Zach Britton Wasn’t Even the AL’s Best Reliever

Zach Britton has recently found himself at the forefront of baseball consciousness for a lot reasons, mostly positive, some negative, albeit through no fault of his own. He had a supremely excellent season in a tightly tailored, typical closer’s role for the Orioles, and his non-usage in last week’s wild-card game has almost become a caricature, a metaphor for outdated laissez-faire managerial strategies.

He will certainly receive many, many Cy Young votes, and might even walk off with the award. In my piece here last week, I compared his 2016 performance to some of the top seasons produced by AL starters, utilizing granular batted-ball data, and found that, while Britton does at least belong in the conversation, he didn’t deliver as much production to his club, in a year that is admittedly without a runaway choice among starting pitchers. What if I told you, however, that Britton didn’t even have the best season among AL relief pitchers?

I’m not going to repeat my subjective analysis of Britton in last week’s piece. Yes, what he did was historic. And just his 80% grounder rate, alone: wow. I mean, wow. He’s been really, really good for three years now, after a brief, unsuccessful run as an MLB starter, and took things to another level altogether this season. His 0.54 ERA was crazy low, but there’s quite a bit of good fortune in that number. His 1.94 FIP is much more reflective of his actual performance; his “Tru” ERA, which takes into consideration the exit speed and launch angle of every batted ball he allowed, is in that range at 1.65.

Britton was great. But if his name is part of the award conversation, Andrew Miller’s has to be, as well. It’s been a crazy road to this point for the consensus top talent and former sixth-overall pick from the 2006 draft. As a prospect, the Tigers raced him to the big leagues, but he battled injuries and mechanical issues for the better part of the next five years as a starting pitcher. He was sent first to the Marlins in the Miguel Cabrera deal and then to the Red Sox in a minor deal following an 8.54 ERA, nearly two-and-a-half-base-runner-per-inning abomination in 2010. He racked up little more than service time to that point, before finding himself in Boston.

When he became a free agent following a 2014 season that he concluded with the Orioles — with which club he was Britton’s teammate — he had all of one career save. He signed with the Yankees that offseason and was utterly dominant in 2015, working in a fairly traditional role and earning 36 saves. The Yanks added Aroldis Chapman in 2016, and proceeded to deploy a three-headed bullpen monster in the first half of this season, with Chapman typically finishing games. Miller went to the Indians at the trading deadline, and here we are.

Miller hasn’t been used as the primary “closer” in Cleveland, either, though he has clearly been their best reliever. We got a glimpse of the Indians’ vision for the big lefty in Game 1 of the ALDS, summoning him in the fifth inning and letting him hang around for a while. It isn’t about “save” situations; it’s about “game” situations, and efficient utilization of resources.

How do the campaigns of these two relief aces match up? Let’s explore both pitchers’ plate appearance frequency and production by BIP type data to draw some conclusions. First, the frequency data:

Plate Appearance Outcome Frequencies, 2016
Miller % REL PCT
K 44.7% 212 99
BB 3.3% 40 3
POP 1.4% 43 9
FLY 27.2% 87 16
LD 17.1% 83 4
GB 54.3% 121 94
Britton % REL PCT
K 29.1% 138 99
BB 7.1% 87 58
POP 0.0% 0 1
FLY 8.8% 28 1
LD 11.3% 55 1
GB 80.0% 179 99

It must be noted that the percentile ranks above are expressed relative to the pool of ERA-qualifying AL starters. Percentile ranks don’t fully reflect the peaks in either pitcher’s profile. Both possess K rates in the 99th percentile of the starter population, but Miller’s otherworldly 44.7% K rate dwarfs Britton’s 29.1%. It also underplays Britton’s incredible grounder rate: his 80% figure obviously is in the 99th percentile, while Miller’s relatively unspectacular, but still awesome, 54.3% grounder rate places him in the 94th.

Neither pitcher allowed many line drives, with Miller (17.1% liner rate) occupying the fourth percentile and Britton (11.1%) the first. Liner rates are the one BIP type that doesn’t correlate well from year to year for most pitchers, but it’s fair to say that it’s quite difficult to square up the baseball against either of these guys. One notable difference between the two that shouldn’t be overlooked is Miller’s walk-rate advantage. It certainly appears that Miller has left any mechanical inconsistencies far behind; that lean, 6-foot-7 frame is purring like a kitten. His BB rate sits in the third percentile; Britton’s is actually below league average, in the 58th.

Frequencies only tell us part of the story. The production by BIP type data is a solid proxy for authority allowed, the main remaining missing piece of the puzzle:

Adjusted Production by BIP Type, 2016
FLY 0.444 1.167 178 98
LD 0.667 1.042 120 85
GB 0.148 0.148 36 97
ALL BIP 0.298 0.546 93 81
ALL PA 0.159 0.187 0.292 40 35 1.45 1.67 1.68 1.47
FLY 0.313 0.688 72 102
LD 0.700 0.850 105 107
GB 0.235 0.296 110 124
ALL BIP 0.235 0.296 40 45
ALL PA 0.161 0.220 0.203 36 40 0.54 1.50 1.94 1.65

The actual production allowed on each BIP type is indicated in the batting average (AVG) and slugging (SLG) columns, and is converted to run values and compared to MLB average in the REL PRD (or Unadjusted Contact Score) column. That figure is then adjusted for context, such as home park, team defense, luck, etc., in the ADJ PRD (or Adjusted Contact Score) column. For the purposes of this exercise, sacrifice hits (SH) and flies (SF) are included as outs and hit by pitches (HBP) are excluded from the on-base percentage (OBP) calculation.

