Do Clutch Tendencies Exist? Breaking Down Hitters’ Approach in High-Leverage Situations

Written by Ari Goldberg

Baseball fans love to discuss whether certain hitters are “clutch” or “big moment players,” claiming that they thrive off hype moments and aura. But do these players truly elevate their performance when it matters most or do the clutch hitters just perform at a high level all the time? Do elite hitters succeed in high-leverage situations because of a strong mental edge, or are they actively adjusting their plate approach to perform better when it matters most? In simpler terms, are there elite hitters or elite tendencies?

Exploring Plate Discipline Variables

In my project, the variables I use to represent plate discipline are the percentage of first pitches are swung at (FP%), the percentage of 2 strike at pitches that are swung at (2-strike %), percentage of pitches in the zone that are swung at (Z-swing%), percentage of pitches outside the zone that are swung at (O-swing%), and a simple percentage of pitches with no swing (Take %). These variables are able to help describe the approach a player takes during their at-bats on a spectrum ranging from a batter who is very aggressive, swinging at everything, to patient, who takes more pitches and swings less.

Data Set Up

To evaluate hitter performance across different game contexts, I segmented the data into four situational categories: (1) all at-bats, (2) two-out at-bats, (3) late-inning at-bats (defined as inning 5 or later in the Kernels Collegiate League and inning 7 or later in the Prospect League), and (4) at-bats with runners on base.

Each pitch was grouped by batter, creating a dictionary where the key was the batter’s name and the value was a DataFrame containing all pitches faced by that player. From these pitch-level datasets, I computed a series of approach and outcome metrics for each hitter, including:

• xwOBA
• Strikeout rate (K%)
• Walk rate (BB%)
• Two-strike rate (2-strike%)
• First-pitch swing rate (FP%)
• Zone contact rate (Z-Contact%)
• Zone swing rate (Z-Swing%)
• Chase rate (O-Swing%)
• Take rate
• Hard hit rate (percentage of balls hit ≥ 90 mph)
• Δ Run Expectancy (change in expected runs across a play)
  

Although not all of these variables were used in every model, they allowed for flexible testing across multiple approaches.

To isolate situational changes, I calculated the difference in each metric between the situational subsets (2 outs, runners on, late innings) and the baseline (all other pitches) on a per-batter basis. I then applied regression models to these within-player differences to determine whether consistent patterns emerged across the full sample — specifically, whether certain approach adjustments or traits were statistically linked to better performance in high-leverage contexts.

Expected Weighted On-base Average

The first dependent variable that I wanted to look at was xwOBA. With 2 outs, almost all of the tested variables were statistically significant. Increasing FP%, Z-swing%, O-swing%, or Take % would lead to a lower xwOBA. Swinging at any pitch, whether in the zone or not, on the first pitch or not, would lead to a player getting on base less.

In my regression for later innings, multiple variables came back statistically significant. First, the intercept was -0.0243, meaning that xwOBA for batters drops by 2% when all the other variables remain the same. Swinging more at 2-strike pitches led to decreased xwOBA, while taking more pitches led to an increased xwOBA.

In situations with runners on base, none of the variables were statistically significant, suggesting that success in these situations depends more on other factors, like randomness, than on raw approach metrics alone.

We see that with runners on, there is little to no indication that changing your approach affects your on-base percentage, in later innings, xwOBA actually tends to decrease regardless of approach, and with 2 outs, increasing your swing rate will lead to a batter getting on base less. In late innings, the clearest effect on xwOBA is that taking more pitches boosts xwOBA, and swinging with two strikes decreases it. One potential explanation is that when relievers come in for those later innings, the control and stuff of their pitches tends to be worse than starters. Therefore, swinging at less pitches in 2 strike counts could help you draw a walk instead of swinging at a strikeout. Also, the fact that taking pitches regardless of where they are leads to better outcomes could lend itself to the combined effects of poor location and bad contact quality by the majority of batters. If the pitches are more likely to be balls than strikes, and a swing is not likely to produce a base hit, it would be advisable to take more pitches and hope for a walk or a HBP. In the 2-out scenarios where swinging leads to less players getting on base, we likely see the same explanation from earlier. Pitches not being able to consistently find the zone, especially when they’re wary of giving up a 2-out hit, leads to players ending up with easy 2 out walks, which would raise xwOBA.

Hard Hit Ball %

Since I looked at outcomes first with on-base percentage, I next wanted to see if there was any indication of hitting quality changing in these situations. In the MLB, a hard hit ball is measured as a batted ball event with an exit velocity over 95 MPH. Because the KCL and Prospect League do not hit the ball as hard as the MLB, I decided to lower the range for hard hit balls to 90 MPH.

In all 3 of the different situations that I looked at, none of the variables came back statistically significant in the regressions. However, an interesting point is that in both the late innings and runners on situations, the intercept in the regression model was statistically significant, with p-values of 0.007 and 0.024, respectively, well below the 0.05 threshold.

In late innings, HH% dropped by 4.3% and with runners on base, HH% dropped by nearly 8%. If you hold every other variable constant, i.e assuming batters maintain a consistent approach across scenarios, the quality of their hits decreases in the two situations. As the graph illustrates, most players experienced a decline in Hard Hit % when runners were on base compared to their overall performance. However, a handful of players showed notable increases, suggesting that some may elevate their quality of contact in higher-pressure situations. That said, there were no consistent trends among these batters that I was able to identify which might explain the improvements.

