Contact and Strikeouts: The Mystery of Nick Castellanos

When the Tigers decided to swap Prince Fielder for Ian Kinsler — side note: Kinsler is already +1 WAR ahead of Fielder on the season, reinforcing just how great a deal that was for Detroit — it was done, in part, to open up a spot on the field for top prospect Nick Castellanos. Fielder’s departure meant that Miguel Cabrera could move back to first base and Castellanos could take over at the hot corner, improving their defense at both positions. And Castellanos’ minor league track record suggested he would bring some offensive upside to the table as well.

So far, he’s basically lived up to expectations. He came in to the day with a 98 wRC+, right in line with what Steamer and ZIPS forecast before the season began, and that was with a .256 BABIP; give him some positive regression on that front, and the overall package looks like a slightly above average offensive player. While Castellanos is more of a good-at-everything-great-at-nothing kind of hitter, his most positive attribute so far has been his below average strikeout rate; at just 15%, he’s striking out about 25% less than a league average hitter this season.

But while we could have expected a better-than-average strikeout rate from Castellanos based on his minor league track record, the fact that he’s striking out so rarely is actually pretty weird. Because Nick Castellanos, for the first few weeks of 2014, has the 9th lowest contact rate in all of baseball. He has about the same contact rate as Curtis Granderson and B.J. Upton. He’s making contact less often than Ryan Howard, Giancarlo Stanton, and Adam Dunn.

The relationship between contact rate and strikeout rate is very strong. Here’s a plot of the two metrics for all qualified hitters in 2013.

ContactKs2013

That is a very strong linear trend, and as you can see from the regression equation, the r squared between the two is .85, meaning that nearly everything you need to predict strikeout rate can be found by looking at contact rate. Contact rate doesn’t perfectly predict strikeout rate, but it is the dominant variable, and knowing contact rate gets you most of the way to knowing strikeout rate.

This is pretty common sense, of course. Guys who swing and miss not only are going to swing and miss at more two strike pitches, but also get to more two strike counts to begin with, since they were swinging through zero-strike and one-strike counts that higher contact hitters would have put in play, ending the at-bat before a strikeout was ever possible. It isn’t any kind of revelation to say that contact rate and strikeout rate are highly correlated.

But now, look at the same graph, only focusing on 2014 data.

ContactKs2014

Because we’re dealing with smaller samples, the plot is a lot more scattered; this will even out as the season goes on, and by the end of the year, the 2014 plot will look a lot like the 2013 plot. You’re always going to get more variance in smaller samples, so we shouldn’t be surprised that the spread is wider here.

But even with that wider variance, look at the lower left hand corner of that graph. Note the point just above the 15% K% mark, the furthest dot from the line. That’s Nick Castellanos and his 67% contact rate. To the right of him is Marcell Ozuna, who also has a very low strikeout rate relative to his contact numbers, but every other hitter in baseball who is making contact less than 70% of the time has a strikeout rate over 25%, and many of them are closer to 30%. In fact, if you use the regression equation to create an expected strikeout rate based on contact rate, Castellanos’ xK% is actually 30%; he is striking out half as often as his contact rate would suggest.

No hitter in baseball is further from their expected strikeout rate than Castellanos, in fact, no one else is even all that close. Ozuna, as mentioned, is also in the low contact/moderate strikeout group, but his 18% K%/28% xK% difference is only 10 points off the regression’s formula. Castellanos is far and away the biggest outlier when it comes to low contact and low strikeout rate.

And no, this is not sustainable. Last year, Hunter Pence had the biggest positive difference in terms of strikeouts versus expected strikeouts based on contact, and he beat the regression by six percentage points. Only eight of the 141 qualified 2013 batters diverged (in either direction) more than five percent from their expected K%. Castellanos does have the profile of a guy who will strike out less than his contact rate would suggest — aggressive hitters who swing a lot get to fewer two strike counts than guys who work the count, so their swings and misses come more often in zero-strike or one-strike counts — but something is going to give.

Of the two metrics, contact rate stabilizes quicker than strikeout rate, which probably isn’t great news for the Tigers. If Castellanos keeps swinging and missing anywhere near as often as he has been, his strikeout rate is going to spike. An uptick in BABIP may very well offset the rise in strikeout rate, and this isn’t any kind of sign that Castellanos is about to stop being a productive hitter, but he is going to stop being this kind of productive hitter. You just can’t swing and miss as often as he has been and post better-than-average strikeout rates.

For reference, are are the top five and bottom five hitters in variance from expected strikeout rate for the start of the 2014 season.


