2016 Weighted Arsenal Scores

Around this time last year (edit: actually, it was more like sometime in 2014), Eno Sarris introduced the Arsenal Score. It was, and still is, a novel concept: for every pitcher, evaluate each of his pitches based strictly on their strikeout- and ground ball-inducing tendencies. Each pitch would be evaluated relative to its contemporaries — in other words, Corey Kluber’s slider would be compared to all other sliders in the league.

I’ll speak for Eno when I say the original Arsenal Scores weren’t meant to be especially rigorous. They received some flak for being mathematically inaccurate — to which I say, it doesn’t really matter. Originally, Eno calculated separate Z-scores for the ground ball rate (GB%) and swinging strike rate (SwStr%) — called “Z-BIP” and “Z-Whiff,” respectively, in the results to follow — of each pitch for every pitcher. The aggregate Z-scores — two Z-scores times X number of pitches — comprise the full Arsenal Score.

This time around, I propose a few tweaks:

  • GB% + PU%: Ground balls are excellent, but pop-ups are the theoretically optimal batted ball outcome — the ball-in-play equivalent to a strikeout. Thus, I added pop-ups (PU%) to ground balls to capture critical outcomes, albeit simplistically.
  • Pitch weights: Eno had equally weighted every pitch in a pitcher’s repertoire. This isn’t heavy criticism; I simply saw this as a flaw in which, for example, the game’s low-key best pitch might be overrated if it wasn’t actually thrown very often. Instead, I chose to weight each pitch by the actual frequency at which it was thrown.
  • Z-score weights: This is more experimental than anything, and less of a critical update than the first two bullet points: Eno weighted the SwStr% Z-score twice as heavily as the GB% Z-score based on each statistic’s correlation with various ERA and FIP metrics (see original post). I’ll provide three separate scores: (1) equal weights, (2) Eno’s “2/3” weights, and FIP weights by which strikeout (or, in this case, strikeout inputs) are weighted 6.5 times more heavily than ground balls.

It’s true — this isn’t a mathematically legitimate way to do things. Frankly, I don’t care. Again, this isn’t (and wasn’t) mean to be anything more than a crude but interesting way to evaluate pitchers, especially in terms of identifying guys who may have breakout potential. A mathematically valid approach would likely involve building a regression equation that ultimately results in something closely reflecting FIP. It wouldn’t necessarily be a redundant exercise — a piecemeal FIP could be really fascinating — but it’s much more labor-intensive than this was perhaps ever meant to be.

So, let’s dig in. But first, the nitty gritty: I excluded all pitches thrown fewer than 50 times last year. Accordingly, a pitcher’s repertoire might not incorporate all his pitches. One can argue the pitches a pitcher throws least frequently are his least-effective. Their omissions, therefore, likely inflate the scores you see below by some undetermined amount for each pitcher. The “Freq” column indicates, from 0 to 100 percent, how much of the pitcher’s repertoire is accounted for in his scores. A lower “Freq” should be interpreted as the scores being more volatile and also perhaps inflated; a score of 100% means you have nothing to worry about. Lastly, I notice some names, such as my boy David Phelps, who don’t show up as starting pitchers here. It all depends on how Baseball Prospectus’ PITCHf/x leaderboard (P.S. thank you!) classifies a pitcher; I toggled by starting pitchers only.

Also: the nominal magnitude of the scores themselves mean nothing. Use them strictly for ranking purposes. Everything is relative.

With that said, here are your top-100 Weighted Arsenal Scores for 2016 starting pitchers , sorted by their FIP-weighted scores. (Note: Sort any column by clicking on its header!)

