Prospectus Feature: The RISP Mystery

In this article, I looked for evidence that some hitters perform better with runners in scoring position (RISP) than in other situations. (I didn’t find much.) To help me define perform, I calculated run expectancies with RISP and discovered that of common batting metrics (batting average, on-base percentage, slugging percentage, and OPS), slugging percentage correlates best to driving in runs. So I looked for hitters who slugged higher with RISP than in other plate appearances.

A couple commenters suggested that this is a low bar. After all, they reasoned, hitters should have a higher slugging percentage with RISP than without, for a number of reasons:

· Pitchers have to pitch from the stretch (though the perception that pitchers give up velocity by pitching from the stretch doesn’t appear to be true)

· Infielders must position themselves to hold runners on, opening up lanes for basehits that wouldn’t exist with the bases empty

· Similarly, with fewer than two outs and a runner on third, infielders may move up to prevent a run, creating additional opportunities for singles

· Batters face lesser pitchers, for the simple reason that good pitchers don’t allow as many batters to reach scoring position as poor pitchers. Last year the 25 pitchers who faced the most batters had a combined ERA of 3.26, FIP of 3.33, and DRA of 3.31 (weighted by innings pitched). The 25 pitchers who faced the most batters with RISP had a combined ERA of 3.88, FIP of 3.83, and DRA of 4.30

All of this makes a lot of sense. And, in fact, batters do perform better with RISP. From 1998 to 2015, players with 400 or more plate appearances in a season batted .279 with RISP and .275 without. That makes sense, given the points noted above. And on-base percentage was considerably higher with RISP, .368 vs. .338, boosted by additional walks, especially with first base open, and over 98 percent of intentional walks occurred with RISP. And slugging percentage… Wait a second. Slugging percentage was lower with RISP, .446 to .447. How can that be?

My first thought was: Mix of players. All those walks, especially the intentional ones, with RISP, are not distributed evenly. Pitchers are more likely to give a pass to a good hitter than a bad one with RISP. So maybe good hitters have a higher slugging percentage with RISP, and bad hitters have a higher slugging percentage with RISP as well, but since the bad hitters get a disproportionate share of at-bats due to all those walks, they drag down the overall average.

To check that, I ranked all hitters with 400 or more plate appearances in a season by TAv and divided them into quintiles. Then I calculated the difference between batting with and without RISP for each quintile and averaged them.* Here are the results, with the difference expressed as batting with RISP minus batting without RISP. For example, for hitters in the second quintile, strikeouts as a percentage of plate appearances were 0.52 percentage points less frequent in plate appearances with RISP (16.147 percent) than in plate appearances without RISP (16.654 percent).

There are some noteworthy trends there. Batting averages are flat or up for all five quintiles. On-base percentages rise, a consequence of not just the higher batting averages but also more walks, particularly intentional walks, especially to the best batters. Slugging percentages are uniformly down, though much more so for the top three quintiles than the bottom two. Singes are up for the best hitters, and doubles and triples are up for everybody. So why is slugging percentage lower? Because home run rates decline for everyone, especially the hitters in the top two quintiles. Top quintile batters hit a home run in 5.65 percent of their at-bats with no RISP, or once every 17.7 at bats. With RISP, their rate falls to 5.14 percent, or once every 19.4 at bats. Home run rates for second quintile hitters decline from 3.95 percent without RISP (once every 25.3 at bats) to 3.66 percent with RISP (once every 27.3).

Why would home run rates be lower with RISP—while, especially confounding, other extra-base hits go up? When we think about the general factors affecting plate appearances with RISP, infield positioning isn’t a factor when we’re talking about home runs, but we’ve already established that batters with RISP are facing lower-quality pitching. Shouldn’t that result in more homers rather than fewer?

Here are some other factors that affect batting with RISP:

· The situations occur in favorable environments, like Coors Field, or Wrigley with the wind blowing out. But that should promote home runs.

· Batted ball location is more important than when there are no RISP. Grounders hit to the right side of the field are more likely to advance a runner from second than grounders ahead of the runner. Similarly, flyballs to right are much more likely to advance a runner from second than flyballs the other way. That means that for right-handed batters, there’s a premium in going the other way, but that may present challenges to the hitter, affecting his ability to hit the ball a long distance.

