August 27, 2010
Ahead in the Count
The Clutch and The Shifted
Picture this: David Ortiz steps to the plate in a tie game with runners on first and second in the bottom of the ninth inning. He gets into a deep count and lines a base hit over the right side of the infield to score the winning run. You’ve seen this time and again.
Ortiz is probably the most famous “clutch hitter” in baseball today, which naturally gives rise to arguments about articles you’ve read at this very website, debunking the myth of said clutch hitting skill. Across the population of major-league hitters, the guys who were clutch hitters in 2009 were no more likely to be clutch hitters in 2010 than the guys who were chokers… at least for the most part.
Now picture this: David Ortiz is leading off the ninth inning. He rips the same line drive toward right field, but with no one on base, the shift is on. The second baseman is halfway into right field, the shortstop is on the right side of second base and playing shallow right-center field and the third baseman is playing where the shortstop normally plays. What happens to that batted ball? Frequently, it gets caught by the second baseman.
However, with runners on base, the second baseman is playing closer to second base and the ball gets through. For most hitters, the difference is small between the kind of defensive alignment that prevents hits and the kind that guards against double plays or holds runners on. However, for left-handed sluggers who routinely face “The Shift,” the defense is severely limited against them with runners on base. As a result, hitters like Ortiz are actually “clutch,” and not because they have the mental fortitude to come through in big situations, but simply because higher leverage situations occur when there are runners on base and that is when it is easier for them to shoot a base hit through the infield.
Ortiz actually hits fewer home runs with runners on base—only 4.2 percent versus 6.3 percent when bases are empty. However, his BABIP with runners on base is .319, compared to .280 with the bases empty. As a result, he gets a lot of RBI and the magic “clutch” tag from fans. Sure, he hits plenty of home runs and some of those are bound to be in clutch situations, but those are magnified in the context of getting so many clutch singles and doubles already, making the whole package seem clutch.
For right-handed sluggers, defenses are not really able to put three infielders on one side of the field. The first baseman has to stay close enough to first base that the second baseman also needs to play on the right side of infield to avoid a wide-open hole. Their hits are bound to be distributed more randomly.
Two years ago, I ran a simple test to see whether this effect was significant—did lefty sluggers really increase their BABIP with runners on, and by a larger amount than righty sluggers? In that article, I simply took the top 20 in career slugging percentage among hitters with 3,000 plate appearances, eight of whom were left-handed and 12 of whom were right-handed. The gain in BABIP with runners on for the eight lefties was .022 points, compared to just .009 points for righties. That was statistically significant, and certainly evidence of the effect I suspected—lefty sluggers are structurally clutch. Of course, all eight lefties in that small group were not always shifted against. I certainly did not recall a shift against Todd Helton, but he was in the top 20 in career SLG, and was included in that study. The problem was that I did not have data on who had been shifted against most of the time, so I did not have a perfect way to test it.
However, I decided to do something more intricate this time. Since there was no way to watch every baseball game of the modern era (roughly 1993-2010, in which run scoring has been much higher), I decided to ask a collection of fans who had. I went to SBNation.com, the mother ship of hundreds of sports blogs, including well-trafficked ones about each of the 30 major-league teams. I posted a FanPost, easily visible to visitors of the sites, and asked them to tell me who on their teams had been regularly shifted against, and who their teams had shifted against regularly within their division. I provided the visitors with a list of all lefties and switch hitters who hit at least 20 home runs in a season at some point from 1993-2010.
The posters were fantastically helpful, more than I ever could have imagined. Within days, there were over 400 comments on the 30 FanPosts, giving me a solid list of 27 sluggers who were regularly shifted against. There were another dozen or so ambiguous players, but I generally erred on the side of including anybody who a poster said was shifted against most of the time, unless they were contradicted by another poster. It was a judgment call in a few cases, but regardless of the criteria I used, the results were clear.
Then I came up with a control group of 38 righties who were the regular sluggers of their era. The group was a great control group, since both groups had the exact same split of batted ball locations: 32 percent pulled, 51 percent up the middle, and 17 percent to the opposite field.
The results were very clear. Left-handed sluggers had a statistically significantly higher BABIP with men on than with bases empty, as well as a smaller but also statistically significant increase in BABIP with runners in scoring position versus with the bases empty.
However, they were not more “clutch” in other ways. The groups of southpaws and northpaws had similar changes in home run and strikeout rates in different runner setups.
