Strike% - Ball% Analysis
I look into the differential between Strike Rate and Ball Rate, evaluating the statistic as a whole and looking at which pitchers stand out.
I got into this in this morning’s Daily Notes, and I wanted to finish the job here.
If you read these pages, you know I like to look at these two statistics when evaluating pitchers.
Strike% = (Fouls + Called Strikes + Whiffs) / Total Pitches
Ball% = (Called Balls) / Total Pitches
These two things correlate highly to K-BB%. That should be obvious, as strikeouts are about strikes, and walks are about balls.
The question I want to answer here is: Is there any use in taking Strike%-Ball% in addition to K%-BB%? If K-BB% works just as well or better, then we don’t need to add on a new stat. Too many statistics can obfuscate. You can miss the forest for the trees, so to speak. I don’t want to talk about Strike%-Ball% if it’s not useful for better predicting the future, so let’s see if it is. I don’t know the answer right now, but I figured it would be fun to publish my findings as I go through them.
My hypothesis is that Strike%-Ball% will be a predictor of K-BB%. It should stabilize more quickly since it’s taken at the per-pitch level rather than the per-PA level. The data sample grows a few times faster, so we should be able to find some signal in it more quickly.
K-BB% vs. Strike%-Ball%
Correlation with ERA
Taking 2023-2024 data and filtering to pitchers with at least 25 starts.
Correlation Coefficients
K-BB% vs. ERA: -0.577
Strike-Ball% vs. ERA: -0.384
Scatter plots:
So, we immediately see that K%-BB% is more strongly correlated to ERA. This is no surprise. A strikeout means an out was recorded, while a strike does not necessarily mean that. A walk means a man was put on base; a ball does not necessarily mean that. So, K-BB% is much closer to box score results and, therefore, correlates better with the ERA measure, which is a measure of box score results.
Does Strike%-Ball% Predict Future K-BB%
What I did was split last season into two parts:
March-June
July-October (regular season only)
I then found every qualified pitcher’s K-BB% and Strike-Ball% for both halves, and also found where they ranked in both stats for both halves.
First, here are the correlations:
First Half Strike-Ball% Correlation Coefficients
With 1H K-BB%: 0.77
With 2H Strike-Ball: .638
With 2H K-BB%: 0.581
To put that in words:
First Half Strike-Ball% predicted First Half K-BB% quite well
First Half Strike-Ball% predicted Second Half Strike-Ball% moderately well
First Half Strike-Ball% predicted Second Half K-BB% moderately well, but note quite as good as it predicted Strike%-Ball%
What we’re really aiming to do is predict second-half K-BB%, since K-BB% predicts ERA pretty well. We’re trying to develop a predictor of a predictor, which is probably one step too far removed, but we’re having fun here, right?
First-half K-BB% predicts second-half K-BB% better than first-half Strike%-Ball% does, but it’s closer than I imagined it would be.
1H K-BB%: .665
1H Strike-Ball%: .581
Next, I looked at the biggest 1H K-BB% “underperformers,” judged on the rank differential with 1H Strike%-Ball%.
#1 Sandy Alcantara
1H Strike%-Ball% Rank: 15th
1H K-BB% Rank: 63rd
Difference: 48
The hypothesis would suggest that we would see a better second-half K-BB%.
1H K-BB%: 12.1%
2H K-BB%: 15.3%
So yes, in this one case, we saw the hypothesized movement. We could have been sitting there last June saying, “Alcantara’s K-BB% should improve moving forward, as evidenced by this much higher Strike%-Ball%,” and we would have been right. But one example doesn’t prove anything!
The answer! Of all sampled pitchers, the average difference between their first-half and second-half K-BB% was -0.3%.
Thirteen pitchers had these rank differences of 20 slots or greater, so we would have expected all of them to improve in K-BB% in the second half. The average improvement was +1.9 points to K-BB%. Eight of the 13 improved in K-BB% in the second half, and four of them improved by more than a full point.
Here are the names, shown with their K-BB% differentials.
Alcantara +3.2
Lorenzen -5.1
Mikolas +0.7
Dunning +4.4
Nelson -0.9
Jon Gray +0.6
T Anderson +1.9
Kyle Freeland -0.6
Luke Weaver +0.8
Clarke Schmidt +0.5
Logan Allen -2.0
Julio Urias +1.9
Rich Hill -3.5
It was hardly a rule to trust, but we see more improvement than decline here, which is a small victory for the hypothesis.
The guys on the other side of the rank differential show the same general movement. Only eight of the 26 pitchers with differentials of -10 or less improved their K-BB% by a point in the second half. Twelve of the eighteen at the bottom of the list lost more than one point in K-BB%.
So yes, the hypothesis is that pitchers with big differences in where they line up in Strike%-Ball% vs. K-BB% should move in the same direction moving forward in K-BB%. It’s a slight change, and it’s hardly 100% of the time, but there is a little bit of something there.
Applying the Result
We’re going to apply this very lightly. I’m not making any hard recommendations based on this. But here are the names that we might expect K-BB% improvement from moving forward due to their strong rankings in Strike%-Ball%.
#1 Keaton Winn
Strike%-Ball%: 14.8% (31st in league)
K%-BB%: 11.8% (114th in league)
#2 Joey Estes
Strike%-Ball%: 20.8% (6th)
K%-BB%: 14.4% (81st)
#3 Mitchell Parker
Strike%-Ball%: 15.6% (26th)
K%-BB%: 14% K-BB% (90th)
#4 Bryan Woo
Strike%-Ball%: 25.8% (1st)
K%-BB%: 16.7% (52nd)
#5 Nathan Eovaldi
Strike%-Ball%: 16.0% (25th)
K%-BB%: 16.0% (63rd)
#6 Spencer Schwellenbach
Strike%-Ball%: 23.2% (2nd)
K%-BB%: 17.6% (40th)
Those names have a lot in common. Besides Eovaldi, they are all rookies or second-year pitchers, and besides Winn, they all have very low ball rates and below-average SwStr%.
I think that the second point is the driver here. These are pitchers generating high strike rates primarily by throwing a ton of pitches in the zone. That will turn into extra called strikes, but not necessarily strikeouts, because the hitters will swing at close pitches with two strikes. The SwStr% on each pitcher:
Winn: 13.2%
Estes: 11.6%
Parker: 11.3%
Woo: 11.2%
Eovaldi: 14.5%
Schwellenbach: 14.8%
It’s pretty enticing to see the higher marks here on Schwellenbach, Eovaldi, and Winn. They seem to be doing everything you want to post a high K-BB%, but not quite getting the results.
So I’ll finish this post by predicting that those three names will generate higher K-BB% marks for the rest of the season, provided they stay healthy.
I will even set a reminder for myself to go back and check to see if we were right!
As for actionable advice, I’m not really going to give any on this one. It might be worth acquiring these three names if you believe in what I’m saying, but I think I haven’t proved the hypothesis firmly enough to change anything with it.
It's a lame ending, but that’s what happens when you start writing before you know how it’s going to turn out! Talk to you later.