Pebble Hunting: Extrapolating the Breakdown of Traditional Defense

One of the most interesting things about extreme infield shifts is how unextreme they are. They are like some lame grownup’s idea of extreme, a little bit of flash and inconvenience but ultimately very safe. The shift was invented by sane people. Real extreme comes from insanity, and it makes us deeply uncomfortable.

Everybody’s talking about the football coach who never punts​—4th and 15 at his own five-yard line, he’s going for it. That’s fearless. It’s hard to think of a baseball equivalent, one that would work or even one that might work. Russell Carleton this week explored the listener-suggested idea of having the left and right fielders swap, depending on batter handedness, to make sure the better defender gets more attempts to field the ball. The gory math supports the use of the relatively conservative proposal, but Carleton concludes what we can't help but conclude:

In United States culture, it’s not nice to look like you’re trying too hard, especially for such a small reward and when you look weird doing it. As the infield shift has become more and more popular, I find it interesting to hear some of the reactions from teams that don’t shift. Usually, they have moralistic underpinnings. We’re gentlemen here. We don’t shift. We could, but that’s for the riff-raff to do. Brian McCann would totally shoot this down. Still, they’re leaving runs on the table. I have to wonder at what price come pride and tradition.

Meanwhile, Grantland/Crashburn Alley writer Michael Baumann recently came out against defensive shifts. Baumann’s rationale is a bit complex: He thinks it’s actually too easy to beat, and considers hitters’ refusal to beat it a self-destructive act of stubbornness that requires paternalistic intervention. More simply, though, we might say he considers the whole business annoying.

So Baumann is against shifts. Let’s say I’m for them. This is not an inconsequential divide. We’re each likely to live about 50 years more, and in those 50 years the shifts quite possibly will get more extreme. When they do, there will be people who want them outlawed, particularly if they work. It’s time to choose a side, people.

I’m on the side of disruption. The key to disruption: small steps.

Step 1. Corner outfielders swap positions in sacrifice fly situations. There might not be a lot of teams with substantial mismatches between their left and right fielders’ overall defense, but it’s not hard to find three-grade differences in throwing arms—Yasiel Puig and Carl Crawford in Los Angeles might be a five-grade difference. Batters go the other way two-thirds of the time on fly balls, Russell found, and one might hypothesize that batters trying to hit a fly ball in a sacrifice situation would go the other way even more often, since pulling the ball is a good way to roll over it. The unambiguousness of throwing ability, and the ease of implementing this tweak—one switch per game isn’t nearly the time burden of eight—makes this a very low-stakes strategy. It’s also unlikely that a manager would have to worry about being second-guessed.

Big Change because: It breaks down the idea that players are affixed to their positions. As it is, even the radical infield shifts are matters of shading, not redefining (with one exception being Brett Lawrie; as a former second baseman, his shift from third base to shallow right field is the exception that proves the rule). If players were moving hundreds of feet from play to play depending on situation, the idea of a “position” would be watered down, if not endangered.

Step 2. Shortstop plays third base in bunt situations. There are really only two plays in baseball in which a team can say it knows what its opponent is going to do. When the catcher stands up and holds out his glove, we know with 99.9 percent certainty that the next pitch will be a ball—but, as the entire point is to keep a play from happening, no countermeasure is really possible. But when the batter is a pitcher and he squares for a bunt, we can say with better than 90 percent certainty that he's going to try to bunt. Defenses are arranged based on uncertainty, on the probability that the ball is going to be hit to some locations more often than others but that there’s way too much variance to leave any square footage uncovered. The team’s best defender plays where the shortstop plays because it’s slightly more likely that a batted ball (particularly one requiring his particular skills to field) will go there than anywhere else. When the opponent declares that to no longer be true, the defense should adjust by putting its best defender where the ball is going. No matter who the Braves have playing third base over the next five years, Andrelton Simmons will be better at charging bunts, better at fielding bunts, better at throwing to first and better at getting the lead runner out. It’s a very small gain, but also a very low-effort one to pursue.

Big Change because: It changes the shortstop’s role from that of a defensive back (responsible for covering one thing well) to free safety, covering any part of the field most likely to require his standout skills. If a fly ball pitcher is facing a fly ball hitter in a fly ball situation, that might mean the outfield. It might mean second base. In a bunt situation, it’s certainly third base.

Step 3. Four outfielders in a no-doubles situation, a fly-ball situation, or an infield shift situation. In any shift that leaves a big part of the field uncovered, there’s a tension between the OBP and the SLG components of offense. Late in games, when defenders hug the lines to avoid extra-base hits, teams are more likely to give up single because they’re out of position, but they weigh the value of each, and they make that choice. If they really wanted to prevent doubles, they could do it—everybody plays outfield!—but each double prevented means more singles, fewer outs, etc. Clearly, six outfielders is too far, but there’s no reason to think that teams’ current modest concessions are exactly far enough. How about this dumb thing:

As for infield shifts: Ryan Howard hit, if Brooks Baseball is to be believed, 15 groundballs to the left side of the infield in the past two years. Five were hits—one was a double—and the other 10 were outs. (Some of these were probably hit in non-shift situations.) Meanwhile, he hit 95 fly balls that stayed in the park, and 17 of those were hits—13 of them doubles. Assume for a moment that he has no ability to hit more groundballs against the shift than he has shown. Turn those 10 groundball outs into singles but eliminate eight of the 13 doubles, and the defense would clear a small profit.

