On Allen Craig, the Magic Man

Allen Craig, so far, has been one of baseball’s all-time best hitters with runners in scoring position. Yes, as in ever. This year, he’s leading the league in batting average in that split. Last year, he led the league in batting average in that split. He is the leader, all-time, in such situational batting average, and his lead isn’t small. If you sort by OPS, Craig’s near the top of the historical leaderboard — between Albert Pujols and Joey Votto. Either the man has a knack for delivering important hits, or he just makes people feel that way.

The other night, Craig batted in the bottom of the seventh in a game between the St.Louis Cardinals and the Cincinnati Reds. The Reds were leading 5-4, but the Cardinals had the bases loaded with two outs. J.J. Hoover threw Craig a first-pitch fastball, up and away, and Craig swung and took it up and away for a deciding grand slam. For the Cardinals, it’s been a magical season when it comes to runners in scoring position, and no one’s had more magic than Craig. And for him, this is just more of the same. Which leads one to wonder: What’s going on with Allen Craig, anyway?

Craig, of course, is simply a good hitter. Good hitters deliver good hits. But over his career, with runners in scoring position, he’s batted .392. He’s slugged .635, and if you put it all together, you get a 194 wRC+. Moreover, Craig just recently passed 400 career situational plate appearances, so we’re not talking about a one-season fluke. What Craig’s done is the kind of thing that gets an analyst’s attention. While no firm conclusion is offered below — I haven’t solved the Allen Craig mystery, either — this begs to be investigated. Here are some assorted points. This is the result of a deeper look into Craig’s career splits, born of a reluctance to just simply say he’s going to regress.

(1) The strikeouts

When you see a split that’s unusually big, you wonder if it’s capturing something real. When you wonder if it’s capturing something real, you look for other indicators, signs of a changed approach or whatnot. With the bases empty, Craig has hit grounders 42% of the time. With runners in scoring position, he’s hit grounders 42% of the time. There’s nothing there. But some people have already observed the difference in strikeouts. With the bases empty, Craig has struck out 20% of the time. With runners in scoring position, he’s struck out fewer than 13% of the time, and that’s a significant gap. That’s an increase in balls in play, and that could be the result of a deliberate process. The drop in strikeouts convinces some people that Craig really is quite different.

I’m not going to argue passionately otherwise, but I’ll note that Craig’s strikeout split isn’t unprecedented. I looked at all the active players with at least 250 plate appearances with the bases empty and with runners in scoring position. I calculated their split strikeout rates, and the correlation between them is .92. Here are the top five guys with the biggest strikeout-rate reductions, in percentage points:

  1. Travis Ishikawa, -7.4 percentage points with runners in scoring position (K%)
  2. Allen Craig, -7.1
  3. Travis Snider, -6.9
  4. Jonny Gomes, -5.6
  5. Wilson Valdez, -5.5

Jeff Mathis ranks sixth. I don’t know what this list is telling me, if it’s telling me anything at all. For most players, strikeout rates with runners in scoring position look a lot like strikeout rates with the bases empty. Craig’s been different, but Travis Ishikawa has been just as different. (And do we have any particular opinions about Travis Ishikawa?)

This is a point neither for nor against sustainability. Craig isn’t the first player to trim his strikeouts when there are runners to drive in.

(2) The sample

Let’s just put this out there: This is Craig’s fourth season in the majors. When a guy has been showing a trend over four seasons in the majors, one is inclined to believe in it. After all, four years is a lot, right? Sure, but we’re not looking at overall performance — we’re looking at splits, and Craig just passed 400 plate appearances with runners in scoring position. He has the equivalent of a fraction of one whole year in that split, so we can read into it only so much. Over a stretch of 393 plate appearances between 2011 and 2012, Craig posted a .979 OPS. Over his next 416 plate appearances, he posted a .770 OPS. The cliche is that baseball analysts are always talking about fluctuation and randomness. Fluctuation is real, and it’s important for us to understand and appreciate.

(3) The complication

Over more than 400 plate appearances with runners in scoring position, Craig’s batted .392. That is, of course, extraordinary. If one were to look for explanations, maybe Craig’s gotten easier pitches to hit. Maybe he thrives against pitchers working out of the stretch, and maybe he’s fond of different defensive alignments. Maybe Craig just mentally bears down when there’s someone else to bring home.

But then there’s this: Craig has batted 268 times with a runner on first. He’s hit .254, with a .710 OPS. His strikeout rate is the same as it is with nobody on. If there’s something about different pitches, we don’t see it here. If there’s something about pitchers working out of the stretch, we don’t see it here. Hitting with a runner on first doesn’t have the same feeling as hitting with runners in scoring position, so this leaves open the psychological possibilities. But Craig’s been worse with a man on first than he’s been with no men on. That’s difficult to reconcile with a belief that we’re not observing interesting randomness.

