Two weekends ago, my Nationals lost a tough series to the New York Mets, largely due to clutch late-game hitting from the Mets and some poor relief pitching by the Nationals. What they did do right, however, was grind out tough at-bats against both Noah Syndergaard and Jacob DeGrom, two of the best starting pitchers in all of baseball. They gave DeGrom a hard time in particular, forcing him to throw 101 pitches over just 5 innings in a game the Nationals ended up winning 7-4.
The Nats' performance prompted a short text thread between my father and I - if only the Nationals always focused on working the count like that! They’d be unstoppable! It’d be like those crazy Red Sox - Yankees series!
With that cloud of thoughts lingering in my mind, I decided it might be fun to dig into the statistics and see if the numbers might back up what we’ve always seemed to feel - that teams working deep counts tend to have the best offenses!
Brainstorming Influential Factors
- Seeing more pitches during a plate appearance increases the likelihood that a batter sees “their pitch” and that a pitcher might make a mistake pitch for the batter to crush
- Seeing more pitches also helps acclimatize batters to the delivery and pitch repertoire of the pitcher, aiding familiarity with future pitches and likely improving hitting productivity
- Forcing pitchers to work deeper into counts accelerates the pace at which a starter builds their pitch count and can force relievers into the game earlier than expected
- Working deeper counts increases the likelihood that the batter will draw a walk, placing a runner on base and denying the pitcher an out in exchange for all those pitches
- Working deeper counts also can make it more likely for the batter to strike out, if they find themselves gravitating to two-strike counts where the pitcher just needs to make one pitch
One thing to note as we think about the Pitches Seen -> Offense relationship is that there is likely an interaction effect between hitting quality and pitches per plate appearance which makes evaluating either independently a bit tricky - let me try to sketch it out here:
- The best hitters may work deep counts more often to get to the pitches that they want to hit
- Pitchers are likely to be cautious pitching to the best pitchers, because they know that a mistake pitch is going to be punished
- Both of these factors lead to the inflation of pitchers per plate appearance when pitching to the best hitters
- When good teams have lineups featuring many strong hitters, this effect is particularly pronounced
- Therefore, when evaluating those teams, with many good hitters and also a relatively high number of pitches seen per plate appearance, it is tough to tell exactly what is driving the pitchers per plate appearance metric up - innate plate discipline? Or a residual effect of pitchers treading carefully around good hitters? Tough to tell…
Pitches per Plate Appearance by Team
One thing we can do right off the bat is trace these “deep count” trends by team over the last five seasons. As mentioned earlier, it turns out that the Yankees and Red Sox do indeed tend to work deep into the count, though the number of pitches seen per plate appearance by the Red Sox has dipped in recent season while the Yankees has increased notably.
Real-Life Examples: The Yankees and Marlins
While the Yankees have been doing some absurd things at the plate over the last few seasons, setting a number of MLB records for home runs, the Marlins have really struggled on offense, and currently have the second-lowest number of runs scored in all of baseball. With that in mind, it’s interesting to see where each team shakes out when we plot offensive output (as measured by Weighted Runs Created Plus) against Pitches per PA…
Statistical Trends - Scatterplots by Metric
To see how other statistics might support or contradict our narrative, I threw together a number of scatterplots to help visualize connections and distributions between variables. Each dot represents a “team season” over the last five years, and the trendline and confidence interval provide some intuition into how the each metric changes in response to increases in Pitches per PA.
One thing about the final two charts which I think is interesting is that they jive very closely with the rise in “Three True Outcomes” in baseball (which I’ve discussed previously). Together, they suggest that hitters who see more pitches are slightly less likely to hit the ball when they swing, but significantly more likely to hit for extra bases when they do - supporting the rise in overall wRC+ we see as Pitches per PA increases, even with a slightly increased strikeout rate and lowered contact rate.
Hitting Metric Correlation Matrix