Sonny Gray: The Anti-Chris Sale
Earlier this week, I took a look at the AL Cy Young race, utilizing batted-ball metrics to address the respective candidacies of David Price, Dallas Keuchel and Chris Sale. To make a long story short, I concluded that Sale’s strikeout-to-walk (K/BB) superiority, along with solid contact management skills that have been obscured by his horrendous team defense, placed him on top. Some readers expressed incredulity in the comment section, not believing that even the worst defense in the game could cost a pitcher one full point of ERA.
Today, let’s look at the counterweight to Chris Sale, Sonny Gray. Though he wasn’t quite the same guy in the second half as he was in the first, he has wrapped up a very strong campaign, especially in the more traditionally accepted statistical categories. He’ll finish third in the AL in ERA (2.73) and is likely to receive his share of down-ballot Cy Young votes, possibly enough to nose out Sale for third place overall. Sale’s ERA is 0.78 higher than his FIP, and as we saw the other day, 1.02 higher than his “tru” ERA, which incorporates Statcast batted ball data. Gray finishes 2015 with an ERA 0.73 lower than his FIP. What gives, and what is the true talent level exhibited by Gray this season?
I first saw Sonny Gray pitch in the summer between his junior and senior year of high school, at the East Coast Pro Showcase, then held in Wilmington, NC. He was a standout, in more ways than one. He stood five inches or more shorter than his pitching contemporaries, and likely weighed about 155 pounds soaking wet. His stuff, however, stood out even more than his slight stature; he ran his fastball into the upper-90s and spun a tight, hard curve ball. His arm action, delivery and athleticism were all solid. For me, he was the best pitcher at the showcase, and a surefire #1 pick the next summer.
Gray committed shortly thereafter to Vanderbilt, making it very clear that he would attend, barring his receipt of an offer he absolutely could not refuse. He wasn’t quite as dominant during his senior of high school as he was in Wilmington, and no club opted to pony up a Godfather offer to a sub-six foot right-handed pitcher.
Gray was electric at times during his collegiate career, but wasn’t always at his best during his junior season, causing him to drop into the middle of the first round. The A’s were there to catch his fall at 18th overall, and have been thankful for their good fortune ever since.
He didn’t exactly dazzle in his minor league career, at least at first. He was jumped to Double-A in his second pro outing after signing, but then struggled through a difficult first full pro season in 2012, striking out only 97 batters in 148 innings at Midland. He made Baseball America’s Top 100 prospect list prior to that season, but never returned.
Each year, I compile my own ordered minor league lists of top full-season-league position player and starting pitcher prospects based on performance and age relative to league and level. These basically serve as follow lists, with the orders then tweaked based on traditional scouting methods. He didn’t meet the standards for inclusion on my post-2012 list. After a solid bounce-back season at Triple-A Sacramento in 2013, in which he struck out a batter per inning and posted a good but not great 1.44 ground out/air out ratio, he did check in at #62 on my list. Still, not exactly the stuff of which future staff aces are typically made.
He made 10 starts for Oakland at the end of that 2013 campaign, and was immediately superior to the pitcher he had been in the minor leagues. He maintained his Triple-A K rate, his grounder rate spiked, and he benefited from pitching in a spacious park for a club that valued defense. He has been one of the most consistent pitchers in the game since his arrival in the majors.
How much of this success is attributable to Gray himself, and how much accrues to these surrounding, contextual factors? Let’s take a detailed look at his 2015 performance by examining his plate appearance frequency and production by ball-in-play (BIP) type data. First, the frequency information:
| Gray | % | REL | PCT |
| K | 20.3% | 100 | 56 |
| BB | 7.1% | 95 | 60 |
| POP | 2.5% | 78 | 30 |
| FLY | 28.2% | 92 | 25 |
| LD | 16.6% | 79 | 3 |
| GB | 52.7% | 116 | 87 |
Gray’s K and BB rates both sit squarely in the average range, with both trending down (K rate from a percentile rank of 68 to 56, BB from 71 to 60) a bit from their 2014 levels. His K/BB ratio is not special, especially for a top-of-the-rotation pitcher. He is clearly a significant ground ball generator; his 2015 grounder rate percentile rank of 87 is essentially unchanged from 90 in 2014. Conversely, one would expect his pop up (30) and fly ball (25) percentile ranks to be quite low; in fact, those marks are actually up slightly from 13 and 10 last season.