(4) The clutch factor

Read that a guy does well with runners in scoring position, and you’re going to see him as clutch. You might want to believe he’s just able to focus extra hard when the pressure is on, that he can achieve a higher level. According to FanGraphs, Craig owns a 140 career wRC+ in high-leverage situations, and a 136 wRC+ in the other situations. According to Baseball-Reference, Craig owns an .879 career OPS in high-leverage situations, and an .849 OPS in the other situations. Those differences are, basically, too small to worry about. To whatever extent Craig’s performance with runners in scoring position has to do with his being clutch, by other measures he hasn’t been particularly clutch or unclutch. By WPA, his career Clutch score is actually slightly negative, not that that’s something we talk about often.

(5) The spray chart

As you can imagine, Allen Craig has been asked about Allen Craig, before. Here’s an article from the USA Today from not too far back. Said Craig:

“The name of the game is giving yourself the best chance to succeed,” he says. “If you’re trying to pull the ball in that (RISP) situation, you leave yourself susceptible to anything away. It’s always good to keep the ball in the middle of the field. That’s not to stay I never try to pull the ball with runners in scoring position, but for the most part, it’s trying to hit the ball hard on a line somewhere.”

According to Baseball-Reference, with the bases empty, Craig has hit 56% of his balls in play up the middle. With runners in scoring position, that jumps to 64%. That supports the idea that Craig has just been looking to make contact and spray the ball. The strikeout rates also support the same idea. More confusing: Craig hits for more power with runners in scoring position. So if he’s cut down his swing, he hasn’t hit like it. Craig hasn’t traded power for contact and spraying. He’s gained contact and power and spraying.

To get back to the main point, what we see here is the slightest hint of the idea that something is different in Craig’s head when there are runners on second and/or third. So that could be a thing.

(6) The words

On the other hand, from the same article:

“I feel like I have the same focus every time,” says Cardinals All-Star slugger Allen Craig, the game’s best hitter with RISP. “I’ve just had a lot of opportunities with runners on base this year. It’s nothing I try to explain.”

And the Cardinals’ general manager, John Mozeliak, speaking not specifically about Craig, but about his team:

“At some point, you typically expect some regression to the mean or some normalcy,” says Cardinals general manager John Mozeliak, who is among the new breed of GMs who place great value on analyzing statistics. “Having it be sustained year in and year out isn’t a good strategy. But it could defy and be a one-year outlier.”
“I don’t think from a statistical approach you can say this is sustainable for additional years,” Mozeliak says. “For us, enjoy the ride as we take it.”

Craig doesn’t think he’s a different hitter with runners in scoring position. The guy in charge of the Cardinals figures this’ll eventually all balance out. He didn’t even try to suggest this is part of some organizational plan. Mozeliak spoke of his own team like a FanGraphs writer, and you figure he wouldn’t have done so if he believed even a little bit that this situational success is sustainable. Presumably, the Cardinals have investigated, being the team responsible for all this statistical confusion. Ask them and they’ll shrug their shoulders.


This is the part I hate writing: Allen Craig has a really interesting career split. It’s my belief that, over time, things will balance out. That’s such a stereotypical analyst perspective, but the fact is that most things regress, and regression spoils a lot of would-be fascinating stories. Craig, for his part, denies his approach changes much depending on the situation. The guy who gave Craig his contract doesn’t think there’s more to good hitting than good hitting. The only reason to believe something’s different is what the numbers say, and the numbers are based on limited samples. And so on and so forth — you know this part by heart.

Allen Craig, at least, has made a lot of people curious. Allen Craig, at least, has caused people to notice. Allen Craig, at least, has left me not completely certain he’s going to regress all the way. I don’t know if there’s really something about Craig, but I do know that I don’t know that. And that has to count for something.

Print This Post

Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.

45 Responses to “On Allen Craig, the Magic Man”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Josh M says:

    I don’t get it, if he changes his approach with RISP (not trying to pull the ball) why not use that approach all the time?

    Vote -1 Vote +1

    • Kumar says:

      The fielders change their position when there are runners on. This can definitely pave the way for a different approach from the batter that wouldn’t work otherwise.

      Vote -1 Vote +1

    • Wildly Inappropriate Comment Guy says:

      Probably has something to do with his huge cock.

      +11 Vote -1 Vote +1

  2. Highball Wilson says:

    The runner on second is stealing signs, obviously.

    +24 Vote -1 Vote +1

    • Alex Anthopoulos says:


      Vote -1 Vote +1

    • Trent Phloog says:

      I’m not sure if you were kidding or not, but this thought occurred to me seriously. With a runner on first, he hits like Nick Punto. With a runner on second, he hits like Babe Ruth.