2016 Weighted Arsenal Scores
Rk Name Z-Whiff Z-BIP Equal Weights 2/3 Weights FIP Weights Freq Pitches
1 Max Scherzer 1.98 0.26 2.24 4.23 13.16 100% 5
2 Clayton Kershaw 1.85 0.97 2.82 4.67 13.02 100% 3
3 Noah Syndergaard 1.70 0.92 2.62 4.33 12.00 100% 5
4 Michael Pineda 1.69 0.53 2.22 3.91 11.51 100% 3
5 Jose Fernandez 1.67 0.40 2.07 3.74 11.26 100% 4
6 Mike Montgomery 1.56 1.11 2.67 4.22 11.24 93% 4
7 Lance McCullers 1.39 1.27 2.66 4.06 10.33 98% 3
8 Matt Shoemaker 1.40 0.41 1.81 3.20 9.50 99% 4
9 Corey Kluber 1.36 0.55 1.91 3.27 9.38 100% 5
10 Yu Darvish 1.41 0.11 1.53 2.94 9.30 98% 5
11 Chris Archer 1.30 0.76 2.06 3.37 9.23 100% 3
12 Carlos Carrasco 1.30 0.75 2.05 3.34 9.17 100% 5
13 Jharel Cotton 1.16 1.25 2.41 3.57 8.79 84% 3
14 Cole Hamels 1.22 0.74 1.96 3.18 8.67 100% 5
15 Justin Verlander 1.28 0.19 1.47 2.75 8.50 99% 4
16 Jon Gray 1.20 0.63 1.83 3.03 8.45 100% 5
17 Sean Manaea 1.18 0.71 1.90 3.08 8.41 100% 3
18 James Paxton 1.17 0.75 1.93 3.10 8.38 100% 5
19 Danny Duffy 1.26 0.08 1.35 2.61 8.30 99% 4
20 Dylan Bundy 1.17 0.64 1.81 2.98 8.24 97% 3
21 Alex Reyes 1.07 1.15 2.22 3.29 8.10 87% 3
22 David Price 1.14 0.49 1.63 2.77 7.92 100% 5
23 Kenta Maeda 1.13 0.55 1.68 2.82 7.92 100% 5
24 Nick Tropeano 1.17 0.21 1.38 2.56 7.84 100% 4
25 Kevin Gausman 1.06 0.86 1.92 2.98 7.77 100% 5
26 Robbie Ray 1.10 0.63 1.73 2.82 7.75 100% 5
27 Vince Velasquez 1.15 0.01 1.16 2.32 7.51 100% 5
28 Madison Bumgarner 1.09 0.41 1.50 2.59 7.50 100% 4
29 Francisco Liriano 1.01 0.88 1.88 2.89 7.42 100% 4
30 Tyler Glasnow 0.98 0.99 1.97 2.95 7.39 97% 2
31 Jacob deGrom 1.02 0.65 1.67 2.69 7.30 100% 5
32 John Lackey 1.07 0.31 1.37 2.44 7.24 100% 5
33 Stephen Strasburg 1.05 0.37 1.42 2.47 7.21 100% 4
34 Danny Salazar 1.00 0.66 1.66 2.65 7.15 100% 5
35 Chad Green 1.09 0.01 1.11 2.20 7.12 97% 4
36 Chris Sale 1.03 0.39 1.43 2.46 7.12 100% 4
37 Rich Hill 0.98 0.71 1.69 2.67 7.09 98% 3
38 Cody Anderson 1.02 0.25 1.28 2.30 6.90 94% 3
39 Marco Estrada 1.01 0.34 1.34 2.35 6.87 100% 4
40 Drew Pomeranz 0.94 0.74 1.68 2.61 6.82 99% 5
41 Jose De Leon 0.93 0.80 1.72 2.65 6.82 87% 2
42 Masahiro Tanaka 0.93 0.75 1.68 2.62 6.81 100% 6
43 Michael Fulmer 0.89 1.04 1.92 2.81 6.80 100% 4
44 Junior Guerra 0.93 0.77 1.70 2.62 6.79 100% 4
45 Zack Godley 0.91 0.79 1.70 2.61 6.71 100% 4
46 Jake Arrieta 0.86 1.06 1.92 2.79 6.68 100% 5
47 Tyler Anderson 0.88 0.94 1.82 2.70 6.68 99% 4
48 German Marquez 0.96 0.31 1.27 2.23 6.54 88% 2
49 Jeremy Hellickson 0.91 0.58 1.49 2.40 6.48 100% 5
50 Joe Ross 0.92 0.48 1.40 2.33 6.48 99% 4