Then there’s his extremely low liner rate allowed, evidenced by a percentile rank of 3. Liner rates fluctuate much more so than those of any other BIP type; there aren’t many pitchers who can maintain such performance over the long haul. The early returns suggest that Gray has a chance to be such a pitcher, as his 2014 liner rate percentile rank was 26. While some regression should be expected from his rock-bottom 2015 liner rate, perhaps less should be expected from Gray as compared to many of his peers.
Gray’s BIP frequency profile is very strong, and it propels him above the average starting pitcher status that his K and BB rates would seem to imply.
Frequency rates can only tell us so much; to fill out the rest of the picture, let’s get a read on his authority level allowed by examining Gray’s production allowed by BIP type data:
| Gray | AVG | OBP | SLG | REL PRD | ADJ PRD | ACT ERA | CALC ERA | FIP | TRU ERA |
| FLY | 0.133 | 0.265 | 34 | 87 | |||||
| LD | 0.645 | 1.073 | 106 | 107 | |||||
| GB | 0.212 | 0.228 | 69 | 106 | |||||
| ALL BIP | 0.276 | 0.403 | 71 | 89 | |||||
| ALL PA | 0.216 | 0.271 | 0.314 | 69 | 84 | 2.73 | 2.67 | 3.46 | 3.22 |
The big takeaway here? Once adjustment is made for contextual elements such as home park and defense, Gray is transformed from an excellent contact manager into merely a good one, whose BIP frequencies are quite good, but who allows relatively authoritative contact.
Gray has allowed an incredibly low .133 AVG-.265 SLG on fly balls this season, good for a 34 REL PRD, or Unadjusted Contact Score. Adjusted for context, that figure moves up to an 87 Adjusted Contact Score. Chris Sale’s fly-ball Adjusted Contact Score was better, at 77, though he allowed much more damage in the air (.161 AVG-.473 SLG) because of a smaller ballpark and much worse outfield defense.
Gray induces tons of grounders, and the limited output he has allowed on them (.212 AVG-.228 SLG, 71 Unadjusted Contact Score) might suggest that they are hit extremely weakly as well. That isn’t the case, as context drives his ground ball Adjusted Contact Score way up to 106. Yes, the grounders allowed by Gray are hit harder than league average. He has been the beneficiary of strong infield defense, which encompasses positioning, overshifting, etc. Sale, on the other hand, allowed much, much weaker grounder contact — his ground ball Adjusted Contact Score was 74 — but again allowed much more damage (.292 AVG-.306 SLG) on that BIP type.
On all BIP, Gray has yielded only a .276 AVG-.403 BIP, for a 71 Unadjusted Contact Score that is even better than Dallas Keuchel’s. Adjusted for context, that mark rises to 89, a bit higher than Keuchel’s Adjusted Contact Score of 83. Add back the Ks and BBs, and Gray’s “tru” ERA is 3.22, 0.49 higher than his ERA, but 0.24 lower than his FIP. His ERA gives undue credit to Gray for factors such as his ballpark and defense, while his FIP doesn’t give him quite enough credit for his ability to contain authority in the air. Chris Sale is a superior pitcher to Sonny Gray in every way, though their respective ERAs would tell you otherwise. The difference between their “tru” ERAs (2.46 for Sale, 3.22 for Gray), actually is exactly the same as the difference between their FIPs (2.70 vs. 3.46). Both pitchers are better than their FIP, by exactly the same amount.