· Relative to other plate appearances, the value of a single is elevated. In 2015, 59 percent of runners on second scored on a single, so the run expectancy of a single with RISP is either 0.59 (lead runner on second) or 1.59 (runners on second and third). (Obviously, the single also increases the team’s run expectancy for the inning; I’m just talking about run expectancy of a single plate appearance.) Without a RISP, a single has no immediate run expectancy (unless Billy Hamilton’s at first).

· The elevated value of the single reduces the value of a home run or triple, relative to the single. Sure, a home run or triple produces two and one runs, respectively, with a runner on second, but that’s true of a runner on first as well. But a single’s worth a lot more with RISP than it is without. That’s also true, though less so, of doubles, since 45 percent of baserunners in 2015 scored from first on a double. That’s less than the 59 percent who scored from second on a single.

· Both infielders and outfielders may shade in slightly with RISP. That doesn’t affect balls hit out of the park. However, a drawn-in infield decreases the likelihood of a groundball driving in a run while increasing the likelihood of a single.

· Shallow outfielders decrease the chance of a bloop single and make it harder for a runner on second to score on a short single. If outfielders play even a step or two in because of this, the risk of a ball hit over the outfielders’ heads, falling in for extra bases.

· Walks have less value than usual (unless there is also a runner on first). So pitchers may throw fewer of the over-the-plate pitches that wind up in the stands.

· Strikeouts have greater value, since they generate an out in which the baserunners (virtually always) cannot advance.

· Fly balls to the outfield with a runner on third are worth far more than they are in any other situation. Fly balls to right with a runner on second, as noted above, have value as well.

· Bunts are more valuable but are also generally more difficult to execute.

Some of those factors increase the likelihood of a home run being hit. Some have no impact. But some could explain what we’re seeing with RISP: More singles, and more doubles and triples, but fewer home runs.

Further, Rob Arthur of FiveThirtyEight, who provided invaluable help on this project, surveyed Statcast data and found that exit velocities on batted balls average 88.56 mph without RISP and 88.02 mph with RISP. That may sound like a trivial difference, but Rob noted that physicist Alan Nathan has found that of all basehits, home runs are the most sensitive to exit velocity. The difference of just over half a mile an hour in exit speeds between batted balls with and without RISP is probably sufficient to explain most or all of the decline in home run rates with RISP. Better hitters, who hit the ball harder, are the most affected.

This still leaves unanswered the question of why exit speeds are lower with RISP, resulting in fewer home runs, while other basehits are more frequent. I’ll present five possibilities, two involving hitters and one involving pitchers that could affect exit speeds, and two involving fielders that wouldn’t.

Batters may be changing their approach, emphasizing run-scoring singles (which, if they’re hit hard enough to the outfield, may result in more doubles and triples as well) at the expense of home runs.

Pitchers may be changing their approach, targeting fewer pitches over the plate that could be hit for home runs.

Right-handed hitters may try to go the other way, which may sap their power. Last year batters hit .334 with a .602 slugging percentage (.268 isolated slugging, home run every 16 at bats) on pulled batted balls and .312 with a .453 slugging percentage (.142 ISO, 55 at bats per home run) when going the other way. Why would they make that sacrifice? Because batters got flyballs on 19 percent of batted balls when pulled and 53 percent of batted balls when hitting to the opposite field.

Infield positioning (holding runners on, playing in to hold runners at third) may result in more singles and, on balls hit down the line, doubles or triples.

Outfield positioning, specifically outfielders playing in to keep runners from second from scoring on a single, may increase the probability of doubles and triples.

Admittedly, I used the word may in each of the last five paragraphs. I suspect that each is at play, contributing the change. Circling back to the original point, batters, in aggregate, do not generate higher slugging percentages with RISP. RISP have a salutary effect on batting average, on-base percentage, singles, doubles, triples, walks, and strikeouts from the batters’ perspective. But home run rates decline, particularly for better hitters. So the next time you hear fans bemoan a slugger who seems to go deep more often with the bases empty than with runners on, think to yourself: Feature, not bug.

*I took an arithmetic mean of the difference in each of the 18 years rather than an at-bat- or plate appearance-weighted mean. The reason for this is that I wanted to weight each season exactly equally. A weighed mean would’ve given higher-scoring seasons a larger impact, and I didn’t want to do that.