Searching For Starlin Castro's Other Half

EDITOR'S NOTE: Many of you remember Naveen Nallappa's statistically-oriented articles here. While Naveen was unable to continue writing them due to work commitments, his brother Rahul is just as knowledgeable about advanced metrics. So I've brought him on to write occasional articles of this type. Please welcome him, and his first article is about a player who's been quite a hot-button topic this season.

In recent years, we have often wondered whether Starlin Castro has all of his mental energy focused towards the game, or whether half of his mind is in another place. This piece is not about Castro's mental lapses, as they have clearly been put on the back burner. Rather, it is to contemplate whether Castro has experienced an epic degeneration in his ability to hit.

As it stands, Castro's season wRC+ is a meager 51, ahead of only Royals second baseman Omar Infante. In other words, Castro has provided offense that is 49% worse than league average. Upon realizing that Castro put up a 115 wRC+ last season, one may find the title of this piece shockingly generous.

Cursed by the BABIP Gods?

When analyzing the performance of a player on the basis of about half a season, it's necessary to identify how big a role luck has played. For hitters, this typically comes down to their Batting Average on Balls in Play (BABIP).

Castro's BABIP for the ‘15 season has been by far the lowest of his career. At .274, it is .45 points lower than his career BABIP. But this does not necessarily mean he has been getting unlucky, as the quality of contact he has been making has been exceptionally poor.

Using the most recent formula to calculate a hitter's expected BABIP based off of their batted ball profile, we see that Castro's low BABIP is not a product of bad luck. Using the xBABIP formula developed by Alex Chamberlain of Fangraphs, we see that Castro's xBABIP has dropped from .271 in 2014 to .234 in 2015. (Some of you may think that these BABIPs are too low, and that the formula is biased downwards. While you may be correct, that still does not diminish the results when evaluating Castro's past two seasons.)

Examining this graph, the three most important takeaways are:

1. Castro's HARD% has dropped nearly 25%

2. Castro's LD% has fallen nearly 30%

3. Castro's GB% has increased nearly 25%

Typically, a change in a hitter's HARD% occurs as the player ages, as diminishing strength levels lead to softer contact. However, it is unlikely that Castro would have experienced such decreases in strength levels entering his age 25 season. What is more likely, and more consistent with points 2 and 3, is that Castro has been fooled more frequently by opposing pitchers.

Struggling with the Heater

Fangraphs "Pitch Values" provides us with a metric to evaluate how well a certain batter has performed against a certain pitch. In particular, the metric tallies up changes in the run expectancy after every pitch while factoring in the type of pitch thrown. These seemingly miniscule changes add up over the course of the long season and provide us with a good idea of which pitches a hitter favors and which pitches a hitter hates.

Last season, Castro established himself as an extremely good fastball hitter, producing a wFB of 16.1.In other words, compared to the league average hitter, Castro performed 16.1 runs better when hitting the fastball, good for 18th in the NL. This year, he ranks dead last, and it isn't very close. Castro has a wFB of -17.7, significantly worse the -12.1 posted by runner-up Angel Pagan.

Connecting Out of the Zone

Upon first glance of Castro's PITCHf/x Plate Discipline, there doesn't appear to be too much of a difference between this year and last. However, the two largest differentials are his O-Swing% and O-Contact%. O-Swing% is simply the frequency in which a batter swings at pitches outside the strike zone, and O-Contact% is the frequency in which a batter makes contact when he swings at a pitch thrown out of the strike zone.

For Castro, both of these percentages have risen; his O-Swing% has risen from 30.4% to 34.2%, and his O-Contact% has risen from 62.8% to 69.1%. By multiplying O-Swing% by O-Contact%, we arrive at a valuable metric: the frequency in which a batter swings at pitches out of the zone and makes contact. This is valuable because a hitter is much less likely to do damage on pitches outside of the zone. In Castro's case, the compounded metric has risen from 19.1% to 23.6%.

Though this difference may seem negligible, (i) that's nearly a 25% increase, and (ii) the following graphic expounds on how poor Castro is at hitting pitches out of the zone.

Are Pitchers Approaching Castro Differently?

Pitch Type

Upon examining the distribution of pitches Castro sees, two things jumped out at me:

1. Castro is seeing roughly 24% fewer curveballs and 7% more sliders.

2. Castro is seeing roughly 9% more four-seam fastballs.

It appears that pitchers have picked up on Castro's weaknesses. According to "Pitch Values", apart from 2012, Castro has always struggled heavily with the slider, and has always been roughly league average at hitting curveballs. This season, the discrepancy has been even larger, and as a result, pitchers are substituting sliders in for curveballs.

Castro's recent struggles with the fastball have similarly led pitchers to throw the pitch more often.

Pitch Location

From these visuals displaying the distribution of location of the pitches thrown to Castro, it appears that pitchers are targeting the outside corner with slightly more regularity. As evidenced in the last visual, Castro has not been particularly effective at hitting outside pitches, nor has he been good at laying off pitches off the plate outside.

Conclusion

Castro's dismal performance has not been heavily influenced by bad luck, as there is more than enough evidence suggesting a decline in skill. Despite showing up worse in several indicators, it is likely that everything stems from Castro's inability to hit the fastball. Though this is only speculation, it appears that the same principle that governs pitchers may also apply to hitters. Pitchers are often told that they must establish their fastballs in order to fool a hitter. For offspeed pitches to work at their best, the hitter must have the threat of a heater at the back of his mind. In other words, all offspeed pitches work off the fastball. For hitters, an inability to hit the fastball makes them more susceptible to offspeed pitches, or in Castro's case, the slider. Interestingly enough, Castro has struggled mightily with the slider yet has performed league average against the curveball. Consistent with this theory, the curveball deviates from the plane of the fastball much earlier than the slider, and so Castro recognizes the pitch quicker out of the hand and hence does not have a problem hitting it. Meanwhile, the slider tends to mimic the fastball for a much longer time, and so Castro is more likely to be fooled at the last moment.

Whether this theory is accurate or not, Starlin Castro has his work cut out for him if he is to return back to his All-Star form.

All data, graphs, and tables are courtesy of Fangraphs.