The difference in their average velocities allowed by BIP type is quite stark. Gray, with his 34 fly ball Unadjusted Contact Score? He’s allowed average fly ball velocity of 90.3 mph, in the average range. Sale is over a half standard deviation better than average, at 88.7 mph. Liners? Sale is over a full standard deviation better than average at 90.2 mph, Gray in the average range at 93.0 mph. Though Gray has a much more pronounced grounder tendency than Sale, the latter squelches grounder authority to a much greater extent; Gray’s average grounder velocity allowed is over a half standard deviation higher than league average at 87.2 mph, Sale’s is over a full standard deviation below at 81.5 mph.
On all BIP, Gray has allowed an average velocity of 88.8 mph, over a half standard deviation higher than average, while Sale’s mark of 85.8 mphis just under two standard deviations lower than average. There really isn’t much of a comparison between them in any regard. Gray is a reliably good pitcher, but Sale is a force of nature whose impact has been muted by the sea wall that is the White Sox defense.
The ultimate goal of baseball player evaluation is to separate the player’s true talent from the context surrounding it. It’s pretty difficult business, but there is an ever-increasing array of tools at our disposal to assist us in our efforts. Batted-ball data is far from perfect; in fact, in past articles I have focused on some of the aspects of Statcast that limit its usefulness, at least in the short term. It does, however, allow us to take the next steps beyond FIP and xFIP as we strive for more precise measurement of pitchers’ true talent levels.
Fascinating, thanks. I didn’t read the comments on the prior article but I find it slightly amusing that there are people who read Fangraphs who don’t think context matters.
Great article. The conclusions were strong, but well supported.
People who don’t believe the White Sox defense can cost a pitcher a run a game haven’t watched White Sox Baseball.
Sale’s BABIP is .326. If you look down the fWAR leaderboard, you don’t get to a number higher than that until number 14 when you find someone at 327. Who is that, you might ask? Jose Quintana. There are only two other pitchers in the top 50 that give up hits on batted balls more often – Gio Gonzalez and Andrew Cashner.
The White Sox have a decent pitching staff and very good starters. Yet their team BABIP Is worst in the AL by a big margin. So either their pitchers give up harder contact than most (which you would not expect based on the batted ball rates and their ability to miss bats in general) or the guys playing behind them are bad, bad, bad.
the anti-Christ sale
get em cheap, while they’re hot
The funny thing is, the A’s don’t have a strong defensive squad. I understand park factors are the other big part of the equation here and you have two pretty extreme parks comparing Sale and Gray, but this mainly reinforces how bad the Sox defense is.
I was thinking the same thing! semien at short was particularly bad this year. lawrie did not put the same UZR as he did in the past in TOR either. As you say, must be more about CHI bad defense than good OAK defense.
One factor not mentioned here is that Sonny is likely one of the best defenders at his position.
How dependent are the production scores on the classifications of batted ball types? I ask because there’s a slight difference between how you/Fangraphs categorize each batted ball type and how BBREF does (not a big surprise, or a big deal).
The numbers at bbref are a bit different, but especially for fly balls, we’re talking about a small number of events. Bbref gives Gray “Credit” for 13 hits on fly balls all year. Sale was at 21. How much adjustment do these low-n figures get?
Also, what should we think about the A’s infield defense? By UZR, they’re uniformly awful. By DRS, they’re not bad. But how does a mediocre-at-best IF turn harder-than-average balls in play into stunningly good ratio of outs to hits?
Don’t know off the top of my head how much the A’s shift, but I’m pretty sure that neither system uses data from plays with shifts in effect. If the A’s shift a lot, then your sample size of plays goes down considerably and the already suspect defensive numbers get even shakier.
I’m not saying that UZR or DRS aren’t useful, you just have to take what they say in context and with a grain of salt.
Tony’s data seems to be outcome oriented, and doesn’t care how hard the play was, just that the outs were made. This would make shifting teams theoretically better in Tony’s data and worse in UZR/DRS, I believe.
