Atlanta Sports Today

Atlanta Sports News Continuously Updated

Player performance is weird: an illustrated example

by Leave a Comment

Putting process, results, and luck in a blender.

Let’s start this post with a Choose-Your-Own-Adventure analogue. You are a head honcho of a major league front office. It’s the offseason after your team made the playoffs in exhilarating fashion, not just to eke into the playoffs, but to coast into them among a wave of superior performances. But, that’s all over, and you find yourself needing to pour your brainpower into the roster construction game once again. Your team happens to have a hole in a corner outfield spot, and you know it needs to be filled externally somehow, because you don’t have anything in the minors quite ready enough to take on the mantle of a lineup regular. There are, of course, lots of choices out there in the offseason. You can outbid other teams on free agents. You can make trades. You can scour the waiver wire. Writing up the full list of choices would be daunting. Let’s crystallize them, for the sake of making one particular, if long-winded (but hopefully entertaining) point.

Let’s condense the choices to two, both of which cost only money. One player is generally thought of as a good one, if potentially risky due to some injuries in his past. He’ll cost money, yes, but not a king’s ransom nor a duration that could hamstring the team later. The other player is going to go for much less of a commitment, but largely because expectations for him are quite modest, without much buy-in into a resurgent prior year for which you are quite grateful, because it aided your team’s quest for the postseason. Here’s how these two players compare, at the time you’re making your decision.


Is this a particularly hard adventure to choose? Well, not really. Player B really only has durability going for him, but he’s older. In every performance respect except perhaps some defensive considerations, Player A comes out on top. We can hem and haw about price and availability and likelihood of signing and all that, but when it comes down to it, the choice seems fairly clear.

Let’s fast-forward through midseason of the next year. How did these two choices do?


Boy, that’s a big gap. That’s a much bigger gap than anything suggested in the prior table. If you followed the basic logic above and chose Player A because, well, he was generally better and also projected to be better (duh), you’ve been rewarded in spades. If you bucked conventional wisdom and chose Player B (for some reason, you weirdo), your insouciance has been punished accordingly, and you’ve now sunk over 350 plate appearances into well, well below average production.

Okay, I apologize, this has been a boring choose-your-own-adventure story. There wasn’t much choice, and the moral at the end was pretty boring: duh, pick the guy that’s probably a good player, not the guy more likely to be a bad player, you weirdo. In case you haven’t figured it out by now, Player A is Michael Brantley, a top 40 position player in baseball so far this year, having his best season since his breakout in 2014 (6.5 fWAR). Meanwhile, Player B is Nick Markakis, signed by the Braves to a cheap deal in lieu of any other option, and scuffling around the sub-1-win mark for most of the year.


If you array every single outfielder that changed teams between 2018 and 2019 (i.e., the ones conceivably “available” because they did, in fact, change teams in the offseason), here’s the list of such outfielders better than Michael Brantley so far in 2019, by fWAR:

  • [null set]

Meanwhile, if you do the same for Nick Markakis, you get:

This also doesn’t include Mike Tauchman and Kevin Pillar, who are tied with Markakis with 0.4 fWAR and have fewer PAs than him so far in 2019.

So, ha ha, you get it. Nick Markakis, what a disaster, right? Should’ve signed Michael Brantley, yeah, yeah, we get it, no need to belabor the point across 700 words.

And yet, and yet… that’s not my point at all! Instead, this is about how tenuous all of the above can be. Not only “can be,” but is. Take a look.

Offense

At first glance, there’s not much to say in trying to bridge the gap between Brantley’s 141 wRC+ and Markakis’ 107 mark. And yet, here we go.

You are hopefully familiar with the Baseball Savant “expected statistics” leaderboard at this point. Here’s a snippet. (Sorry, no column headings.)


A .379 wOBA and a .342 wOBA are big differences (third column from the left). The .355 xwOBA and the .352 xwOBA (second column from the left)? Ehhh, not so much. The rightmost column pertains to the gap between a player’s wOBA and his xwOBA. Nick Markakis has been moderately unlucky, underperforming his xwOBA. Michael Brantley has been quite lucky (top 70ish among the 359 players with 100+ PAs so far this season; 10th among the 70 players with 350+ PAs in 2019).

