Using Game-By-Game Fastball Velocity to Influence SP Sit/Start Decisions

Eno Sarris’s latest article and starting pitcher ranks include details about pitcher recency and how starting pitchers tend to “go in and out of funks”. It’s a good article, you should read it. But, I was much more intrigued by a Rob Arthur article Eno referenced from 2017 that I had never read before. Arthur and Greg Matthews summarize their research that seeks to predict in-season starting pitcher highs and lows, or hot and cold streaks, based on a pitcher’s fastball velocity. As a fantasy manager whose starters recently gave up 19 earned runs in one night, my interest was peaked by this gem of an article.

While Arthur and Matthews do not write from a fantasy baseball perspective, the implications on the fantasy community are great. Imagine if you could base a start/sit decision for a starting pitcher on something as simple and easy to look up as fastball velocity. As usual, I’m standing on the shoulders of giants as I write this article. I won’t be recreating any models, but simply taking a look at whether or not we can use Arthur and Matthews’ research to benefit our fantasy teams. Here goes.

I highly recommend reading Arthur and Matthews’ article before going on, but here’s the main point of focus for the rest of my writing:

“we can also predict whether a pitcher will be hot or cold in the future. Using just the first two months’ worth of 2016 data, we tried to predict every pitcher’s subsequent fastball velocity.7 Our model was able to predict how hard the next pitch would be better than a guess based on the pitcher’s season-long average would be able to, suggesting that it’s able to pick up on when a pitcher is hot or cold at any point in the season after June 1.”

I won’t be using a model to predict, instead, I’ll look at qualified starting pitchers’ game-by-game four-seam fastball velocity and I’ll try to determine if an uptick in a single starts velocity compared to the same season average relates to a “good” performance in the following game. Whew. That was a mouthful. Here’s a walkthrough example. Prior to June, Jordan Montgomery’s four-seam fastball velocity was 92.4 (Pitch Info) and in his last two starts in May he threw it above that average, 92.6 on May 24th and 93.5 on May 31st. Both are slightly above average, but better yet, he increased over that two start period, increasing his average from one start to the other, games 9 and 10 below:

Jordan Montgomery Fastball Velocity Per Game

So the question is, did Montgomery perform well in the very next game? We can look at his game logs and pick out a few things that help us measure that:

Jordan Montgomery Game-Logs
Date Opp W ERA IP TBF H R ER HR BB SO GSv2
2022-06-11 CHC 1 0.00 7.0 28 5 0 0 0 0 5 76
2022-06-05 DET 0 2.84 6.1 24 5 2 2 0 1 5 64
2022-05-31 LAA 1 1.29 7.0 25 4 1 1 1 1 4 66
2022-05-24 BAL 0 3.00 6.0 22 4 2 2 1 0 5 60
2022-05-19 @BAL 0 5.40 5.0 21 7 3 3 1 0 3 43

In the visual above, we noted Montgomery’s May 24th and May 31st increasing fastball velocity and in his following start, June 5th, he performed well. Was it the start of his career? No. But from a game score perspective, he was above average. However, his June 11th game looks like a gem, where he went seven innings and gave up no runs. In that June 11th game, his game-average fastball velocity (93.0) stayed above his season average (92.4).

Now that we have a working example let’s apply this way of thinking to the 2022 season as a whole. I’ll go through the same process with all pitchers who are qualified pitchers as of today. I’ll find their season-long four-seam fastball average through May and I’ll look for pitchers whose last two games in May had fastball velocities above their season average. There were only seven pitchers who qualified under those whacky parameters:

Fastball Velocity Improvers, Last Two Starts in May
Name Season Average Through May Second to last start in May Fastball Velo Last start in May Fastball Velo
Gerrit Cole 97.6 97.7 97.8
Miles Mikolas 93.2 93.8 94.4
Kevin Gausman 94.4 95.0 95.3
Chris Bassitt 92.7 94.1 94.2
Jordan Montgomery 92.2 92.8 93.2
Tyler Wells 93.7 94.1 94.1
Tarik Skubal 94.1 94.3 94.7

Of these seven pitchers, here are the improvement rates in their very next start:

ERA: four out of seven improved, allowing fewer earned runs to score in their following start (57%).

Strikeouts: three out of seven struck out more batters in their next start (43%).

Game Score (GSv2): three out of seven improved in their next start (43%).

Walks: six out of seven walked fewer batters in their next start (86%).

Hits: four out of seven allowed fewer hits in their next start (57%).

Yes, I know. I’ve taken someone else’s research on a large sample and I’ve applied it to a tiny one. But, if we want to make this actionable, we really don’t have much of a choice. We are trying to determine if upticks in fastball velocity in two games can help us determine if the very next game will be a good one or not. If we believe there is validity to using this system and we want to use it, we need a way to crunch data and supply a list of pitchers who qualify under these conditions. I’ve written some python code that takes in savant pitch level data to accomplish this and I’ll work on automating it to run reports on a regular basis if there is an interest. Here are the pitchers who are currently qualified, who have shown increased four-seam velocities over their last two or three starts, and who’s last two or three start’s fastball velocities have been above their season average:

Starters With Game-By-Game Pitch Velocity Increases
Name Season Average Third Most Recent Second Most Recent Most Recent
Miles Mikolas 93.13 92.37 93.18 93.45
Taijuan Walker 93.93 94.26 94.41 94.59
Taylor Hearn 94.34 94.72 94.86 94.87
Sandy Alcantara 97.82 98.30 98.13 98.28
Ian Anderson 93.73 94.05 93.83 94.10
Nick Pivetta 93.23 94.12 93.53 94.00
Jose Quintana 90.74 91.19 91.58
Patrick Corbin 91.55 91.84 92.65
Jon Gray 95.45 95.64 96.30
Germán Márquez 94.94 95.75 96.14
Jordan Montgomery 92.24 92.40 93.27
Shohei Ohtani 97.19 97.55 98.29
Joan Adon 94.80 95.10 95.11
Josiah Gray 94.07 94.34 95.15

There are some interesting things happening in this list. First, we see pitchers who have been bad to start the season and are likely hanging out on the wire. Second, we have a few pitchers who have been slowly ramping back up from injury. Finally, we see pitchers who have had a few good outings recently. If you are rostering any of these pitchers there is reason to believe that their next start could be good, or just as good as they have been performing, i.e., decreasing the likelihood of a stinker. If they are on the wire and you feel like streaming, the list above might be a good place to start but take caution. Take a look at their peripherals and matchups to better inform your decision.





6 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
viceroymember
16 days ago

I find myself always streaming pitching and wasting time hunting down recent tends. This is a cool idea and would be interested in seeing the code! May be nice to track streaming results using this method