51 John Gant 0.89 0.63 1.52 2.41 6.41 100% 3
52 Blake Snell 0.96 0.15 1.11 2.07 6.41 100% 4
53 Drew Smyly 0.97 0.09 1.06 2.03 6.38 100% 4
54 Julio Teheran 0.94 0.25 1.19 2.13 6.37 100% 5
55 Garrett Richards 0.87 0.63 1.50 2.37 6.28 96% 4
56 Adam Morgan 0.93 0.13 1.07 2.00 6.19 100% 5
57 Luis Cessa 0.84 0.70 1.55 2.39 6.17 100% 4
58 Eduardo Rodriguez 0.93 0.05 0.98 1.91 6.11 100% 5
59 Alex Wood 0.76 1.15 1.92 2.68 6.11 100% 3
60 Rubby de la Rosa 0.78 1.06 1.84 2.62 6.11 100% 4
61 Daniel Norris 0.90 0.24 1.14 2.04 6.09 100% 5
62 Jon Lester 0.84 0.64 1.48 2.31 6.09 100% 5
63 Collin McHugh 0.86 0.48 1.34 2.19 6.06 100% 5
64 Matt Andriese 0.86 0.46 1.32 2.17 6.04 99% 4
65 Zack Greinke 0.83 0.64 1.47 2.30 6.03 100% 5
66 Carlos Martinez 0.74 1.18 1.93 2.67 6.02 99% 4
67 Bud Norris 0.79 0.88 1.67 2.46 6.01 100% 5
68 Tim Lincecum 0.88 0.27 1.15 2.03 5.99 100% 5
69 Julio Urias 0.84 0.47 1.31 2.15 5.94 99% 4
70 Matt Moore 0.85 0.38 1.22 2.07 5.88 100% 5
71 CC Sabathia 0.70 1.07 1.77 2.47 5.60 99% 5
72 Jorge de la Rosa 0.75 0.73 1.48 2.23 5.59 100% 5
73 Nate Karns 0.80 0.32 1.12 1.92 5.53 100% 4
74 Matt Harvey 0.78 0.46 1.24 2.01 5.51 98% 4
75 Taijuan Walker 0.75 0.61 1.36 2.11 5.49 98% 4
76 Henry Owens 0.87 -0.17 0.70 1.57 5.48 86% 3
77 Johnny Cueto 0.67 1.06 1.74 2.41 5.44 100% 6
78 Kyle Hendricks 0.70 0.92 1.61 2.31 5.44 100% 4
79 Steven Matz 0.67 1.04 1.71 2.38 5.40 100% 5
80 Aaron Nola 0.66 1.08 1.74 2.41 5.39 100% 4
81 Carlos Rodon 0.77 0.38 1.15 1.92 5.38 100% 4
82 Adam Conley 0.77 0.34 1.11 1.88 5.32 100% 3
83 Jason Hammel 0.78 0.27 1.05 1.82 5.32 100% 5
84 Nathan Eovaldi 0.68 0.86 1.55 2.23 5.31 100% 5
85 Ian Kennedy 0.81 0.03 0.84 1.65 5.28 100% 4
86 Ervin Santana 0.72 0.55 1.27 1.99 5.21 99% 4
87 Robert Gsellman 0.63 1.10 1.73 2.36 5.21 91% 4
88 Reynaldo Lopez 0.74 0.36 1.11 1.85 5.20 100% 3
89 Marcus Stroman 0.58 1.39 1.97 2.54 5.13 100% 6
90 Chris Young 0.88 -0.61 0.27 1.15 5.11 99% 2
91 Tim Adleman 0.77 0.07 0.85 1.62 5.11 100% 4
92 Cody Reed 0.66 0.81 1.47 2.13 5.09 100% 4
93 Felix Hernandez 0.66 0.78 1.44 2.10 5.07 99% 5
94 Mike Foltynewicz 0.73 0.34 1.07 1.79 5.06 100% 5
95 Clayton Richard 0.52 1.63 2.15 2.66 4.99 99% 4
96 Joe Kelly 0.65 0.77 1.42 2.07 4.98 91% 4
97 Steven Brault 0.74 0.17 0.91 1.65 4.97 100% 4
98 Aaron Blair 0.67 0.61 1.28 1.95 4.97 100% 5
99 Kyle Gibson 0.65 0.77 1.42 2.06 4.97 100% 5
100 Dallas Keuchel 0.57 1.25 1.82 2.39 4.96 99% 5
SOURCE: http://www.baseballprospectus.com/pitchfx/leaderboards/
Click column headers to sort!