Miller didn’t allow many fly balls, but he allowed quite a bit of damage (.444 AVG-1.167 SLG, 178 Unadjusted Contact Score) on the ones he did. That’s a bit misleading, however, once you adjust for exit-speed/launch-angle allowed. His Adjusted Contact Score drops to an almost exactly league average 98. What gives? Well, Miller allowed six 100-plus mph fly balls this season. All six left the yard. As a frame of reference, 60% of such fly balls left the park in both leagues combined this season.

Miller was also a bit unlucky on liners, allowing a .667 AVG-1.042 SLG (120 Unadjusted Contact Score). Once adjusted for context, his Adjusted Contact Score drops to 85. He did get a bit lucky on the ground, with a 36 Unadjusted Contact Score that rises to 81 once context is taken into consideration. On all BIP types, Miller posted a 93 Unadjusted and 81 Adjusted Contact Score. Add back the K and BB, and his “Tru” ERA of 1.47 also exactly duplicates his ERA, and is better than his FIP. The big lefty is a solid contact manager who is raised to the inner-circle elite level by his high-impact K/BB skills.

As indicated in last week’s article, there is a dynamic at work in Britton’s profile that renders his BIP type-specific Contact Scores somewhat meaningless. In 2016, just over 14% of all balls in play didn’t register speed/angle data in 2016. Typically, this is the result of weakly hit balls hit at very high and very low launch angles. For Britton, however, over 36% of the batted balls he allowed didn’t register — the vast majority of them weakly hit ground balls. Batters went 2-for-59 on these batted balls. His overall 45 Adjusted Contact score takes those unregistered BIP into consideration and accurately reflects his exceptional contact-management skill. Just over 9% of BIP allowed by Miller were unregistered, so his BIP-type specific Adjusted Contact Scores are representative of his above-average, though not elite, contact-management ability.

Adding back the K and BB to Britton’s line, and his “Tru” ERA is actually a bit higher than Miller’s at 1.65. This is much closer to his FIP than to his unrealistically low ERA of 0.54. Britton allowed three 100-plus mph fly balls this season, and only one of them left the yard. When you are talking sample sizes this small, a couple extra or fewer well-hit balls going over the fence can be the difference between two excellent pitchers.

So what does all of this mean with regard to overall value? To measure each hurler’s “pitching runs” for the season, you simply multiply their innings pitched totals by the difference between their “Tru” ERA and the league average, per nine innings. Miller pitched more innings than Britton and, as we concluded just above, was slightly more effective on a per-inning basis. Using this method, Miller contributed 22 pitching runs to Britton’s 19, behind the triumvirate of Chris Sale, Justin Verlander and Corey Kluber, all of whom round to 26 pitching runs, in that order of finish.

Lastly, let’s visit the concept of leverage. Backers of Britton like to cite his advantage in this area when comparing him to these and other starting pitchers. Won-lost records of traditionally deployed closers are usually and rightly disregarded, as the only way for them to earn a win is to blow a lead. Britton only had three decisions all year (2-1), as he was used in a tightly restrictive manners, entering only seven tie games, and pitching over one inning only seven times all season.

On the other hand, Miller entered 17 tie games this season, and pitched over one inning eight times with Cleveland alone, in just the last two months of the season. His 10-1 won-lost record is far from meaningless, but rather a direct reflection of games in which he outpitched his opposition with the game on the line.

There were two historic pitching performances turned in by relievers in the American League this season. With regard both to quantity and quality, Andrew Miller nosed out Zach Britton, only to likely be ignored in the Cy Young Award voting. With a little luck, Miller might be able to “settle” for some team-oriented hardware as the Indians move through October.

newest oldest most voted

When you mentioned leverage, why did you actually quite win-loss record instead of leverage?


I only bring this up because in your last (awesome) article, you completely left off leverage, and I and others questioned that. Now you pay it lip service before discussing win-loss record instead, never once actually using leverage. I think leverage is absolutely relevant to any discussion of relief pitcher season value, much more so than win-loss record.

Since you write these articles, I know that you are capable of properly using the statistics, so your reticence to use leverage must be because you don’t like leverage as a measure. What issues do you have with it? Do you think it doesn’t measure what it purports to? That it’s misleading in some way? I would love to hear your thoughts on this subject.


I am afraid the answer is most likely: because it does not fit the narrative.

Any stat that tries to take into account leverage will favour Britton over Miller this year, whether it is WPA, SD/MD or Clutch, they all say Britton has been better in high leverage situations.


I agree W/L is pretty useless for relief pitchers. I think leverage is interesting but not a good way to measure the talent of a player. pitchers don’t get to choose the situations they pitch in and batters can’t fully control when they do/don’t get hits. it’s a contextual statistic, like the RBI stat.