PitchingBot's Standout 2020 Performers

Here I examine the pitchers that PitchingBot likes the most from the 2020 season. Seeing who has the best stuff, command, and who throws the best pitch of each type.

This article is part of a series on my pitch evaluation model PitchingBot, an overview of the model can be found here.

Introduction

PitchingBot gives me run values and predicted events for every pitch thrown. Grouping the results by pitcher we can see the predicted best performers from pitch quality alone. For each of these rankings I filtered by a minimum of 150 pitches to avoid small sample effects. The rankings are made by taking the top rates of predicted runs saved.

In addition to this post, I've made an app which shows PitchingBot's grades on a 20-80 scouting scale. This includes ratings for the pitchers who didn't make it into this post, find it below:

https://mlpitchquality.shinyapps.io/pitcher_app/

Top Pitchers

1. Gerrit Cole

Credit: ESPN

There's no surprises at the top of this list. In 2019 Gerrit Cole signed the most valuable contract ever for a pitcher and this is why. The raw quality of his pitching is second to none. PitchingBot predicts Gerrit Cole to be worth over a run above average per hundred pitches. PitchingBot ranks his fastball as his most dominant pitch, followed by his slider, then curveball, and the changeup brings up the rear being league average. 
His fastball, slider, and curve all have well above average stuff and command ratings. PitchingBot doesn't think his changeup stuff is anything special, however he locates it extremely well. Remember, PitchingBot doesn't know anything about the other pitches in his arsenal when it gives ratings, which makes it difficult to get accurate changeup rankings since their value is tied to their relationship with the fastball.
These rankings are measuring the rate of runs saved based on pitch quality. Gerrit Cole, a starter who threw the fourth most pitches in 2020, outranked all other pitchers including relievers. To achieve such a result in a rate statistic over such a large sample is a testament to his consistency and ability.

Gerrit Cole		Predicted runs saved above average per 100 pitches
Pitch Type	Number thrown	Overall	Stuff	Command
All Pitches	1308	1.04	0.52	0.56
Four-seam Fastball	690	1.28	0.43	0.55
Slider	311	1.00	0.78	0.84
Knuckle Curveball	232	0.70	0.76	0.21
Changeup	75	0.00	-0.31	0.57

Here is Cole's best pitch of the season according to PitchingBot, with over 50% probability of a swinging strike:

2. Blake Treinen

Credit: Tony Gutierrez / Associated Press

Blake Treinen is known for his exceptional 2018 season where he posted an ERA of 0.78. He was non-tendered by the A's after a disappointing 2019 but bounced back in 2020 as a key part of the Dodgers' bullpen. His appearance at number two on this list is due to his power sinker which he threw over 50% of the time. His secondary offerings include an above average slider and cutter.

Blake Treinen		Predicted runs saved above average per 100 pitches
Pitch Type	Number thrown	Overall	Stuff	Command
All Pitches	407	1.02	0.69	0.24
Sinker	226	1.44	0.77	0.23
Slider	107	0.91	0.90	0.38
Cutter	38	0.84	0.39	0.60
Four-Seam Fastball	36	-1.07	-0.16	-0.57

Here's Blake Treinen's best pitch of 2020, a slider just off the outside corner.

3. Jacob deGrom

Credit: David / Wikipedia

Jacob deGrom had a stellar 2020, leading the league in strikeouts with an uncanny ability to keep adding velocity to his pitches each season. He's the favorite for the 2021 NL Cy Young award for good reason, PitchingBot rates his fastball, slider and curveball as elite, with an above average changeup to round the arsenal off.

Jacob deGrom		Predicted runs saved above average per 100 pitches
Pitch Type	Number thrown	Overall	Stuff	Command
All Pitches	1135	1.00	0.54	0.44
Four-Seam Fastball	510	0.96	0.44	0.16
Slider	403	1.39	0.70	0.98
Changeup	192	0.25	0.50	-0.04
Curveball	30	1.35	0.34	1.10

deGrom's best pitch was this 3-2 changeup to Andrew Benintendi, which was fouled off.

4. Richard Bleier

Credit: KA Sports Photos

The final two entries on this list are surprises to me, relievers who went under the radar with ERAs of ~2 in a short sample size for not great teams. Richard Bleier ranks highly based on his sinker command, he dotted the bottom corner of the zone, outside to righties and inside to lefties. This led to a 2% barrel rate with the average launch angle of batted balls being -4.5˚.

