucky LeMahieu never found left field. Either way, it got our attention.
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.
So what’s that foul out that looks about 20 ft out of play behind home?
Probably in a game played at Oakland?
Oakland’s not on the Rockies schedule for 2017.
I’m not too familiar with all baseball parks, but it has to be some park with lots of foul ground right? That’s why I guessed Oakland.
From Baseball Savant Spray Chart:
DJ LeMahieu grounds out softly, third baseman Justin Turner to first baseman Adrian Gonzalez.
Game Date: 2017-04-08
Pitch: Slider
Batted Ball Velocity: 68.7 mph
Distance: 4 feet
Uhh, no.
Where the ball first bounced, maybe? The batter swung away in a bunt situation, then just wound up topping the ball anyway?
Missed it by *this* much, StatCast.
Computers don’t make mistakes, humans do.
Holy crap, that ball would have had to decelerate at a rate of 1270 ft/s^2 to stop at 4 ft. That’s nearly 40 g’s!
Nice piece, but a RFer in his “normal” position probably makes the catch that Pollock makes, no? Any chance we can get the hit % on that batted ball?
According to state cast, it has a hit% of 69%. Nice. Though this metric does not include “radial” data. So it treats a ball hit to dead center the same as one down the right field line despite the fact the second one is always a hit, and the first is sometimes a hit. That is probably the metric we would need to judge the shift here.
That’s actually wrong, hit probability does pay attention to the spray vector, its just not data we have access to. It is recorded, though, and teams have access.
You will see radically different hit probabilities on the same 375 foot fly balls hit with the same EV/LA profiles.
Statcast collects and uses a lot of data that isn’t publically released.
Hit probability doesn’t take fielder starting point into consideration, just EV/LA/vector on the field.
The Statcast catch probability stat, however, looks at the actual fielder position and the amount of distance/time required to make it to the ball’s endpoint. So this ball would have a much different statcast Catch Probability than it does Hit Probability, since it was not a hard ball to field (but it is usually a hit due to positioning).
Just because the only data publicized by statcast is EV and LA doesn’t mean its the only data it’s collecting and using to give you hit probabilities. It’s also recording batted ball spin data and including that (which is why statcast will give balls that look otherwise the same in EV/LA different projected distances on home runs, due to spin).
I thought the exact same thing at first, but after watching the video clip I’m not so certain. The ball had no airtime, and I think J.D. Martinez in the normal RF position would have had to move more than Pollock had to in order to get it. And Pollock – who’s faster than Martinez – didn’t even have enough time to set-up under it, he caught it mid-run.
Are you sure a normally positioned right fielder wouldn’t be closer? It looks like it was hit to straight away right field to me.
TKDC still has a good point, though.
This also assumes that a “normal” positioning would actually be the alternative to this. Does any team play a guy with LeMahieu’s profile just straight away? I bet if you looked at how teams have actually defended LeMahieu this year, this is easily an extra base hit almost every single time.
Here is a pretty extreme shift that the Padres did to Matt Duffy a couple years ago. http://mediadownloads.mlb.com/mlbam/mp4/2016/04/27/632916583/1461790888945/asset_1200K.mp4
The result of that shift is a triple instead of a double. So in this case, the damage was relatively minor since he would have been in scoring position regardless.
Look at the % of the balls he hit to the opposite field that went for singles – aren’t they trading a moderate risk for a relatively low impact event (a single) for a low risk of a relatively high impact event (inside the park home run)?
I don’t have the numbers for this, but this just intuitively seems incorrect, particularly with no runners on base (which would shift the value of getting that single).
I think if you have your fast centerfielder as the guy playing to the left of center, you’re not really risking too many ITPHRs. I don’t see any dots that look like they’d be too far away to hold him to a triple, and I’d think that most of the balls hit over there would still only be doubles.
But the way I’d think about it is this: LeMahieu is already being shifted pretty heavily (I’m just assuming this is true). And he’s still doing way more damage to right field than left field. So why not shift him more? Until he’s getting more hits (or better hits) to left than right, it seems logical to continue to stretch the defense to defend over there, where all the damage is happening.
Looking at wOBA constants, a single is valued at .876. So you can look at how many singles you turn into outs vs how many outs turn into triples as an example. A triple is valued at 1.550. Seems risky, but maybe not as much as you would expect?
Any idea if the dimensions and XBH rates for Coors play into this alignment? It almost seems like a no doubles defense for DJ, but with less space to cover at home, you’d really be trading singles for the risk of an IPHR, vs. at Coors where those are more likely to be doubles and triples.
Not so sure an inside the park HR is really at risk here. LeMahieu has to run 360 feet in a circle which effectively means he has to run nearly 400 feet with rounding bases. Since the “left” fielder is positioned just left of center, the absolute max distance he has to run to retrieve the ball is about 225 feet or so, even for one right down the line and in the corner.