There is, of course, a bit more nuance to this. Brantley’s home park is a favorable hitting environment that often sprinkles in havoc to wOBA-xwOBA gaps: visiting hitters have, as a group, outperformed their xwOBAs by .018 so far in 2019, and have done so with notable gaps in all but one of the past four-and-a-half seasons. (Overall, Minute Maid Park has always resulted in average xwOBA overperformance, though I like to display visitor-only numbers to filter out any tendencies by home hitters.) SunTrust Park, meanwhile, plays basically neutral with respect to xwOBA-wOBA gaps over its history, and has delivered unto away hitters only an .009 gap so far this season. Brantley is getting lucky, and some of that is the park he plays in (massive .062 xwOBA outperformance at home), but those are his immutable circumstances. Brantley doesn’t really control his xwOBA outperformance nor the dimensions of the park he chose to play in this year, but both have vindicated his signing. Markakis similarly lacks control over these aspects, but they haven’t turned out anywhere near as favorably for him. (Markakis is outperforming his xwOBA by .008 at home.)

We can also look at shifts. For those of us that have cursed the heavens every time Markakis hits a groundball to second, we might think that his tendency to hit the ball into the ground makes him prone to be an xwOBA underperformer in a way that undercuts the “well Brantley is just lucky” argument. But, we can tell from this leaderboard that Markakis is really only shifted 11 percent of time by teams, and the effect is so modest as to be basically unnoticeable over this small sample (.328 wOBA in 41 shifted PAs; .344 wOBA in 324 un-shifted ones). The effect is more pronounced for Brantley in this respect: .344 wOBA in 111 shifted PAs and .383 wOBA in 244 un-shifted PAs. I’m also partial to a different frame of analysis:

  • Nick Markakis — 13 balls in play at “infield level” (launch angle of nine degrees or lower) with the shift on; .237 xwOBA versus 0.067 wOBA
  • Michael Brantley — 49 balls in play at “infield level” with the shift on, .298 xwOBA versus .286 wOBA.

Markakis isn’t really underperforming his xwOBA entirely because of the shift, though it isn’t helping. It’s not helping Brantley either, though it’s not really hurting him, either. In any case, Markakis gets shifted so rarely that you can’t point to them as a reason why his xwOBA undersells him. After all, he’s no Kendrys Morales. a career .040 xwOBA underperformer.

In any case, dice up how you’d like, and you end up with something like this:


Is it surprising to anyone at this point that on Michael Brantley’s Statcast page, Nick Markakis’ 2018 season is listed under similar batters? Or that on Nick Markakis’ Statcast page, Michael Brantley’s 2018 season is listed under similar batters? It really shouldn’t be.

(As a counterpoint, I suppose you could bring up that Brantley’s DRC+ is 12 points higher than Markakis’ mark, at 124 and 112, respectively. Still, there’s much less wiggle room there than the 34-point difference in their wRC+s.)

Brantley was projected to hit better than Markakis, and therefore, he was projected to have better hitting results than Markakis. Brantley has indeed had better hitting results than Nick Markakis. But has he actually hit better than Nick Markakis? Ehhhhhh.

Fielding

Fielding is always a bit cagey, and as we get more data, I’m not sure the fog of war actually clears at all. Note also that a quarter of Brantley’s plate appearances so far have come as a Designated Hitter, for which he suffers a pretty substantial value penalty.


In many ways, this table seems more clear-cut. Michael Brantley has been better at fielding… mostly. The DRS is a half-win swing, enough to account for a full win over the course of the year. The UZR, which is what’s actually used in fWAR, is even larger, nearly three-quarters of a win. The Statcast “jump” metrics similarly agree that Markakis covers less ground than Brantley, due to slightly worse reaction times and routes, though their “burst”-related acceleration to the ball is similar.

And yet, and yet… the Outs Above Average (OAA) and Catch Probability Added (CPA) numbers are interesting. Under these, it is Brantley that comes out below par, and Markakis slightly above it. Contrasted with UZR and DRS, we arrive at one potentially-interesting conclusion. To the best of my knowledge, UZR and DRS do not rigorously adjust for starting position. OAA and CPA do, by definition — that’s how they work. Brantley has the advantage in non-starting-point-adjusted range, and he also seems to have an advantage in moving to get to the ball… yet he actually appears to trail Markakis in recording outs. The perhaps-obvious inference: Brantley’s positioning, combined with his below-average jump, is superior, enough to give him +5 rPM and +1.1 RngR (the range components of DRS and UZR, respectively) despite -3 OAA/-3% CPA. Meanwhile, Markakis’ positioning, combined with his own poor jump, has led to -1 rPM and -3.2 RngR, despite +1 OAA/+1% CPA.