Richard Bleier		Predicted runs saved above average per 100 pitches
Pitch Type	Number thrown	Overall	Stuff	Command
All Pitches	219	0.93	0.24	0.91
Sinker	113	1.61	0.51	1.46
Slider	53	0.27	-0.19	0.72
Cutter	41	-0.24	0.04	-0.49
Changeup	12	1.97	0.44	1.77

His best pitch was this double play ball to Jeff McNeil, which is thrown in the perfect location to induce a groundball. Despite this McNeil gets good contact on it. It's funny how baseball works sometimes, an excellent pitch ends up with a bad outcome, a hard hit burner. But then that ball is caught and they turn the double play, what a rollercoaster.

5. Joely Rodríguez

Credit: Smiley N. Pool / The Dallas Morning News

Joely Rodriguez managed to sneak into the number 5 spot and I'm not really sure why. PitchingBot loves his changeup and this props up his ranking. The stuff and command models thought it was a good pitch but nowhere near as good as the overall model predicted. 

He had a 2.13 ERA for the Rangers in 2020 throwing a three pitch mix of a sinker, changeup and four-seamer. He threw these pitches in decent spots, going high with the four-seam fastball, low with the sinker, and even lower with the changeup. This model of pitching is not unique to him, so it is strange to see PitchingBot give such a high rank. This illustrates one of the drawbacks of machine learning models, a lack of explainability, we can't easily dig into PitchingBot's brain to figure out why it gives certain ratings.

Joely Rodriguez		Predicted runs saved above average per 100 pitches
Pitch Type	Number thrown	Overall	Stuff	Command
All Pitches	234	0.91	0.27	0.21
Sinker	92	0.76	0.46	0.30
Changeup	70	2.29	0.59	0.40
Four-Seam Fastball	66	-0.05	-0.32	0.11
Slider	6	-2.08	0.47	-2.16

His best single pitch was this sinker to strike out Shed Long Jr.

Stuff Rankings

PitchingBot allows me to make predictions based on stuff and command alone. The pitchers who had the nastiest stuff in the 2020 season are listed below. It's interesting to see that PitchingBot is familiar with the Joe Kelly has great stuff meme.

Player	Pitches thrown	Predicted stuff runs above average per 100 pitches
Aroldis Chapman	223	0.86
Tanner Scott	331	0.83
Joe Kelly	171	0.76
Chaz Roe	159	0.76
Eric Yardley	436	0.75

And here is the pitch which had the highest stuff rating:

Command Rankings

The command rankings rate the players who threw the ball in the right place according to PitchingBot. Brandon Leibrandt makes an appearance, meeting the minimum pitch threshold exactly. He had 7 walks in 9 innings pitched last season, not ideal when we're supposed to be measuring good command. After having a look at his walks it seems there were a few tough calls which went against him. In addition, there is much more to command than walks, where the ball is thrown in counts with fewer than 3 balls matters just as much.

Player	Pitches thrown	Predicted command runs above average per 100 pitches
Richard Bleier	219	0.91
Evan Marshall	391	0.70
Kenta Maeda	1077	0.69
Brandon Leibrandt	150	0.69
Marco Gonzales	1057	0.68

The pitch which had the maximum command rating in 2020 was, ironically, a ball which led to a walk!

Finally I'll run down the rankings for each pitch type:

Four-Seam Fastball Rankings

Player	Pitches thrown	Predicted runs above average per 100 pitches
Gerrit Cole	690	1.28
Rowan Wick	161	1.27
Walker Buehler	363	1.19
Nick Anderson	174	0.96
Jacob deGrom	510	0.96

Sinker Rankings

Player	Pitches thrown	Predicted runs above average per 100 pitches
Justin Topa	103	1.81
Richard Bleier	113	1.62
Blake Treinen	226	1.44
Luis Castillo	291	1.37
Eric Yardley	287	1.14

Slider Rankings

Player	Pitches thrown	Predicted runs above average per 100 pitches
Daniel Norris	106	1.59
Tanner Scott	130	1.52
Ross Stripling	128	1.45
Adam Ottavino	160	1.41
Chris Stratton	146	1.39

Curveball Rankings

Player	Pitches thrown	Predicted runs above average per 100 pitches
Joe Kelly	108	1.62
Aaron Nola	306	1.40
Seth Lugo	118	1.26
Lucas Sims	163	1.25
Shane Bieber	363	1.19

Changeup Rankings

Player	Pitches thrown	Predicted runs above average per 100 pitches
Wandy Peralta	148	1.97
Evan Marshall	151	1.89
Zac Gallen	212	1.59
Trevor Rogers	103	1.59
Alex Claudio	126	1.58

Splitter Rankings

Player	Pitches thrown	Predicted runs above average per 100 pitches
Jeremy Jeffress	113	1.05
Kevin Gausman	270	0.99
Alex Cobb	283	0.98
Masahiro Tanaka	204	0.68
Hector Neris	192	0.54

Thanks for reading if you made it this far! If you didn't see the name of a pitcher you were looking for, then check out the app to see what PitchingBot thought of their 2020 season.