Revisiting whether clutch is a skill

A few days ago, there was an interesting conversation brought into my Twitter feed, all spurring from this Tweet by known Yankees Twitterer Max Wildstein.

It’s been an interesting argument that has been one of the main topics that divides old and new school baseball. Both sides have fair points, the source of those points mostly differentiating along the lines of experience and evidence. Pressure is a real psychological effect. Everyone has experienced it and has felt it to some degree. The questions are a.) how much does it actually impact the performance of a hitter and b.) can it be used in evaluation?

Luckily, there’s data for everything and splits for almost every situation, including leverage, which is determined by the inning, base runner(s), and score. High leverage is what you would consider a pressure situation, while conventional wisdom would suggest low and medium leverage situations as the opposite. To gauge the split, here are the top qualified hitters in high leverage situations since 2002. To what should the surprise of nobody, it’s basically a list of ten of the best hitters of recent memory.

Then for non-high leverage situations, it’s the same story. Just a bunch of good hitters.

Again, none of this should be a surprise. In basically every critical situation, you should always prefer to have the best possible hitter up at the plate. Preferring someone that’s perceived like... David Freese... or Eric Hosmer would be an extremely flawed process.

But let’s make this all relative for a second. Who has performed the best in pressure situations relative to their performance in non-pressure situations. Someone who would be perceived as a “clutch” hitter. Examining the differences between high and low leverage offensive performance would at least be a step in the right direction for this form of analysis. Here are the top 10 differences from 2019, showing hitters who have a large portion of their offensive production in high-leverage situations.

Looking at the opposite side, here are hitters who have had a large portion of their offensive production in low leverage situations, most of whom haven’t performed well in high leverage situations.

Now that we’ve assessed most of the baselines, we can now get to examining perhaps the biggest question. Is being clutch a repeatable skill? To help eliminate actual hitting skills, we’ll be examining the difference in offensive production between low and high leverage situations and how that difference correlates year-to-year.

For the sample sizes, we’ll be looking at every qualified hitter from 2014 to 2019, giving us a sample size of 434. First, lets look at how consistently hitters distribute their offensive performance through leverage situations. Starting with the year-to-year correlation of high-leverage wRC+.

FanGraphs

And... nothing.

Some single examples that can illustrate include 2014-2015 Josh Harrison (207 wRC+ year one, 28 wRC+ year two), 2014-2015 David Ortiz (201 wRC+ year one, 60 wRC+ year two), 2017-2018 Cody Bellinger (187 wRC+ year one, 46 wRC+ year two), and 2018-2019 J.D. Martinez (231 wRC+ year one, 19 wRC+ year two). There is zero consistency in high leverage performance. Considering high leverage situations make up such a small sample of total plate appearances (less than 10 percent on league-average), this makes it hard for any player to build up a consistent basis for performance. Combine that with the fact that almost every player can’t put up a consistent performance in the clutch and it shows that this high leverage stats are pretty much useless in evaluation.

As for performing in non-high leverage situations, it’s a different story. Why? Well because it accounts for roughly 90 percent of a player’s season and as we’ve seen, most hitters tend to hit near the same level they did the previous year. There is a decent correlation for year-to-year offensive performance in non-clutch situations (R = 0.52).

FanGraphs

Now as I mentioned above, to help make everything relative, we’ll examine the year-to-year correlation for difference in offensive performance between non-high leverage and high leverage situations.

There you have it! No correlation, meaning being “clutch” isn’t a repeatable skill. High leverage performance could not be any more inconsistent than it already is. To summarize, it’s look to be as clutch is not a skill, as in general it is not repeatable. It’s a random distribution of offensive production that messes with the perception of a player’s performance. It can be a fun stat to look at, sure, but in no way should it be used in evaluation of performance.

Patrick Brennan loves to research pitchers and minor leaguers with data. You can find additional work of his at Royals Review and Royals Farm Report. You can also find him on Twitter @paintingcorner.