Pitch Combos Part 2: Batted Balls Only

Introduction

Last week, I posted an article on the best pitch type and location combinations in the strike zone. In that analysis, I included swinging and called strikes and assigned these pitches an xwOBA of zero. I noted that one could argue this choice gives too much weight to strikes relative to batted balls. Today, I am exploring the exact same question except exclusively on batted balls. Other than that, my methodology is exactly the same, so I encourage you to check out that article before reading any further.

Results

LHP on LHB (Left handed pitcher on Left handed batter)

Best:

Pitch Type	Zone	Average xwOBA
Sinker	9 (Down and In)	0.284
Slider	7 (Down and Away)	0.290
Two-Seam FB	9	0.292
Sinker	8 (Middle-Down)	0.298
Slider	4 (Middle-Away)	0.307

Worst:

Pitch Type	Zone	Average xwOBA
Four-seam FB	5 (Middle-Middle)	0.465
Cutter	5	0.453
Four-seam FB	8 (Middle-Down)	0.429
Four-seam FB	4 (Middle-Away)	0.410
Slider	8	0.407

A lot of shakeup from last week’s leaderboard. A few constants: fastballs continue to get crushed in the middle of the zone and away in the zone generally seems the best place to go. The effectiveness of down and in once again surprised me based on lefty hitters “dropping the barrel” on that location. Out of all the data, the batted ball only lefty-on-lefty split seemed to be the biggest outlier in terms of “best” zones, and I suspect it’s partially because it has the least amount of observations, giving it more statistical “noise” (only 7746 of the roughly 700,000 total 2018 pitches were lefty-on-lefty matchups that resulted in fair contact).

RHP on LHB

Best:

Pitch Type	Zone	Average xwOBA
Changeup	7 (Down and Away)	0.296
Four-seam FB	3 (Up and In)	0.299
Sinker	7	0.303
Two-seam FB	7	0.329
Slider	7	0.335

Worst:

Pitch Type	Zone	Average xwOBA
Two-seam FB	5 (Middle-Middle)	0.500
Sinker	5	0.472
Four-Seam FB	5	0.452
Four-Seam FB	4 (Middle-Away)	0.451
Two-Seam FB	8 (Middle-Down)	0.444

In terms of contact prevention, down and away seems key here.

LHP on RHB

Best:

Pitch Type	Zone	Average xwOBA
Four-Seam FB	1 (Up and In)	0.281
Changeup	9 (Down and Away)	0.294
Two-Seam FB	9	0.310
Sinker	9	0.330
Cutter	4 (Middle-In)	0.353

Worst:

Pitch Type	Zone	Average xwOBA
Two-Seam FB	5 (Middle-Middle)	0.495
Four-Seam FB	8 (Middle-Down)	0.473
Four-Seam FB	6 (Middle-Away)	0.461
Sinker	5	0.457
Four-Seam FB	5	0.448

Again, down and away seems like a good contact manager.

RHP on RHB

Best:

Pitch Type	Zone	Average xwOBA
Two-Seam FB	1 (Up and In)	0.283
Cutter	9 (Down and Away)	0.286
Slider	9	0.287
Curveball	9	0.291
Two-Seam FB	9	0.301

Worst:

Pitch Type	Zone	Average xwOBA
Cutter	5 (Middle-Middle)	0.487
Four-Seam FB	5	0.463
Four-Seam FB	8 (Middle-Down)	0.458
Four-Seam FB	3 (Up and Away)	0.439
Changeup	5	0.437

Like the previous article, down and away and up and in have great success in righty-righty matchups.

Conclusion

Zone 8, or middle-down in the strike zone, stood out to me. While it did not make an appearance at all in the previous analysis, it consistently showed up here in the “Worst” leaderboards. If you look back at this article’s tables, you can see that in every batter/pitcher split, a fastball in zone 8 appears. Its previous absence indicates that while zone 8 gives up poor contact, it gets enough takes and swings and misses to make it, at the very least, not the worst of the worst.

I also noticed that high pitches never showed up in the “Worst” leaderboards for both analyses, not even once. Pitching up in the zone more, even with breaking balls, could be an underrated strategy.

There doesn’t seem to be much evidence of lefties loving the pitch down and in. In fact, this data indicates down and in could be an effective spot against lefties for both right-handers and left-handers. I tried searching for “lefties dropping the barrel” and “lefties down and in” on Google, but could not find anything definitive about the cliche’s origin. Regardless of the origin, the data does not back this cliche up.

I prefer my previous results to these. After mulling it over, I’ve concluded called strikes and especially swinging strikes are far too valuable to omit in any pitch analysis. Additionally, omitting these strikes makes combos that induce more contact look better than they should, while doing the opposite for pitches inducing whiffs and takes. However, I could see focusing exclusively on batted ball data useful for pitchers who don’t generate many swinging strikes/called strikes.

Future Research

While leaderboards are good for outlier detection and fun to look at, it misses the middle portion of the data, which can still be very insightful. I have an idea for building a pitcher statistic based on every single pitch thrown. The model would take in many features of a pitch, including velocity, location, movement, spin rate, prior pitch data (to account for sequencing), and likely more, and would output probabilities of different outcomes (probability of a swinging strike, a home run, etc.). After summing up a pitcher’s probabilities from every pitch thrown, you could formulate a statistic that evaluates pitchers on their fundamental underlying data; their actual pitches. However, I would likely need a machine learning model, possibly a neural network, in order to approach this problem, so until I become comfortable with implementing machine learning with R, this idea is on hold.

Thank you for reading! Please comment any other baseball topics you are interested in reading about or any thoughts you had. If you enjoyed my analysis, please follow me on Twitter for future posts.