      What’s the likelier explanation: The mountains of data showing that “clutchness” isn’t a thing are all wrong; or the runner on second is giving Craig some kind of an edge?

      Vote -1 Vote +1

  3. DrBGiantsfan says:

    You really could just dispense with the articles on this site and post a single statement in larger letters:

    All Baseball Statistics Regress to the Mean

    Vote -1 Vote +1

  4. Career OPS in innings 1-3: .918.
    Career OPS in innings 7-9: .788.

    Career OPS “late and close:” .676.

    When is a difference meaningful and when is it just a thing? Late in games, with the tying run on deck or closer (situations that should be clutchy), he not that good. He performs well with men in scoring position early in games and late when the score isn’t close. That can’t be a skill, especially when he and his GM don’t seem to think so.

    This is just an idle thought, but could the Cardinals have developed some sort of sign stealing routine that’s gone unnoticed? I’m not talking about cheating, I’m talking about a relay from the guy standing on 2B. Theoretically possible explanation for their RISP numbers. No evidence to back it up, just wondering.

    Vote -1 Vote +1

    • orthotds says:

      If the sign stealing theory were correct, we’d expect to see other Cardinals hitters over the span of Craig’s career post similar splits with RISP, right?

      I guess you could complicate things further and say that maybe some guys are see-ball-hit-ball type hitters and don’t even want to know what’s coming, but at least a decent portion of hitters would benefit from knowing what type of pitch the next one will be.

      It would seem weird if the Cardinals were stealing signs and basically feeding the info only to Allen Craig.

      +7 Vote -1 Vote +1

      • Aaron Lehr says:

        While Craig is the headline, the story doesn’t end there. Cardinals take up half of the top 10 hitters with RISP this year, and lead the NL as a team by a large margin, so it isn’t just Craig. Not necessarily trying to support the sign stealing idea, just adding some information.

        Vote -1 Vote +1

        • Sparkles Peterson says:

          “Screw you, Freese. You’ve already got a clutch reputation!”

          But really, if the explanation for Craig’s RISP numbers is sign-stealing, what was it last year when the Cardinals got worse with RISP?

          Vote -1 Vote +1

    • thistakesgumption says:

      I would expect ops to decline late in games because you’re batting against relivers/closers. need OPS normalized against league average in those situations to say anything meaningful

      +15 Vote -1 Vote +1

      • jessef says:

        you’re also batting against garbage time mop-up guys sometimes. Probably more often against better relievers but not 230 points of OPS more often.

        Vote -1 Vote +1

        • Lanidrac says:

          Those mop-up guys are usually going to be pitching earlier in the game after the starter gets yanked early.

          Vote -1 Vote +1

    • tz says:

      Here are some 2013 stats from bbref, splitting by whether there was a runner on 2nd base vs. not (ie. 2, 12, 23, and 123 vs. 0, 1, 3, and 12 base-states):

      MLB total, no runner on 2nd: .714 OPS (excluding IBB)
      MLB total, runner on 2nd: .680 OPS excluding IBB
      Difference: .034 worse with runner on 2nd base

      St. Louis, no runner on 2nd: .710 OPS (excluding IBB)
      St. Louis, runner on 2nd: .825 OPS excluding IBB
      Difference: .115 better with runner on 2nd base

      I can potentially understand why having a runner on 2nd base could distract a hitter, all else being equal. But there is no reason to expect there to be such a big improvement as what the Cardinal show.


      +14 Vote -1 Vote +1

      • tz says:

        Forgot to put a after the word better. I’m not trying to make a bold assertion here.

        Vote -1 Vote +1

        • Dave Cornutt says:

          Yes, interesting. That does somewhat support the notion that the Cardinals have developed some novel way to steal signs from 2B and relay them to the hitter. I’m trying to think of some way to further isolate this (e.g., day vs. night games, or isolating cases where the runner at 2B steals 3B), but I can’t think of anything that would produce a large enough sample size.

          Vote -1 Vote +1

        • olethros says:

          Maybe they’re stealing location rather than pitch type?

          Vote -1 Vote +1

        • Tim says:

          Dave, I would expect it to show up in their swing frequency by location, and/or possibly by pitch type, as compared to the same in non-RISP situations.

          Vote -1 Vote +1

      • jessef says:

        not that it’s the case, but it’d also explain why the players and front office would chalk it up to statistical noise

        Vote -1 Vote +1

  5. DrBGiantsfan says:

    Since Allen Craig is a pretty good hitter overall, he could regress to HIS mean and still be pretty good with RISP, no?

    Vote -1 Vote +1

    • Scraps says:

      “Craig, of course, is simply a good hitter. Good hitters deliver good hits.”