Put another way, the very fastest players circle the bases just under 14 seconds. Per Jeff Zimmmerman’s home-to-first piece from earlier this year we know that LeMahieu is slightly slower than average so his time circling the bases is probably closer to 17 seconds. I have to think a normal OF can run 225 feet and get the ball in to home with a relay throw faster than that. And remember, that’s only a ball right down the line. That’s also only at Coors Field which has some of the deepest, if not the deepest, fences down the line.
Inside the park HRs happen when defenses are in regular alignments. All it really takes i a bizarre bounce that no one can expect, that causes the ball to bounce about 100 feet away for it to happen.
Assuming everything you have put, the batter would be running at 21 feet per second. Assuming the CF is faster, we can place him at 27 feet per second making him get to the ball in 8 1/3 seconds if everything is perfectly played, and there is no spin on the ball that makes it behave in anything but a regular straight line. This puts the batter on second base at the exact moment the fielder gets to the ball. But that fails to neglect that the fielder cannot just sprint to the ball and pick it up top speed and throw it in. They have to slow down, pick the ball up, and set up in position to actually throw the ball with strength and accuracy. By the time this happens, the runner would be halfway between 2nd and 3rd, or basically where the third base coach would normally waive almost any runner around for an attempt for a run.
It is more than reasonable to think that almost any hit in the general direction of the left field line would result in an Inside the park HR, and at very minimum, the attempt.
Crazy bounces only have an impact when the fielder is standing right at the wall and that wouldn’t be the case here.
As to total time – 8.333 seems a bit optimistic (Usain Bolt at absolute top speed is about 38 ft/sec for reference), let’s say 9 seconds for a typical CF. A 90 mph throw travels 132 ft/sec, however it’s on an arc so let’s say 125 ft per second or roughly 3 seconds to get from the left field foul pole at 347 ft to home. That’s 12 seconds. Time to gather and throw the ball and make the transfer on the relay throw can’t be 5 seconds. That would be more like 2 seconds. So now we’re at 14 seconds total vs. 17 seconds for LaMehieu to round the bases. Even if you drop LeMahieu’s time to 16.5 seconds, he’s still out by a bunch. Certainly the fielder could drop the ball picking it up and the relay throw could be offline but honestly, that sounds like out by 30+ feet most of the time. And remember, that’s only on a ball that is hit exactly down the line and stops at the wall, basically a best (or worst for the defense) case scenario that would happen an incredibly small percentage of the time.
8.3333 seconds assumes the CF is sprinting top speed from the crack of the bat to the exact location of the ball. Obviously this is not possible, and 9 seconds is still optimistic. In order to actually field and throw, the fielder needs to slow down, come to a complete stop, and then throw the ball. If the fielder continues sprinting past the point of the ball it makes fielding the ball much more difficult, and nearly impossible to get anything on the throw into the infield. If they continue to go top speed it would have to be a highlight reel shortstop leaping throw style play, that has to cover nearly double the ground with a much more possibility for error. In reality, this entire process of running to the ball, getting the ball off the round, setting to make the throw (in which the fielder’s momentum will be not going in the direction of the play) and then throwing the ball, would probably take about 10.5-11 seconds for almost everyone. Given your assigned speed to LaMahieu that would put him at 220-231 feet out of the 360 he has to run. Any third base coach would send a runner in that position with a fielder that has to make a throw from the wall.
Quite simply the math here is, if you would normally send the runner from first on a double, you would send them on this theoretical play. It would be the same positioning of the runner and fielder at this point.
It seems you’ve ‘proven’ that inside the park home runs are statistically impossible.
All inside the park HR are either crazy bounces off the wall that go back past the fielder or diving attempts to catch the ball where the ball rolls past the fielder (now laying on the ground) to the wall.
Neither would come into play in this scenario.
The Dbacks have come a long way since last year, both in W-L record, and the fact that they actually have an analytics department now that can be open to the possibility of doing something like this. What were the odds of them trying something like this last year? About ZERO.
This is all anecdotal, and I’m sure you can forgive me for not paying attention much last year, but I seem to remember Chip Hale actually employing quite a lot of defensive shifts.
Mark me down as one of those readers that sees a pretty routine ball to the right-fielder – it certainly had plenty of air under it.
I think it is important to distinguish an over-shift from a shift. Shifts have been around forever. A good high school program will employ shifts… An over-shift is the goofy looking thing that we are discussing today. I could be wrong, but I think the peak over-shift era is over with. Data, results, and common sense have emerged and we are seeing less over-shifts, which is ironic as it was “data” in the first place that gave birth to over-shifts. Over-shifts create a hit for every one that they take away, which is why BABIP hasn’t gone down in the shift era. As this discussion has talked about, the over-shift trades singles for doubles, which makes it a bad idea for almost anyone. The biggest problem with the shift is the errors and unforced errors that it creates. I will call an unforced error a ball that is fielded, but the throw is late, or a routine double play that doesn’t get turned. As far as the “data” goes, errors are more or less outs and the over-shift creates a bunch of those, which are positive on paper, but back-breakers in real life.