(We assumed for a moment that Howard has no ability to hit more groundballs against the shift than he has shown. This is an assumption that Baumann won’t make—he assumes hitters are just refusing to adjust, and shirking their game theory obligations. I’m not sure that I agree. One of the unknowns is whether Howard and his ilk are actually capable of hitting the ball toward an abandoned position. Some hitters can, but it’s not clear that Howard can.)

Big Change because: Four outfielders! Once there are four outfielders, it doesn’t look like baseball anymore, and anything’s possible.

Step 4. Switch third and first basemen depending on batter handedness. This one gets tricky. Unlike swapping the left and right fielders—both of whom can play either position without shaming himself—this sets up a situation where one player would be a tremendous liability. Prince Fielder playing third base. Maybe even a left-handed player playing third base? And, because a third baseman doesn’t have the time to let a ball bounce off his chest and still recover, a higher percentage of plays at third are in the margin where a good defender makes the play and a bad one doesn’t. Probably doesn’t work! But the pull/oppo tendencies that Russell found in fly balls are way higher for groundballs.

He ran the numbers for me, and this is how big the gap is: There are roughly four times more gounders hit toward the corner infielder on the pull side than hit toward the corner infielder playing the other way. (This includes grounders fielded by the corner infielder or the corner outfielder, but not the second baseman, shortstop, or center fielder.) So, when a left-handed batter is up, he’s four times more likely to hit it toward Chris Davis than toward Manny Machado. This seems like an absurd waste of resources.

Russell adds:

This is the same basic construction as the [fly ball] article, except more profound. If you could make the jump that two fielders could be proficient at both first and third, and that one would be comparatively better than the other at both positions, it would make sense to flip them around.

Wow, that's powerful in terms of its effect size because the ball is pulled so much more on grounders than on fly balls… it's probably harder to pull off than the corner OF, but logically, the only thing that you'd have to sell me on is that, all things considered (range, arm, etc.) Smith is both a better first baseman than Jones and a better third baseman than Jones. (I'd also have to buy that the constant switching, both mentally and physically—jogging across the diamond—wouldn't be a net loss.) The difference might not be worth much, but it would be net positive value to have them switch back and forth. I could probably run similar numbers for the SS and 2B.

In theory, you could make the case that there could be a team where the proper thing to do as the batter's handedness switched is to flip six positions around, leaving only the battery and center fielder (and DH) in their proper places.

Big Change because: Unlike corner outfielders, who are essentially mirror images of the same position, infielders’ position identities are strong and specific. This would go much further to breaking down the idea that the position after a player’s name on the lineup card means anything. It’s also an easier shift—it can be done without delaying the game—and so it could make position switching a relatively painless custom.

Step 5. Five-man infield when the pitcher is hitting. Maybe even when the pitcher isn’t hitting, but especially when the pitcher is hitting, and especially when he’s hitting against a hard-throwing pitcher.

In Clayton Kershaw’s career, opposing pitchers have put roughly 115 balls in play. Of those, seven—slightly more than one per year—were either line drives pulled or fly balls pulled. Once per year, a pitcher pulls the ball in the air against Kershaw. The pitcher batting against Clayton Kershaw is 10 times more likely to hit a groundball than to hit a fly or liner to the opposite field. (He’s four times more likely to hit a fly or line drive to center or the opposite field than to pull it.)

Meanwhile, Kershaw has allowed 10 groundball singles through the infield. And of the six non-deep line drive singles he has allowed, at least some (one of two I spot-checked) would have likely been caught in a five-man infield.

Strangely, of the seven balls that pitchers have pulled in the air, six have been hits, two of them going for doubles. If we believe that rate then we’d say there’s practically nothing to lose here; a few singles turn into doubles, but singles turn into outs. Would have to assume that the high rate of hits on pulled flies is a fluke, though. At most, seven outs turn into seven doubles. At the very most. Meanwhile, the five-man infield has between 10 and 16 chances to turn hits into outs. There’s hope here.

What seems more certain is that a fifth infielder should be brought in when the pitcher is bunting. We’d have to hypothesize that there’s practically no chance that a pitcher could pull a fly ball or a line drive against Kershaw in a butcher-boy attempt, so once that bat is squared the left (or right) fielder is totally out of the play. Even if the pitcher doesn’t square early, though, we already know how hard it is for pitchers to pull the ball in the air against Kershaw; it would be especially so if Kershaw were specifically pitching away from this possibility. With five infielders, the corners could crash to just an absolutely ludicrous degree. If the offense adjusts and decides to swing away with such low chances of bunting success, then the hitting team’s manager has had his preferred tactical weapon taken from him.

Big change because: No four-man infield and three-man outfield construct anymore. Just a nine-man defense, as it should be.

My guess is that Steps 1 and 2 work but to an extremely small degree; that step 4 would work to a fairly sizeable degree, at least on a lot of teams; that step 5 would work, but maybe wouldn’t; and that step 3 is stupid. The point isn’t necessarily that any would work. It’s that there’s got to be something out there that would, something that would make me and Michael Baumann a little uncomfortable, and I’m in favor.