      “This is the result of a deeper look into Craig’s career splits, born of a reluctance to just simply say he’s going to regress.”

      Vote -1 Vote +1

  6. JLuman says:

    Is he seeing a more hit-able pitch mix with RISP?

    Vote -1 Vote +1

    • steve says:

      That’s my thought… is that pitchers inherently change the way they pitch with RISP. Usually, that approach leads to more Ks, GB, whatever. However, maybe that approach is less effective against hitters like Craig than most hitters, where such an approach evens out. I have also heard him note that he is a good curveball hitter, and he tends to see more of them with RISP. Can anyone investigate whether any of that is true?

      Vote -1 Vote +1

  7. WL says:

    You should also think about the selection bias in the numbers. When runners are one base there will always be an inferior defensive alignment to keep pickoff threats alive. Plus, the pitcher has already given up at least one baserunner, and if the runner is in scoring position, has given up at least one baserunner or allowed the one to move up. Either way, this suggests the pitcher is doing worse than an average pitcher in an average inning. Both of these would inflate the batting average for a groundball hitter who doesn’t strike out much. Finally, he’s off the charts on wFB, indicating that he crushes fastballs, which (anecdotally) is something that people will be throwing when they’re trying to stabilize an inning.

    All this needs to be confirmed with data, but it wouldn’t surprise me if you looked at RISP AVG minus AVG, this value will be correlated with gb%, wFB, and k%.

    Vote -1 Vote +1

  8. Ann Wilson says:

    He’s got the magic hand.

    +5 Vote -1 Vote +1

  9. Mark says:

    This reminds me of a few of David Ortiz’s seasons when he had huge numbers in RISP and other situational splits, so much so that the Red Sox gave him a silly plaque declaring him the “greatest clutch hitter” in team history. But when you look at Ortiz’s career splits, you see that it wasn’t a repeatable skill. The splits are in close alignment with his career numbers.

    Vote -1 Vote +1

  10. T says:

    Maybe we’re just asking the wrong question. Why is Allen Craig anti-clutch with fewer runners on? Maybe he relaxes when the rest of the team (more than one of his teammates) is behind him. His personality is already laid-back in a team full of assertive players, while his numbers necessarily push him to the forefront.

    Maybe he’s not going to regress to a lower mean.

    Maybe if he learns to relax with no runners on… then we’ll really see how good he is. The world will know fear the day he comes out of his shell.

    Vote -1 Vote +1

    • Scraps says:

      That’s always my thought when hitters are good in the clutch.

      Maybe they’re good in the clutch but average the rest of the time because otherwise their head would explode.

      Vote -1 Vote +1

    • Tim says:

      That would suggest that Craig’s true talent level is w.454, which would make him the sixth-best hitter in baseball history. That strikes me as unlikely.

      Vote -1 Vote +1

  11. rusty says:

    For me, the most instructive comparison is his RISP splits vs his high-leverage splits. I would expect an underlying change of approach or psychological drive to show up in both cases, rather than just the former.

    Vote -1 Vote +1

  12. bob says:

    Any way to see if the alteration in defensive positioning with men on somehow leads to increased babip/his typical hits getting through rather than getting scooped up by a fielder?

    Vote -1 Vote +1

    • BVHeck says:

      I can tell you, anecdotally, Craig does not see infield shifts with any regularity. Cardinals that see the shift most often are Adams and Beltran.

      Vote -1 Vote +1

  13. Mister says:

    What happens to his plate discipline numbers with RISP? It would be really fascinating if it turned out that he actually did change his approach with RISP and swung more often, tried to hit for contact, and send the ball the other way, and then this changed approach just makes him a better hitter all-around. Generally a hitter who goes for contact has to sacrifice some power to do so, but maybe in Craig’s case he is just naturally more comfortable as a contact hitter. If that is the case, then he should adopt this approach for all his plate appearances.

    Vote -1 Vote +1

  14. Tim says:

    I’d like to see less speculation about mystical psychological abilities which we can’t test for, and more about sign-stealing and recognition of a limited pitch repertoire, which we can.

    Vote -1 Vote +1

  15. sambf says:

    With 400 PAs, the batting average is going to have a standard deviation from luck of about 50 points; finding a two sigma outlier–as Craig is– out of a pool of >> 20 hitters, isn’t unexpected just from random chance.

    Vote -1 Vote +1

  16. beelicker says:

    Check out how many times Craig gets AB RISP & with FIRST BASE likewise occupied i.e. with no place to put him & MUST be pitched to?

    Vote -1 Vote +1

  17. Mr Punch says:

    Assuming equal pitching (which it isn’t – I’ll bet clutch existed before there were so many relievers)wouldn’t we expect a good hitter to have a higher BA and fewer Ks with RISP? All he needs, after all, is a single.

    Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>