Pretty sure DRS did and UZR didn’t, which is why, for example, Brett Lawrie was +20 in DRS and +4.5 in UZR back when he’d shift to short RF in Toronto.
In any event, the adjustments for quality of defense can’t be purely about outs/no-outs – we already know Gray gave up X hits on Y GBs, what happens after that in the “Adjusted” step? How hard the ball is hit is one factor, but defense seems to be one, too. I just wondered if the A’s IF defense helped or hurt? Clearly, the BB velocity really, really hurt him.
I was thinking all these things too as reading it. I remember looking at pop% as percent of flyballs the other day, and Gray does very well in that regard.
Also, did we just read an article about how problematic the batted ball speed measures are? There is a lot here we don’t know. For instance we know batters control batted ball speed much more so than pitchers and do we have any sort of park adjustments for these yet? I find it very hard to believe that all parks are calibrated all that similarly.
As someone who consciously tried to provoke this article in my comment on the previous article, I want to thank Tony Blengino for writing it. In that light, two and only two points:
1. We all agree that FIP and FIP- are more predictive stats than RA9 or BRef DRA, which is the specific one I invoked in my previous comment. I suspect most or all of us agree that if one is a GM with the choice to invest long-term in either Sale or Gray at the same cost, one should choose Sale. None of this justifies the statement that Sale has had a better year in 2015 than Gray. Even if 100% of the explanation for Gray’s better year is random variation – and maybe it is – it has happened, that’s what BRef DRA measures, and many of us think that’s a better way to vote for Cy Young. I know that position isn’t widely shared on Fangraphs, but I would appreciate commentators understanding that your disagreement with me on this point isn’t an example of you being smart and me being stupid. If on the other hand you do want to take the time to show me what’s being left out of or miscalculated in BRef DRA, though – that I would listen to.
2. I assume – perhaps wrongfully – that everyone understands that while FIP- is extremely predictive, when a player’s RA9 consistently outperforms their FIP- over several years, there is substantial reason to believe that there is a true talent not being measured in FIP-. We know this happens. Therefore, we need to measure how often it happens, and from that develop a working hypothesis concerning the size of the correlation between FIP- and true talent. The correlation is not 1. Is is .9? .5? (My Bayesian prior is in the .7 range.) The point is, this would be worth knowing, and arguments along those lines would be appreciated. Endlessly repeating the same arguments about why not to vote for the leading choices for this year’s Cy Young – that’s less appreciated.
Great article. Made me curious about home/away splits for Sonny Gray. He has a much higher K% and GB%, as well as a much lower HR/9, on the road. In 2014, the K% is pretty similar home and away but GB% and HR/9 trends are the same as 2015. Maybe he pitches to the park?
The White Sox are the worst defensive team in the league but the A’s aren’t much better. Doesn’t seem to account for all of the difference.
I think the philosophy in Oak is strike guys out when you need to, but pitch to contact when you can and work deep in games. I think Gray could strike more guys out if he wanted to, but he would have far fewer complete, and 7+ inning starts. Especially with the bullpen struggles this season, starters were trying to work deep. Also I would be interested to see the contact quality before this end of year blip of rough outings when the team was dead in the water with nothing to play for, as he allowed half his HR on the year during that stretch, as well as a number of high scoring outings.
K/9 goes from 7.24 in low leverage to 8.2 in high. K/BB goes from 2.67 low leverage to 4 high. This is how OAK operates, look at Kazmir comes over with a higher K rate and almost no starts making it into the 6th, then in Oak K and BB rates both drop and he starts pitching 2+ innings per outing with better contact management, and RA then before.
A’s are second to worst in the league in team defense. On the infield Lawrie, Semien, Zobrist have been bad and Valencia merely average. Sogard is good, but that’s it. I’m not sure how Gray’s defense could have been ranked as average, never mind good.
Good article but I have to admit that when I first saw the title I though it said The Anti-Christ Sale lol.