Throughout the season, I charted countless minor-league pitchers, noting pitch data when available (velocity, spin rate, and movement) and pitch outcomes. In doing so, one pitch outcome specifically drew my attention: called strikes.
Yet, clearly, not all called strikes are created equal.
Pitchers often earn called strikes in unfavorable counts. A called strike in a 3-0 count, when a batter is likely to take, is not worth the same as a called strike in a 0-2 count. Indeed, a called strike in a 3-0 count is actually the most common outcome. The table below sorts called strikes per total pitches for each count for all pitchers in the MLB in 2021:
Table 1: Total Called Strikes in MLB in 2021
|Count||Total Pitches||Called Strikes||Total Swings||Total
Unsurprisingly, a called strike on 3-0 is exceedingly common (55.3 percent), while a called strike on 0-2 is extremely rare (3.9 percent). These types of called strikes should not be weighted the same.
As such, and understandably, called strikes alone have a weak correlation with most important pitching metrics. Given these observations, I hypothesized that weighting called strikes by rarity per count would improve the correlation between called strikes and important pitching metrics, such as strikeout rate.
First, however, I created an adjusted called-strike rate based strictly on non-swing results. Evaluating called strikes rate as a quotient of called strikes and total pitches involves too many other variables, such as, well, swings. Most pitchers with high swinging-strike rates indeed have low called-strike rates. Therefore, as an initial matter, I adjusted the called strikes rate as follows:
aCS% = CS ÷ (TP – TSw)
- aCS% = Adjusted Called Strikes Rate
- CS = Called Strikes
- TP = Total Pitches
- TSw = Total Swings
To test this metric, I pulled count-by-count pitch-result data for all MLB pitchers. In addition, I grabbed various rate statistics for each pitcher, including strikeout rate, whiff rate, deserved-run average (DRA), fielding-independent pitching (FIP), and weighted on-base average (wOBA). The sample excluded pitchers with less than 300 total pitches.
This article only examines the descriptive quality of these metrics—that is, the correlation between two metrics in the same season. How does Adjusted Called-Strikes Rate compare to called-strikes rate?
Table 2: R-Squared Between CS%, aCS%, and Pitching Metrics
Adjusted Called-Strikes Rate has a stronger correlation to the selected pitching metrics (strikeout rate, DRA, FIP, and wOBA) than called strikes rate. Of course, the correlation between aCS% and the selected pitching metrics remains weak.
The next step is to test my initial observation: not all called strikes are created equal. As such, I developed, with assistance from Jordan Rosenblum and the BP Stats Team, Adjusted Called Strikes Plus (aCS+):
aCS+ = 100 × ∑ ((aCS%c × (MLBTP%c – MLBTSw%c)) ÷ MLBaCS%c)
aCS% = Adjusted Called-Strikes Rate
MLBTP% = MLB Total-Pitch Percentage
MLBTSw% = MLB Total-Swing Percentage
MLBaCS% = MLB Adjusted Called-Strikes Percentage
c = count
(The league-average aCS+ is 100.)
How does Adjusted Called-Strikes Plus correlate to these selected pitching metrics?
Table 3: R-Squared Between aCS%, aCS+, SwStr%, Whiff%, and Pitching Metrics
Adjusted Called-Strikes Plus improves even further over Adjusted Called-Strikes Rate, with a stronger correlation to the selected pitching metrics. Nevertheless, the correlation between aCS+ and the selected pitching metrics is still weak. Called strikes alone simply lack much descriptive, and likely predictive, value.
Now, what exactly should Adjusted Called-Strikes Plus tell us? Pitchers with better stuff, deception, pitchability, and/or command should generate more quality called strikes. (A big, knee-buckling breaking ball comes to mind.) Further, pitchers with better control should generate more early-count called strikes and avoid padding their called-strike totals with called strikes in unfavorable counts. On the other hand, pitchers with poor stuff or control likely will generate less quality called strikes and more standard called strikes, such as those in a 3-0 count. While these considerations can work against each other for many pitchers, ultimately quality pitches—particularly, breaking balls—should win the day.
So, which pitchers benefit the most from Adjusted Called-Strikes Plus?
Table 4: Top-20 Starting Pitchers with Highest aCS+*
*This table only includes starting pitchers with 1,000 or more total pitches.