I don’t know to what extent Brantley and Markakis control their own starting positions. Maybe a lot? Maybe not at all? It seems plausible that in this day and age, they’re getting some pretty strict positioning instructions for each batter, though that also may be my own wishful thinking and the actual information they get is more nebulous. In any case, there’s a chance that again, it’s not that the difference between Michael Brantley and Nick Markakis is what’s “failed” the Braves when comparing the two, but rather things in the vortex of, but ultimately beyond, these two respective players. How would Brantley’s -3 OAA translate to DRS if he had the same positioning approach the Braves are giving Markakis? We don’t know, but the above suggests the answer might be “quite poorly.” (The reverse, with Markakis being positioned by the Astros, also applies.)

One place where the data are less equivocal is in the arm scores. Only two outfielders have a worse rARM than Markakis’ -3 (though nine others are tied with Markakis at this integer); Markakis rates eight-worst in ARM. Brantley, by comparison, is pretty quotidian here. Still, a weak throwing arm appears to be the only surefire, keyed-in difference between the two so far. It’s worth some runs, for sure (call it two to five), but the overall performance difference between the two has been over two wins, with a half-win difference in (UZR-based) difference to date, even after accounting for Brantley’s sizable DH penalty.

Where this all comes down to is a pretty expected, generic sentiment at this point: baseball gonna baseball.

All else equal, Michael Brantley was the “good process” choice. (Others, relative to Markakis, also probably represented better process, depending on the player.) Michael Brantley has also been the “good results” choice. This isn’t surprising — it was likely a sucker’s bet that Markakis would outperform Brantley in 2019, except I suppose in the case of a catastrophic-level injury to the latter (which ironically befell not Brantley but Andrew McCutchen, the other interesting and potentially above-average corner outfield candidate…). What is surprising, though, is that while the results have been as expected, the underlying waters are far, far murkier.

Yes, Michael Brantley has had better offensive results. Way better. But have those results been deserved? You could very much make a case for no. It’s a weird world when the guy that should have had better results has indeed had better results, without actually hitting better, but we are definitely in a weird world.

Michael Brantley wasn’t necessarily projected to play better defense (see the Steamer projection way above), but he has anyway. Yet, how much of that has been his own skill, and how much of it has been something beyond his skillset? In this day and age of prescriptive fielding instruction, is positioning a skill? Has Brantley really been a “better” fielder when the OAA and CPA gap between him and Markakis is four outs above average, and four percent of catch probability, in Markakis’ favor?

In some ways, this isn’t too relevant. The Astros are probably more than content with their third-best record in baseball and 7.5-game division lead at the All-Star Break. They’ll probably be in fine shape even if Brantley regresses — they have seven (yes, seven) outfielders with at least 0.4 fWAR this year and a wRC+ of 101 or above. The Braves have been content to ride out Markakis shedding, regaining, and then re-shedding his fWAR as the weeks go by so far, and they’ve clambered to a top-five MLB record and a six-game division lead even while not platooning their right fielder, who currently holds a 58 wRC+ against southpaws. The fear was that the Braves would end up at a place on the win curve where they really needed the extra not-Markakis wins; that fear has not (yet) come to pass, and it seems unlikely that it will. While the Braves don’t actually have much reason to do anything differently at this point — that’s the reality of playoff odds above 90 percent in July — they too are in a position where Markakis has only been their fifth-most valuable outfielder to date. If they haven’t seen fit to downsize his playing time so far, they may as well see fit to just keep running him out there, maybe with a positioning tweak, and see if the wOBA starts catching up to the xwOBA (at least while no one is shifting on him) and the UZR/DRS start catching up to the OAA/CPA over the remaining 71 games of the season.

The best of all worlds is good process and good results. The Braves probably exercised bad process and have gotten bad results, albeit in undeserved fashion. Yet, they’re not suffering for it at all. Maybe that’s the real best, or at least most comforting, of all worlds, one where mistakes and bad moves don’t hurt you in the end.

Reader Interactions

Leave a Reply

Your email address will not be published.