The fact that there were no changes to approach that led to an increase or decrease in HH% does not leave us without any conclusions. The simple answer is that the players who hit the ball harder do it consistently regardless of the situation, while the players who don’t hit the ball hard are also consistent in not hitting it hard. However, since the intercepts in late innings and with runners on were significant and negative, suggesting that hitters in the KCL and CornBelters generate lower-quality contact in these situations. One logical conclusion for the much larger decrease when runners are on is that batters who make it on base more frequently usually have more hard hit balls, and these hitters may not be the ones hitting in key moments after reaching base, skewing the data so that batters who usually don’t hit as hard are overrepresented in this sample. The other explanation is much simpler, clutch performance was not consistently observed in the data when the situations get tense.

Delta Run Expectancy

Since I have now looked at simple batting stats and the quality of BBE, the next category of variable I wanted to test was an advanced analytics stat, Delta Run Expectancy, the most direct way of seeing if a player can change the outcome of a game.

Delta run expectancy measures the change in a team’s expected runs scored from the start to the end of a play. It reflects the value a play adds or subtracts in terms of future scoring potential, based on the shift in base-out state.

First, in 2-out scenarios, we see that swinging in a 2 strike count actually increases the delta run expectancy, while swinging at pitches in the zone leads to less expected runs. We also see that taking pitches raises the expected runs a team can expect to score.

In late innings, we see that the run expectancy actually drops by 0.0082 on average for each pitch. We also see that swinging at first pitches more often leads to less expected runs. Swinging at pitches outside the zone is also shown to lead towards lower expected runs. Finally, we see that taking pitches in general in the later innings increases the number of runs expected.

Finally, with runners on the bases, we see that swinging at the first pitch and swinging at balls in the zone lead to less expected runs, but none of the other variables had any significant results.

Only in 2 outs scenarios does swinging with a 2 strike count actually lead to better outcomes, for one of two reasons. It might be because pitchers have already expended a lot of energy and concentration to get to that point, and therefore that 2 strike pitch is more likely to be hittable for the batters. Another factor is that batters being more aggressive in 2 strike counts to fight off a strikeout actually leads to them getting more swings around on the ball, and eventually, after working the count and staying alive with foul balls, they put one into play that gets down for a base hit. In later innings, taking pitches leads to more expected runs, likely because as a starter begins to tire and as relievers come in, pitch quality tends to decrease, leading to more walks which then translates to more expected runs. We see differing results for first pitch swings between later innings and with runners on, with higher FP% leading to less runs in late innings and more runs with runners on. This could again be due to relievers coming in and offering more hittable first pitches allowing batters to take advantage of a higher FP%. However, with runners on, pitchers might be more hesitant to throw a hittable first pitch and therefore swinging at the first pitch will either lead to swinging strikes, lowering runs expected, or low quality hits that could result in more double plays and outs.

Conclusion

This analysis set out to investigate whether hitters in the Kernels Collegiate League and Prospect League adjust their plate approach in high-pressure situations, and if so, whether those adjustments contribute meaningfully to improved outcomes. While some statistically significant patterns did emerge, particularly around taking more pitches in late innings and swinging with two strikes in two-out situations, the overall picture is more nuanced.

Rather than identifying sweeping changes in behavior that consistently lead to better results, the data reveals a more subtle truth: the most effective hitters tend to be those who are simply good, regardless of context. Their performance in high-leverage scenarios appears less about transforming their approach and more about maintaining consistency. Situational changes, when they occur, often yield modest benefits and may be more about exploiting specific pitcher tendencies, such as fatigue or poor control, than about the batter radically changing strategy.

Importantly, we saw that metrics like hard hit rate and delta run expectancy tended to decline in pressure-filled moments, even when approach metrics stayed constant. This suggests that it’s not just about what hitters choose to do in these situations, it’s also about the quality of pitches they face and the difficulty of execution under pressure. While certain traits like patience and selectivity offer value in late innings, these findings reaffirm that a hitter’s underlying skill level remains the strongest predictor of success.

Ultimately, this study contributes to the ongoing conversation around clutch performance. While it’s tempting to believe in heroic transformations during big moments, the data implies that consistency, rather than clutch, is the defining trait of high-level performers. That doesn’t mean situational awareness is irrelevant, but it highlights that the best way to succeed under pressure is often to stay true to what already works.

Limitations

While there were over 17,000 pitches in total throughout the season, we have 5,000 total pitches with 2 outs, 6,000 total pitches in the later innings, and only 1,500 pitches with runners on, limiting the statistical power in those segments of the analysis. Also, since we are looking at each batter’s individual differences, some of the batters will have much less pitches in certain scenarios, just out of the pure randomness of the batting order, leading to potential misrepresentations in the data. While using Yakkertech to track pitch data throughout the season, exit velocity was sometimes not picked up by the cameras, especially later in games once the sun went down, so there will be much less pitches incorporated into that section of the analysis.

I’m incredibly grateful to Jarrett Rodgers for the opportunity to work with the CornBelters this summer; it was a truly unforgettable experience that deepened my understanding of the game and the world of baseball analytics. I’d also like to give a shoutout to my fellow analytics interns, Cam, Charlie, Max, Noah, and Wyatt, for their insight, collaboration, and support throughout the season. It was a privilege to learn and grow alongside such a dedicated group of future MLB baseball analysts ;).