Bulletproofing Your Arsenal

There are a myriad of strategic intricacies throughout any baseball game that branch off a singular theory. For example, the recent decay in stolen bases is largely tied to better event valuations. Organizations now recognize and apply the real value of each play or strategy. It’s the same philosophy that virtually murdered the bunt. When concepts only marginally enhance a team’s chances of scoring or even diminish them they’re gradually omitted from the third base coach’s signs. That’s Run Expectancy 101. 

There’s a lot of existing information and analysis on pitch by pitch strategy and the semantics of each pitch in a plate appearance. The publicity of metrics and concepts which induce these trends varies. Some aren’t quite as well documented. Some are evident. For example, throwing fastballs up in the strike zone is effective as it plays into modern swing planes. Throwing a curveball while ahead in the count with two strikes forces the batter to take a chance, potentially making weak contact. These (well-documented) examples are indicative of another element of these ‘umbrella’ concepts: their diversity. Some react to modern analytic principles or concepts and some are logic-based. As an aside, these logically derived, often traditional tactics are most often pummeled by baseball analytics in an attempt to debunk the convention, ultimately testing its effectiveness and relevance. Some survive (staying alive, battling in a plate appearance), some don’t (bunting) and some are adapted and developed (high fastballs now an effective putaway pitch). 

Getting back to the ebb of plate appearances, one factor which might be appealing to examine is a pitcher’s consistency. When reflecting on the length of plate appearances, the evolution and shifting of leverage, it is somewhat well understood that batters ordinarily find themselves more successful than they contrarily would later in plate appearances rather at the start. It’s not fair or analytically sound to assume that this is a direct relationship. This may be partially a result of better hitters being more capable and thus more likely to prolong a plate appearance. This would raise the mean batting average, or any measure, when viewing plate appearances with high, potentially double-digit pitch counts. Alternatively, from the pitcher’s perspective, he may tire as the duel continues. Regardless of the impact either of those factors has, the conclusion most reach is a batter becomes more acquainted with his opponent as the plate appearance progresses and thus attains an advantage. There’s a lot of evidence for this effect. Extensive (and expensive) video rooms, the third time through the order tax, and on-deck batters’ urge and interest in getting an illustrative look at a pitcher all establish a nice basket of evidence. Batters like to have familiarity with the pitcher they will face. Even if all the aforementioned evidence is flawed, few will dispute this reality. 

However, this all makes one assumption: That the pitcher is consistent. That’s a fair assumption to make. You don’t get to the highest level of professional baseball by doing something different every day. Pitchers are creatures of habit, finely tuned machines, as are all ballplayers. While pitchers make adjustments, generally, they’re continual or exclusive to one pitch. Consider pitchers who have implemented some hesitation, pause, or delay, a near balk motion seemingly scattered randomly throughout their appearances. Notably, Roberto Osuna and Marcus Stroman call on the maneuver regularly. Others deploy it on occasion as well. Having and effectively using such a tool is naturally a benefit to a pitcher. It disrupts a batter’s rhythm. Some may contend that the only reason it is powerful is that it is so seldom exposed. True or not, any small discrepancy in baseball’s ‘normality’ would alter a batter’s aptitude for making solid contact. 

While Baseball-Reference doesn’t (yet) list whether a batter faced delivery hesitation or not on their splits pages the strategy is paired with a grander concept, as discussed earlier. In essence, any value derived from hesitation is value derived from an irregular approach. Thus, the impact can be estimated in several other ways. For example, what if pitchers could throw various curveballs, accurately controlling the amount of break they get? In essence, some can. That’s just effective pitch tunnelling. A concept that creates uncertainty for batters the same way delivery hesitation does.

A pitch’s effectiveness or how strenuous it is to hit is measured in several ways, but many indicators are in some way associated with velocity and spin. Returning to the previous example where pitchers are creating deception by tunnelling their pitches, or ‘controlling the pitch’s amount and direction of break’ we know that pitchers tailor their spin to ensure they remain deceptive. What about creating deception with velocity? A batter is fooled by a pitch’s velocity when they are sitting fastball and get a hook, or vice versa. What if a batter could guess the pitch type correctly, but still be fooled by the pitch’s speed? 

Analyzing 705 individual authors’ 2231 different pitches (thrown at least 50 times) offered some insight into how using velocity to create deception may be effective. First, it’s important to bear in mind that measuring the standard deviations of each pitch’s velocity, similar to measuring spin rate, is imperfect. For example, consider Steven Wright’s knuckleball. Its spin wasn’t even measured by Stat Cast in 2019. Before that, in 2018 it produced a rate of 1441 which appears high when you watch Steven Wright, but that’s beside the point. That season, 2018, Wright put up a respectable .317 xwOBA across 719 trials with the knuckleball; 197 plate appearances resulting in a .220 batting average. This is merely a demonstration that spin rate does not equate to effectiveness. In essence, spin is most effective at its tails. Another component in the lack of a direct linear relationship is velocity. More velocity creates more natural spin. Thus, it’s difficult to compare pitchers’ spin rates or deem a certain threshold ‘elite’. This is why Bauer units, or effective spin, was created. It accounts for the pitch’s velocity by dividing its spin by its speed. This produces a number closer to the amount of spin which influences movement, the share of spin that matters. This number is closer correlated to productivity than raw spin. 

This is all to contextualize the standard deviation of pitch velocity. The plot shows that while the correlation is weak, pitchers whose velocity is more sporadic do generate lower xwOBAs. The cause for the weak correlation is likely justified the same way spin’s weak correlation is. Spin (and Bauer Units) are very raw, almost like batting average. Spin is effective at its tails, paired with the right velocity, in the right parts of the zone, etcetera. Spin rate does not encapsulate that. The same likely goes for varying your velocity – being deceptive and inconsistent with velocity. The effectiveness of varying a pitch’s velocity is likely dependent on when and how the pitch is used. When it is thrown at what speed. If deployed correctly, it presumably offers the same benefits of pitch tunnelling. Just like spin, where it was once thought that more is always better, perhaps pitchers are working toward an era where high velocity’s importance will pale in contrast to timely and varied velocity.