Nick Markakis and the Terrible, Horrible, No Good, Very Bad Game

Nick Markakis didn't necessarily make history last night, but that doesn't mean he didn't accomplish something highly notable. However, his accomplishment was a rather ignoble feat: his -0.609 WPA mark for the game is, by far, the worst such mark by a Braves batter this year. In fact, it's the second-worst single-game mark by a Brave this year (more on that later), and it's much closer to some of the worst performances all time from a batter in a single game than it is to the way an average "bad day at the plate" transpires.

Before I get ahead of myself, let's lay out some basics. WPA stands for "win probability added," and refers to a player's impact on the probability of his team winning a particular game. Essentially, every time the player takes part in a plate appearance, the game state changes - the batter either reaches base or makes an out, and that increases/decreases his team's chances of scoring a run, and thus winning the game, given the current score. WPA is somewhat similar and somewhat distinct from WAR - WPA measures what actually happened based on the score, inning, runners on base situation, and so forth, while WAR measures how valuable a player's results are, assuming that those results generally occur in average-ish situations.

(Side note: WPA is not really predictive but it's a very interesting and apt descriptive statistic. For example, the Braves' bullpen currently has a WPA of about -2.4 total wins, which means that the bullpen, in terms of actual runs allowed and when they allowed them cost the team about two-and-a-half wins. Meanwhile, an average bullpen has added about 0.6-0.7 wins to its team so far, meaning that in aggregate, the Braves bullpen has really cost the club around 3 wins relative to an average bullpen. But I digress, though hopefully that digression is illustrative of either the overall or its own separate point.)

The way WPA works, in practice, is that each team starts out at a 50 percent chance of winning the game (WPA does not take talent level of the teams into account, it cares only on base-out states and the historical chances of scoring runs in those states, averaged across all teams/talent levels/etc., as well as the chances of winning/losing a game given a certain score and base-out state in a certain inning of the game). Then, as stuff happens, each player's individual play WPA increases or decreases his team's chance of winning. By the end of the game, one team's players will have amassed -0.500 WPA, and another team's players will have amassed +0.500 WPA, because by the end of the game, one team's win expectancy is 100%, and the other team's is 0%. Similarly, the positive WPA earned by one team's pitchers is necessarily mirrored by the negative WPA earned by the other team's batters, and vice versa. The reason for this is hopefully intuitive - batters and pitchers are involved in opposite sides of the same play, and the pitcher is credited/debited for each player in the same way a given hitter is. (Alas, defense and baserunning are not included in WPA. That's what WAR attempts to do.)

So now that we've got that out of the way, here's the kicker: Nick Markakis posted a -0.609 WPA last night. Given that, in order to lose a game, a team, in total, only has to amass -0.500 WPA, you can see why this is kind of a big deal. That means that not only did Nick Markakis, by one measure, kinda sorta single-handedly lose the game, he actually negated the positive contributions towards winning the game from the rest of his teammates. Specifically, the rest of the Braves actually put up a +0.109 WPA for the game, meaning that aside from Markakis, the Braves actually boosted their chances of winning the game from the default 50 percent state up to about 60 percent. It was on the back of Nick's bat that those odds came crashing down.

Specifically, here's how Nick Markakis' terrible, horrible, no good, very bad game proceeded:

With no score and none out in the top of the second, the win probability was knotted at 50-50. Markakis flew out to left, giving the Mets about a two percent boost of probability in winning the game (because Markakis made an out that the Mets had not yet matched). This is basically nothing in the grand scheme of things - it was an out in a tie game, and early on at that.
With one out, down 1-0 in the fourth, Markakis again flew out with Freddie Freeman on first. Given that the Braves were trailing already and there was a runner on base, this hurt a bit more than the earlier fly out: it decreased the Braves' win probability by about 3.5 percent.
Nick then came to bat one inning later, with the bases loaded and two outs, with the Braves again down by one. His weak tapper back to the pitcher cost him 0.110 points of WPA, which is another way of saying he knocked the Braves' chances of winning the game down by 11 percent. That's kind of a big deal (more on that in a bit).
Things would only get worse from there. In the 7th, with the Braves again down, this time by two runs, (the score had changed each time throughout his at-bats) Markakis hit into an obvious double play. While this double play scored a run, scoring one run on a double play in that situation and trimming the lead to one is actually a detriment to a team's chances of winning relative to an average outcome in that situation. This cost Nick and the team about 18 points of win probability. Ouch.
The real kicker came as the Braves battled back in the 9th. With the bases loaded and one out this time, and the deficit at two runs, Markakis hit into a game-ending double play which cost the Braves their remaining 26 percent of win probability, and closed the book on the June 12 game for the club.

In total, Markakis had his win probability stand at -0.609. The guys ahead of him, Jace Peterson, Cameron Maybin, and Freddie Freeman, combined for a 0.527 win probability added on the night. Markakis erased all that and more.

To see whether this was significant, I trolled through the archives of WPA for the games this season, as well as for players. The worst game by Braves regulars in WPA terms (for offensive players) tends to be around -0.150 to -0.175ish, depending on the player. That's pretty reasonable and normal: after all, losing is usually a team effort. It's a little different with pitchers, who can either allow a lot of runs as starters to put the game out of reach, or give up a key hit in the late innings that results in a likely--insurmountable deficit. But no Braves hitter even came close to approaching the -0.609 that Markakis posted last night - I think the highest I saw for a hitter was around -0.200.

There are two further remarkable things about Markakis' performance. First, it was closer than you'd think to some of the worst performances, by WPA, all time. Unfortunately, I don't know how to search this as some kind of big database (which also explains the fuzzy, non-definite numbers used above), but the worst I can find are: Juan Rivera in 2003, and old friend Bob Wickman in 2006. I don't have a subscription to Play Index, but I can tell that there have only been 92 hitters since 2000 to put up a worse WPA than Markakis in a single game. I'm not sure any of them played for the Braves. (If someone has Play Index, it's a very easy query to run and I can update the post with the results.)

What I did find was that this feat was only the second-worst single-game performance by the Braves this season. The worst was, of course, in both statistical and figurative senses, Grilli's allowance of a homer to Dan Uggla. WPA doesn't take into account how much money the Braves are paying an opposing team's player when he makes them lose a game, but maybe it should. In any case, a two-run lead in the 9th is about a 94 percent chance of winning when the game starts; by the time Grilli's outing was over, the Braves had only a 16 percent chance (soon to be a zero chance, at that).

Other outings where players basically gave the game away themselves include two other sorry bullpen outings: Cody Martin allowing 2 runs in the 8th with a 1-run lead to the Nats on May 10 (Zimmerman RBI single, Ramos RBI double), and Jim Johnson allowing a go-ahead 2-run homer to Josh Donaldson in the 8th while the Braves had been up by a run.

So here's to you, Nick Markakis. Hopefully we never see another performance this bad in 2015, or ever again. For the record, Markakis has been fairly solid as a Brave so far: his 113 wRC+ is propped up by a high-ish BABIP but he's also walking a lot, and it's kind of impressive he can maintain a offensive metric that high with a 0.066 ISO. His defense has been better than I expected: he's been average to below average in RF, both in terms of placement among his RF peers as well as the actual metrics on an absolute value basis. He's on pace for around a 2-win season overall, which would make his 2015 a bargain by $/WAR standards (no word on his 2016 or later years), and aside from this game, he really has not dragged the team down.