The Tampa Bay Rays by win probability added: Assigning credit and blame

It's tough to figure out how to feel about the Rays.

They're clearly not a great team at this point, sporting only the ninth-best run differential in baseball, but they are leading their division.

They haven't scored enough runs (third-fewest in the American League), but when you remove context and adjust for park, the offense turns out to be league average (100 wRC+). On the other side of the ball, they own the American League's best ERA, but when you start adjusting that for park, you realize that the rotation (which, to be fair, is missing it's top two starters by preseason projections) has probably not actually been as good as the Yankees, Astros, or Indians (who lead them in both FIP- and xFIP-).

About the only thing you can say with certainty is that this team is resilient and is in the thick of the playoff hunt.

So who gets the credit for the success? Who gets the blame for the failures?

There's a good tool for this. It's called "Win Probability Added" (WPA). It's what FanGraphs uses to make these:

Source: FanGraphs

They're figured by looking at all of the baseball games in the modern era, and seeing how often each team wins when the game passes through a certain base-out state and score.

When something happens in the game (a hit, a strikeout, etc.), the win probability for each team changes, and WPA gives credit to the pitcher and the hitter for the change they brought about. Sounds great, right?

Don't Fall In Love With WPA

There are a lot of things that WPA does poorly. It's context-dependent, so a single with the bases loaded in a tie game gives way more WPA than a home run with the bases empty in a blowout. In fact, a single with the bases loaded in a tie game with two outs will give more WPA to that same single if it came with no outs.

This is a problem if the goal is to evaluate the ability of players, because, while good hitters are more likely to get important hits, clutch hitting is not a repeatable skill, or rather, the difference between the clutch-hitting abilities of major league baseball players are small enough as to not be measurable (and not overly meaningful) in the sample size baseball deals with. For predicting the future performance of a player, we're much better off using context-neutral statistics.

Also, WPA knows nothing about defense, and it doesn't care about position. It gives all credit to a great defensive play to the pitcher, meaning that pitchers will have inflated WPA totals, and good fielders have deflated totals.

The one thing that WPA does well, though, is tell a story (albeit one that ignores defense). It tells us who came though or didn't. It tells us why the Rays won or why they lost. It assigns credit and blame. It's fun.

First we're going to look at the offense. It can be separated into basically three groups. Let's step through from top to bottom

David DeJesus

The 35-year-old outfielder who was almost traded is in a category all of his own. To date he's been worth over a full win above average (1.02 WPA) to the Rays. And the amazing thing is that he hasn't done that by being given especially high-leverage spots or by being clutch. He's just been that good.

Over 133 plate appearances, DeJesus has posted a .333/.383/.504 line, which when converted to a wRC+ shows him to be 57% above average with the bat so far. Sure his .364 BABIP won't last, but there's no getting around the fact that DeJesus has been extraordinary this year (he was pretty good last year, too), and that the Rays were smart/lucky to keep him around.

Here is his most valuable play, a three-run homer against the Marlins worth .248 WPA.

Other Winners

Logan Forsythe is in the midst of an all-star-caliber season, and in addition to that he's also been "clutch," getting his hits at the time they happen most. On the season he's contributed .53 WPA with the bat so far, but also without the bat. His single most valuable play this season came in this game, and was worth .174 wins.

After Forsythe comes a player who's gotten into this "winners section" in a very different manner. While Forsythe has been slow, steady, and consistently good, James Loney has helped the Rays win all at once, without actually playing that well overall. That's because he's been given a few extremely-high-leverage situations and he's come through in those.

There's a stat (WPA/LI) that strips away the advantage players get if they're presented with more important at bats than their peers, and by that metric James Loney has been slightly below average with the bat, just like his 97 wRC+ is slightly below average. But that's not a good story. Loney has come up to bat with the game on the line, and he has gotten those big hits. The result is .44 WPA headlined by two giant WPA-swinging plays.

The biggest, in the 13th inning against the Yankees, worth .367 WPA all on its own:

And this one, pinch-hitting against Toronoto, worth  .331 WPA:

Raise your hand if you thought Brandon Guyer would have a higher WPA at this point of the season than Evan Longoria. Now keep your hand up if you thought that would happen, not due to the context, but because Brandon Guyer has hit better than Evan Longoria. Okay, and now keep those hands up if you thought that would happen even though Longoria bounced most of the way back from his poor offensive 2014. Okay, put those hands down. You're a liar.

The truth is that Brandon Guyer has been very good, hitting 32% above average in 128 plate appearances. That's earned him .44 WPA. His most important play was the first on our lists of most important plays that didn't go down as an RBI. Down one to the Miami Marlins, Guyer lead off against Steve Cishek and doubled to right field.

