Allen Craig: Clutchiest Hitter Ever?

Jeff Sullivan at FanGraphs took on local hero Allen Craig and asked whether his demonstrated clutchiness is a thing. I think it's an interesting question, but I had a somewhat different reaction and approach which I thought was maybe worth documenting.

Sullivan notes all the good things that happen when Allen Craig steps to the plate with a runner in scoring position.

Craig has a 194 wRC+ with RISP; that's a slash line of .392/.450/.637, for those of you who prefer such a thing.

Sullivan also points out that this is not some miniscule split. After three seasons and 1400 PAs in absurdly productive lineups, Craig has a ridiculous 400 PAs with runners in scoring position, coming to the plate with RISP almost 28% of the time. A lot of hitters would kill for such opportunities.

And Craig has definitely made the most of it, hitting like Barry Bonds when he gets the chance.

So, what's behind this? Is it real or Ordinary Baseball Weirdness (OBW, which I propose should be the substitute for the overused "luck" in saber-talk)?

STRIKEOUT RATE

Sullivan notes that Craig has a huge split in his K rates: he strikes out 19.9% of the time with the bases empty and 12.7% of the time with RISP. That sure looks clutch.

THING OR NOT A THING? Not a thing.

What's driving the K rate? Mostly, he's walking more. Craig walks at a paltry 5.5% rate with the bases empty, but at a 10.9% rate with the bases full. Maybe that's him being choosier at the plate with the bases loaded, but that hardly makes situational sense; much more likely, pitchers are more reluctant to give him a pitch to hit because they're afraid of what he'll do to it.

It's worth remembering that he didn't start out hitting in the three and four slots. In 2011, he took as many PAs (82) in the 5th and 6th slots as in the second spot in the lineup. The number three spot was otherwise occupied at that time. In 2012, he started settling in in the #4 spot, but still saw 104 PAs in the #5 spot. In 2011 and 2012, he was often followed by David Freese who is not the hitter Craig is. Also, if first base is open, some pitchers might be less afraid to walk Craig, even if Holliday or Molina follows,  to open up the possibility of a double play. While all 4 of Craig's IBB have come with RISP, that only accounts for about 1% of the difference in his walk rate. Mostly, he's seeing fewer balls with the bases empty (36%) than with RISP (39%).

I hate using a split of a split, but I'll note that his 2013 walk rate shows a much less dramatic split: 6.6% with the bases empty and 8.8% with RISP. That is somewhat suggestive that Craig is actually getting walked less with RISP this season, maybe because he is frequently followed by Yadier Molina.

Basically, if you consider the situational increase in walk rate (which is driven by pitcher's choices, not his clutch instincts), walking an extra 5.4% of the time makes up for almost the entire gap between his K rate with the bases empty and his K rate with RISP. So, I think the split in his K rate comes almost exclusively from him seeing fewer strikes as a hitter with RISP.

This walk-rate split is reflected throughout baseball, although not as steeply as in Craig's case. With the bases empty, the ordinary player walks 7.2% of the time, strikes out 20.3% of the time, and hits for a .250/.310/.398 line, or a 96 wRC+. With RISP, that same average player hits for a .254/.336/.388 line, with a 10.7% walk rate and a 19.2% K rate. There's actually very little difference in average or slugging in that leaguewide split. Slugging actually drops (probably due to no-doubles defense). Mostly, the difference between hitting with RISP and hitting with the bases empty lies in the increased walk rate with RISP and corresponding increase in OBP.

HIS HITTING STATS

So, sure, maybe you buy the notion that his strikeout rates and walk rates are related. But that's only 23-25% of the time, you say. What about the stuff that's happening the other 75% of the time, when there's a ball in play?

Well, that's pretty striking. As I said above, he has a 194 wRC+ with RISP but only a 118 with the bases empty. He has a .245 ISO with RISP, but only a .174 ISO with the bases empty. He his for a .283 average with the bases empty and a .392 average with RISP. Surely, that must be clutch?

THING OR NOT A THING? Probably not a thing.

I dislike being a buzzkill on this point, because I am so much more tempted to believe in this. It just can't not be true because it's such a great story.

Look, if you've been hanging around this site for any period of time, you know that batting average is pretty weird. It fluctuates a lot. It certainly does not stabilize in sample sizes of 400 PAs. That batting average actually driving most of the difference here. And in those 400 PAs, Allen Craig has a BABIP of .404, which is not sustainable by anyone ever.

The ISO is much less troublesome for me. A difference of 70 points of ISO is meaningful, but seems reasonable in a sample size of 400 PAs. If a player had an ISO dip like that two-thirds of the way through a season, I wouldn't be shocked by it.

The thing that ultimately persuades me is that Allen Craig basically hits for the exact same batted ball profile with the bases empty as with RISP.

With the bases empty, Allen Craig hits 24.3 of the pitches he makes contact with for line drives; with RISP, he hits 25.1% line drives.

With the bases empty, Allen Craig hits 41.6% of all pitches he makes contact with for groundballs; with RISP, he hits 41.9% groundballs.

With the bases empty, he hits 34.6% of the pitches he makes contact with for fly balls (6.6% of them on the infield); with RISP, he hits 33.0% of the pitches he makes contact with for fly balls (3.3% of those on the infield).

In a 400 PA (roughly 300 BIP) sample size, his batted ball profile is a LOT more reliable than his batting average. Both groundball and flyball rates stabilize long before 400 PAs (at 80 BIP), and line drive stats are halfway to significance (stabilizing at 600 BIP). By contrast, a player's BABIP is barely a third of the way to stabilizing in that sample size (needs 820 BIP to stabilize), and his average is not approaching reliability (needs 910 AB to stabilize).

The only thing that looks significant-ish is this improved ISO, which stabilizes in 160 ABs. On the other hand, Craig's .071 isolated power split is the least compelling or interesting split to me. I just don't think deviation from the mean for his ISO is that substantial; it's roughly on par with the variation between last season's ISO (.215) and this season's ISO (.149).

The really troublesome split is the batting average split; .392 versus .283. Is he really making better contact with the ball with runners in scoring position?

Here's where I go back to his hitting profile. If Allen Craig had the magic ability to hit the ball harder with runners in scoring position, we'd probably see him hit more line drives and maybe more grounders. If he's hitting for more power, we should see more line drives and more fly balls. He's not. His LD rate, GB rate, and FB rates are all basically indistinguishable in comparing his RISP profile to his bases-empty profile. Unless you believe that Craig is really good at hitting them where they ain't when there are runners in scoring position, he's the same batter with guys on base as without.

As John Mozeliak put it, "enjoy the ride as we take it," don't fret about when he'll regress or hang our hats on him being amazing with runners in scoring position forever. I think when you take a careful look at the most granular stats, the ones most likely to be sustainable in this 400 PA sample size, they mostly show no difference between Craig's hitting profile with runners in scoring position and with the bases empty. I'm glad his success with runners on is happening, I enjoy it while it's happening, but I'm not counting on it continuing.

The other bit of good news - as I noted when I looked at team-wide clutchiness - is that his bases-empty outcomes and his RISP outcomes should regress towards one another. More balls will find gloves while runners are in scoring position, but he'll also hit better in the future with the bases empty and fill the bases for other Cardinals to drive him in. He's not a true-talent 192 wRC+ hitter, but he's also not a true talent 118 wRC+ hitter.