The results seem to validate themselves. Would you hope or expect to see anyone else other than Kershaw, Scherzer and Syndergaard near the top? I won’t break down this list to exhaustion. Instead, I will point out some of the most obviously interesting names (sorted by FIP weights) and include the rank of their scores in the following order: FIP weights / Eno’s weights / equal weights.

Michael Pineda (4th / 6th / 8th): What to do with this guy? Filthy yet simultaneously hittable stuff. Enigmatic. If the home run rate regresses heavily, he could return a hefty profit, but this guy screams of red flags after underperforming his peripherals for two years.

Mike Montgomery (6th / 4th / 2nd): I won’t dive too deep into Montgomery; I featured him exclusively in this September 2016 post. The 10-second version: he threw five pitches that induced ground balls more than 50 percent of the time, and two of those pitches induced spectacular whiff rates. Until further notice, he’s the Cubs’ No.-5. His current NFBC ADP: SP83 (he doesn’t even show up on the starting pitcher page), 315th overall. Absolute bargain.

Lance McCullers (7th / 5th / 3rd): Ah, yes — Weighted Arsenal Scores are immune to walks. McCullers has dominant stuff; if he can chip (er, maybe chunk?) away at his walk rate, he could become a monster.

Matt Shoemaker (8th / 13th / 30th): The FIP weights punish Shoemaker less harshly for his fly ball propensities. Yet 30th among starting pitchers is still pretty dang good, and his swinging strike rate ranked fifth among all pitchers who threw at least 150 innings. At SP66, he’s also a low-risk, high-reward option.

Jharel Cotton (13th / 8th / 5th): Cotton! He has been a Carson Cistulli Fringe Fiver for some time now, and for good reason. Perpetually overshadowed by Julio Urias, Cotton has escaped to Oakland to pursue a full-time gig. His cutter is bonkers, and while his other pitches lag, he has shown real promise even in barely 30 Major League innings. A note, though: his low Freq indicates his score is a bit volatile and likely inflated. Bump him down a few pegs, but at least make sure he’s on your radar if he wasn’t already. Current NFBC ADP: SP71.

Other highly ranked names: Manaea, Paxton, Duffy, Bundy, Reyes, Tropeano (huh!), Gausman, Ray, Velasquez… all in the top-30 FIP-weighted Arsenal Scores.

* * *

Acknowledging the shortcomings of the Arsenal Score approach, I (and I’m sure Eno, too) would love to hear your feedback.

It’s important to remember that this is nowhere near the end-all, be-all of pitcher evaluation. Not even close. Sale, Strasburg, Arrieta, Cueto, Aaron Sanchez, and many more don’t rank particularly well here. There’s obviously a lot that this approach misses. But if you’re a fan of strikeouts, ground balls and pop-ups as shorthand analytical tools, then I hope this exercise points you in the direction of some pitchers you may not have given fair consideration prior to this.

We hoped you liked reading 2016 Weighted Arsenal Scores by Alex Chamberlain!

Please support FanGraphs by becoming a member. We publish thousands of articles a year, host multiple podcasts, and have an ever growing database of baseball stats.

FanGraphs does not have a paywall. With your membership, we can continue to offer the content you've come to rely on and add to our unique baseball coverage.

Support FanGraphs




Currently investigating the relationship between pitcher effectiveness and beard density. Two-time FSWA award winner, including 2018 Baseball Writer of the Year, and 5-time award finalist. Featured in Lindy's Sports' Fantasy Baseball magazine (2018, 2019). Tout Wars competitor. Biased toward a nicely rolled baseball pant.

newest oldest most voted
Pirates Hurdles
Member
Member
Pirates Hurdles

“The results seem to validate themselves.” If you stop at about 20. This seems to fall apart in the middle of the list with some pretty crazy rankings (Cessa>Lester; Green>Sale and so on). Any ideas why this might fall apart beyond the very top?

Anonymous
Member
Anonymous

(in before Luis Cessa and Chad Green are 2017 breakout fantasy MVPs)

wily mo
Member
Member

yeah i love this post but lincecum over urias is a little hair-raising. still, if you want to make an omelet you gotta break some eggs