He did eventually come around to score, but the Rays lost the game in extra innings.

Finally we come to the Rays' real star. As mentioned above, Longoria has hit well, playing nearly every day. His position on the WPA list is not an indictment of his abilities, as he's done his part this season to help the Rays win, with a .34 WPA.

Like Guyer, his biggest contribution of the season also came in a losing effort. It was worth .410 in win probability, but it was about an inch from being a .704 WPA swing.

Below was the sad result of that missing inch.

Source: FanGraphs

I'm avoiding players with very limited playing time on this list, but Jake Elmore deserves a shoutout. He's only been here for 34 plate appearances, but he's made them count, winning .29 of a game above average. With Tim Beckham on the disabled list, we'll all get a chance to see if he can keep the run going (spoiler: probably not).

Finally, that brings us to Steven Souza. Not only has he struck out nearly 40% of the time, the poor guy has been the most unclutch player on the Rays. At the same time, enough good things happen when he does make contact that he's still been a net positive as far as WPA is concerned.

Well, sort of. Before yesterday's game, he was in the negative. Now he's been worth .19 WPA, all because of this .255 WPA jack:

That's about as good an illustration as you can get of why WPA is good for telling stories, and bad for everything else.

But They Play Good Defense

So, the headline of this group is obviously tongue-in-cheek. Per UZR, Kevin Kiermaier has been the most valuable defender in baseball this season. Asdrubal Cabrera has been the sixth-most valuable, and Rene Rivera the 30th-most valuable (and that's not counting the 5.4 runs Baseball Prospectus thinks he's saved with his framing). I'm also fairly confident that Bobby Wilson and Desmond Jennings, when they play, are contributing good value with their gloves.

But there are other guys on the negative WPA list: Joey Butler, Nick Franklin, and Tim Beckham, who's defensive contributions are less certain (sometimes less certain means what it sounds like, and sometimes it's a nice way of saying "poor").

• Joey Butler has hit the ball very well when he's been up, but has been remarkably unclutch, nearly as much as Souza, for a -0.16 WPA.
• Bobby Wilson, unlike Joey Butler, is simply not a very good hitter, and has lost the team just under half a game with his bat so far.
• Desmond Jennings was off to a poor offensive start (-0.55 WPA) over his first 72 plate appearances (a 64 wRC+ that was uncluctchly distributed), before succumbing to an knee injury.
• Nick Franklin gets a pass, as he's only gotten a handful of plate appearances so far, but they've been impressively poor, counting for -0.65 wins.
• Kevin Kiermaier and his -0.69 WPA gets a pass because of his already-mentioned league-leading defense, but he hasn't been good (82 wRC+) and he hasn't been clutch, either. Maybe he shouldn't bat leadoff?
• Tim Beckham has had a few plays that netted him large amounts of WPA, but memorable as they've been, they were the exception, not the rule. Overall, he's cost the team .84 wins with the bat compared to an average expectation.
• And now we get to the true depths. Rene Rivera has posted a full -1.00 WPA. How can it only be one win below average? Well, Rene Rivera has been clutch (defined as performing better when the stakes are higher). Couldn't you tell?
• And rounding out the players who have cost the Rays runs with their bats, Asdrubal Cabrera has been neither good nor clutch. That combines for a -1.50 WPA. That is, in fact, the second-lowest WPA mark in all of baseball, leading only the Giants' Casey McGehee. #ButHePlaysGoodDefense

Conclusion

Win Probability Added is not a tool for analysis. There is no grand conclusion about the Rays playoff chances to be found here. Rather, it's a tool for storytelling. When used correctly, it helps to see the narrative that matches the facts. After all, the goal of the sabermetrically-aware baseball fan is not to ignore narratives, but to latch onto the one that most closely represents reality.

The Rays have been absolutely carried by David DeJesus so far, and have received other good contributions from Logan Forsythe, Brandon Guyer, and Evan Longoria. James Loney hasn't been especially great, but his hits have come at the right time for the Rays, and that's translated to wins. Meanwhile, Steven Souza Jr. has gotten most of his hits at relatively unimportant times, but power matters, and he has enough of it to help his team even when he's not being especially "clutch."

On the other side of the equation are a bunch of glove-first starters* and bench players who have contributed with the bat about as poorly as expected. The ire thrown towards Rene Rivera and Asdrubal Cabrera by Rays fans is deserved, as those two have cost the team wins at the plate at an alarming rate (although both of them provide value in other ways).

*And yes, we all thought Cabrera would be a decent bat and a below-average glove, so who knows anything about anything.

Will the Rays make the playoffs? Win Probability Added doesn't know. Use a better stat if that's what you want to find out. But what it does tell you is how they got to where they are now. Now go forth and assign your credit and/or blame accurately.