Notably, this table includes some of the top pitchers in Major League Baseball, such as Julio Urías, Gerrit Cole, Brandon Woodruff, Corbin Burnes, Walker Buehler, and Jacob deGrom. (We’ll pretend that we don’t see Dylan Bundy at the top.) In fact, most of the pitchers on this table have electric stuff—the type of stuff that can freeze a batter.
Table 5: Top-20 Starting Pitchers with Lowest aCS+*
*This table only includes starting pitchers with 1000 or more total pitches.
Most of these pitchers either have mediocre stuff (i.e., Matt Shoemaker, Josh Fleming), poor command (i.e., Wily Peralta, Carlos Hernández), or some combination of the two. In many cases, pitchers on this list lack an impact breaking ball (i.e., Jake Odorizzi, Adrian Houser).
The next inquiry is whether aCS% and aCS+ improves Called Strikes Plus Whiffs rate (CSW%), a metric developed by Alex Fast, among others, at Pitcher List. As a general matter, these formulas are not strictly comparable. The following formulas do not have a common denominator. Instead, I sought to normalize the rates based upon swing rate.
For aCSW%, I multiplied the Adjusted Called-Strikes Rate by the pitcher’s swing rate and the whiffs rate by the corresponding non-swing rate, and I summed the figures. For aCSW+, I similarly weighted whiffs (actually reducing its descriptive value alone across all pitching metrics), then multiplied aCS+ by the league-average swing rate and weighted whiffs by the corresponding league-average non-swing rate.
CSW+ = 100 × ∑ (CSW%c × MLBTP%c) ÷ MLBCSW%c
aCSW% = aCS% × Swing% + Whiff% × (1 – Swing%)
aCSW+ = aCS+ × MLBSwing% + W+ × (1 – MLBSwing%)
This metric should more accurately measure how often a pitcher obtains a strike without contact.
Table 6: R-Squared Between CSW%, CSW+, aCSW%, aCSW+, and Pitching Metrics
The correlation between Adjusted Called Strikes Plus Whiffs rate (aCSW%) and the selected pitching metrics is consistently stronger than CSW%. Furthermore, Adjusted Called Strikes Plus Whiffs Plus (aCSW+) has the strongest correlation with strikeout rate. However, Called Strikes Plus Whiffs Plus (CSW+) has even stronger correlation to DRA, FIP, and wOBA. This makes sense. Refining a statistic, such as called strikes, based on quality should provide better descriptive value.
Here is the aCSW+ leaderboard:
Table 7: Top-20 Starting Pitchers with Highest aCSW+*
|McCullers Jr., Lance||2788||113|
*This table only includes starting pitchers with 1,000 or more total pitches.
Now, that is a star-studded list!
Ultimately, called strikes alone lack much descriptive, and likely predictive, value with important pitching rate metrics, though there is some correlation. Even weighting called strikes based on count does not fully resolve those issues. That said, aCS% is a simple alteration to called-strikes rate and substantially improves the descriptive value of the metric with the selected pitching metrics. Further, aCS+ consistently shows even stronger correlation to the selected pitching metrics than called-strikes rate. Finally, aCSW% improves upon CSW%, while retaining similar notational elegance. Meanwhile, CSW+ provides superior descriptive value to CSW%, aCSW%, and aCSW+ for DRA, FIP, and wOBA, while aCSW+ provides superior descriptive value to both CSW%, aCSW%, and CSW+ for strikeout rate.
Some fun closing statistics observed in this process:
- Rich Hill led the league with a staggering 22 called strikes in 0-2 counts. That curveball!
- Meanwhile, Julio Urías is adept at getting ahead in the count, leading the league in 0-2 counts and leading all starting pitchers in 0-2 pitches per total pitches.
- Lance McCullers, Jr., amassed 30 called strikes on 3-0 counts, five more than the next highest. Control (and arm health) remain concerns.
- Bailey Falter had zero pitches in a 3-0 count over 552 pitches!
- Like Falter, Richard Bleier simply avoids unfavorable counts, with just three 3-1 counts and one 3-0 count over 685 total pitches.
Table 8: Sortable Pitch Data for All Pitchers with 300-Plus Pitches
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.Subscribe now