Name PA Contact% xK% K% Difference
Nick Castellanos 59 67% 30% 15% 15%
Marcell Ozuna 89 69% 28% 18% 10%
Aramis Ramirez 90 78% 21% 11% 10%
Albert Pujols 97 82% 18% 8% 10%
Andrelton Simmons 78 88% 13% 4% 9%
—– —– —– —– —– —–
Ruben Tejada 67 83% 17% 25% -8%
Joe Mauer 96 82% 17% 26% -9%
Brett Gardner 77 84% 16% 25% -9%
Abraham Almonte 94 73% 25% 35% -10%
Garrett Jones 90 78% 21% 33% -12%

Count-working types like Mauer and Gardner will likely continue to post higher strikeout rates than their contact rates would suggest — at 8%, Gardner had the biggest difference of any hitter in 2013, so this isn’t new for him — and guys like Ramirez and Simmons will likely not regress all the way to their xK% because of how often they swing, but these ranges are going to come down. For hitters like Castellanos, the hope has to be the contact rate is the thing that regresses more than the strikeout rate, but something is going to have to give.




Print This Post



Dave is a co-founder of USSMariner.com and contributes to the Wall Street Journal.


23 Responses to “Contact and Strikeouts: The Mystery of Nick Castellanos”

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

    He must be having a good run of putting 2-strike pitches into play, and/or very few called third strikes against him. His BABIP is low this year, so perhaps he’s just very focused on putting the ball in play with 2 strikes on him.

    Vote -1 Vote +1

  2. ralph says:

    Dave opens the article with this: “When the Tigers decided to swap Prince Fielder for Ian Kinsler — side note: Kinsler is already +1 WAR ahead of Fielder on the season, reinforcing just how great a deal that was for Detroit”

    So since that sort of logic is apparently kosher, I’m pretty sure I’m safe saying the Doug Fister deal was also great for Detroit since Robbie Ray has a 1.93 ERA for Toledo while Fister hasn’t even pitched in the regular season for the Nationals.

    -9 Vote -1 Vote +1

    • ralph says:

      Or maybe Dave’s point is that small sample sizes are only cool to use when they support one’s Narrative of choice?

      I don’t know, help me out here.

      -10 Vote -1 Vote +1

    • Dave Cameron says:

      This idea that small sample data should be completely ignored until it reaches a threshold at which it is no longer considered small sample data is ridiculously stupid. Refusal to accept the most recent information because it is an incomplete picture is no better than drawing firm conclusions based only on that information.

      No one is saying that the first three weeks of the season proves that a trade is good or bad, but it absolutely does improve the odds of this deal working out in the Tigers favor. Incorporating the most recent data into your future forecasts is the only reasonable thing to do. Ignoring the most recent data and anointing yourself the Sample Size Police only ensures that you learn nothing.

      +10 Vote -1 Vote +1

      • mike says:

        It’s not bad a point in general, but it’s quite weak in this context. You got caught in a little bit of hypocrisy. A bigger man quickly acknowledges it and moves on. That you chose to try and rationalize it into a larger point mostly just shows you are not (in this case) that bigger man. JMO.

        -12 Vote -1 Vote +1

        • Costanza says:

          Not necessarily. Since we’re looking at cumulative contribution metric (WAR) to determine who “wins” the deal, the +1 WAR digs a little deeper hole for Fielder to climb out of in order to equal Kinsler’s value.

          IE, (pretending their salary was equal to simplify), Fielder would now have to be 1 WAR better than Kinsler in the future in order for the deal to even out. ZiPS projects Fielder for 2.1 WAR RoS, Kinsler for 3.5 WAR RoS; Fielder would need to produce 4.5 WAR to equal what Kinsler is projected to do this year.

          Sample size is a major problem when you’re looking prospectively, but less so when looking retrospectively. Look at it this way — the logic isn’t “Kinsler will be better than Fielder moving forward because he has been better in 3 weeks this year,” but rather, “Kinsler has provided 1 WAR more value than Fielder, so Fielder has more ground the make up than he had before the season.”

          +18 Vote -1 Vote +1

        • jruby says:

          Umm… Small sample size is a huge caveat for rate stats, but for counting stats, you can’t ignore the fact that accumulation is set in stone. Dave would be crazy if he said something like, Kinsler will outperform Fielder by 1 win every 20 games. But he didn’t say that; clearly his point was “based on salaries, Fielder was going to have to outperform Kinsler to make this a break-even for the Rangers. Now, over 5/6 of the season, Fielder’s going to have to outperform Kinsler by *an extra win* over what we thought he’d have to do.”

          And Dave’s absolutely right about that. The second paragraph of his response can be reframed as “we can’t conclude anything yet, but the error bars are shrinking, and doing so swiftly in the Tigers’ favor.” Calling out Dave for “not being the bigger man” seems downright absurd to me.

          … JMO.

          +16 Vote -1 Vote +1

    • rusty says:

      There has been some talk around DC about whether the Tigers knew Fister was hurt when they moved him… so, yes, it’s possible they “knew” they were getting back more than they gave up.