These two tables summarize the BABIP with bases empty, men on, and runners in scoring position for each of the sluggers in my sample.
*As LHB only, statistics only through 2009
**As RHB only, statistics only through 2009
Compiling all of the balls in play for each of the two groups and computing BABIP leads us to the following table, which tells the story I was looking for with crystal clarity:
Note that the relative increase in lefties’ BABIP with runners on base versus bases empty (as compared to righties’ BABIP increase) is statistically significant at the 99.9 percent level (t = 3.84; i.e. less than 1-in-1,000 chance that you would observe a difference between lefty and righty BABIP spikes with runners on base if they had an equal ability to increase BABIP with runners on base), and the lefties’ increase in BABIP with runners in scoring position as compared with the righties’ is statistically significant at the 95 percent level (t = 2.06).
These lefties were not more clutch in any other way. Looking at home runs and strikeouts, we see that the performance with runners on and with runners in scoring position changed similarly for both groups.
Note that none of these differences are even close to statistically significant, with the highest t-stat coming in at 1.06 (meaning each of the differences had at least a 30 percent chance of being that far from equality of lefties and righties could have occurred by coincidence).
With help from Eric Seidman, I decided to also check whether this was true of all left-handers versus all right-handers, to see if this was simply a fact of handedness, rather than an implication of the shift, and I found that it was the shift.
This BABIP spike is not common to all left-handed hitters, but only to those powerful pull hitters against whom defenses regularly employ the shift.
The implication is clear: there is clutchness, in the sense that left-handed sluggers are able to get more of their hits in crucial situations than right-handed sluggers, even though on the whole, left-handed and right-handed hitters are similar in their ability to hit the ball well in clutch situations.
Next, we have to determine the degree to which this is valuable.
Putting a Dollar Value on Clutchness
Suppose that a player gets 650 PA in a season, and that sluggers get walks about 10 percent of the time. Suppose they put the ball in play (non-HR) about 75 percent of the time like the sluggers in this sample. That means that sluggers put about 440 balls in play in one year.
If the numbers we see above are true: a 20-point increase in BABIP for lefties and a six-point increase in BABIP for righties, then we can look at how many more hits a hitter with the same BABIP would get with men on if he were left-handed and shifted against versus if he were right-handed. In my sample, about 47 percent of balls in play by lefties came with men on, and about 48 percent of balls in play by righties came with men on, so we can say that roughly about half of balls in play for both types of hitters came with men on. If we consider the 14-point difference in BABIP to be about 7 points extra with men on for a lefty and 7 points extra with bases empty for a righty, then that means about 1.54 more hits come with men on for lefties than for righties.
Colin Wyers ran some numbers for me to determine the run value of a non-home run hit with runners on versus one with the bases empty (0.81 vs. 0.34 runs), as well as the relative value of an out (-0.18 vs. -0.39). That means that if we keep the same overall number of hits and outs, but switch one hit that came with bases empty to occur with men on and switch an out with men on to occur with bases empty, the effect would be 0.68 runs extra scored for the slugger’s team. Doing so 1.54 times per season means the lefty sluggers produce about 1.05 more runs with the same batting line as a righty slugger.
That means that a left-handed shift victim is worth about $525,000 more than his right-handed slugging counterpart even if they have the same slash line (since MORP determined that a win is worth $5 million).
How Sabermetrics can Find Clutch Hitters
Outside of perhaps Defense Independent Pitching Statistics, the most contentious issue between traditional fans and sabermetrically inclined fans is the issue of clutch hitting. Sabermetrically inclined fans have seen countless studies looking at how poor the correlation is between “clutch” performance year to year. The primary reason is that the sample of clutch at-bats is usually too small to draw any conclusions.
However, when looking at large populations, we can drive up the sample size and find some interesting facts. In this case, we see that the relative increase in batting average on balls in play with men on for left-handed sluggers who have been victimized by the shift is more than three times the increase in batting average on balls in play with men on for right-handed sluggers. Southpaws do not have some sort of mental fortitude that helps them win games, but their hits come in more important situations.
So, next time you see Ortiz come up to bat with a couple of runners on base in a clutch situation against the Yankees, and your friend tells you that he always comes through in these at-bats, don’t argue. Remember that Robinson Cano and Derek Jeter cannot place themselves where the ball is most likely to go because of the runners on base, and so Ortiz actually does have a better shot at “hitting it where they ain’t.”