      But it’s widely assumed that they could have gotten better offers for Fister from other teams, whereas it’s unlikely they could have gotten much more for Fielder.

      Vote -1 Vote +1

  3. Greg says:

    The low walk rate definitely has something to do with it. Swinging and missing on 0-0 or 0-1 isn’t going to hurt your strikeout rate; swinging and missing on 2-2 will.

    Vote -1 Vote +1

  4. Chris says:

    This is almost certainly just the result of personal bias (and perhaps wishful thinking), but it seems to me that Castellanos rarely swings and misses at the same pitch twice in one at bat.

    He starts off a lot of at-bats swinging and missing on breaking balls low and away, but if the pitcher throws it again he either lays off or puts the bat on the ball. He might also swing and miss again in the same AB on a fastball up, but again, probably not twice. I feel like a lot of his ABs involve two swings and misses, but he then fights back to put the ball in play. Seems like most of his Ks come when he takes a called strike or two and then has to swing.

    Vote -1 Vote +1

    • brett says:

      This seems like a good assessment. I’ve noticed that he swings out of his shoes on early fastballs from time to time as well. I wonder if he’s trying to add power to his game, early in the count. Someone mentioned that the Tigers are trying to get him to pull more baseballs, so maybe that comes into it?

      Vote -1 Vote +1

  5. SucramRenrut says:

    He seems like a thoughtful, professional hitter in interviews I’ve seen, so I expect he either changes his approach significantly with 2 strikes, or learns/adjust well in an at bat as the count deepens.

    Vote -1 Vote +1

  6. SucramRenrut says:

    As the Count Deepens is also a day-time baseball themed soap opera I’m pitching to CBS fwiw.

    +16 Vote -1 Vote +1

  7. Cory says:

    While this isn’t sustainable I’d also be willing to bet he’s only going to improve as the season goes on considering he just turned 22. His contact rate will rise and his strikeout rate will probably slightly increase as well to account for regression. Not a big deal.

    Vote -1 Vote +1

  8. rustydude says:

    I’m disappointed. I haven’t seen a “You must hate Detroit” comment.

    Vote -1 Vote +1

  9. Rafael Ferreira says:

    He’s the second in the majors with 37 LD% and have a low GB (20%). At least so far he’s making good contact(when he makes). The BABIP may have a postive regression but he has a inside out swing and tend to shot the ball to the right center field, in Comerica Park the ball will probably die. Like the Chris said, he tends to expand the zone in the first pitch breaking balls away( may worth a look )but after he shorts up his swing and ussualy ajust. This may explain the low contact rate. May be a wishfull thinikig but i think he will have low SO% bases on the eye test so far.

    Vote -1 Vote +1

  10. MLB Rainmaker says:

    Two interesting and not-terribly-controversial articles from Dave within a week…I’m impressed. I was convinced you only published to get comments.

    Anyway, just an unfounded guess as to what’s going on here, but I’d bet this has something to do with two young, eager, overly-aggressive guys swinging at everything near the zone. Their first time through the league their free-swinging tendencies hasn’t been scouted — pitchers are throwing pitches around the plate and their innate contact skills have allowed them to survive so far. Going forward, I’d guess teams are going to start expanding the zone on these guys and that’s what will drive the regression.

    I just don’t buy that either of these guys get magical better when there are two strikes on them.

    Vote -1 Vote +1

  11. Peter 2 says:

    I suspected that the reason you see a guy like Nick Castellanos having the highest residual in the regression graph has less to do with him personally than the fact that among qualified players he actually has comparatively very few plate appearances. So both his contact rate and K% are fairly noisy measures at this point. Anyway, I looked it up and he is actually dead last in plate appearances among those reaching the “qualified” threshold. Lumping those with 110 plate appearances together with those with 60 plate appearances in this analysis, and drawing conclusions based on a selection of who has the biggest residual, is basically inappropriate in this case because it has much more to do with within-group difference in plate appearances than anything else.

    Basically just a reminder of what you likely already know but still should be careful of: The leaders in any particular rate stat are almost always very likely to be those with the fewest plate appearances or innings pitched, as a trivial property of statistics.

    Vote -1 Vote +1

    • Peter 2 says:

      Which is more likely to regress—contact rate or K%—is only a function of the intrinsic stability of the statistic, but also how deviant from the mean the player is on either stat. Contact rate may be more stable, but he currently makes the 7th worst contact in the league (among qualified players), thus the probability of his contact rate going up in the future is much more likely than it going further down. His K% is better than league average but he is far from being a league leader. Based on all this, and the small sample size, predicting that Castellanos will wind up with roughly league average contact rate and roughly league average K% is a reasonable, albeit boring, bet.

      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>