It’s that time of year again. That last week in July when we swear that teams lose all sense of things big and small and ship otherwise valuable prospects in exchange for late-inning relievers who will pitch a few dozen innings over the balance of the season. It’s a formula that the sabermetric community sometimes finds difficult to rationalize. Relievers pitch so few innings and are so volatile that their value is almost certainly lower than that of the prospects dealt for them.

It’s difficult to look at the trade returns for late-game specialists and understand the thought process. The Cubs seemingly traded a king’s ransom to acquire Aroldis Chapman, a pitcher whose performance is only marred by the domestic violence charges that hang over him. Let’s not mince words either; that marring very real and deserved. For the purposes of this article, though, we're be ignoring that component of this trade, not because it doesn’t matter—it matters immensely—but because it didn’t dramatically diminish his value in the baseball world, which is what this article is about.

This most recent move joins a list of others that shook our understanding of reliever value in recent seasons:

Prospect rank on the BP 101 at time of trade in parentheses. Asterisk indicates midseason top 50 rank where applicable.

Now more than ever, baseball seems to appreciate the immense value produced by team-controlled assets. Make no mistake, that’s how they’re viewed. Players like Manny Machado or Kris Bryant produce tens of millions of dollars worth of surplus value simply because they're forbidden from negotiating their compensation. The result has been a prospect-laden gold rush, with teams rushing to hoard talented young players. This philosophy, in some ways born out of a ruthlessly efficient sabermetric approach to maximizing surplus value, has created a perception that trades like those mentioned above are absurd from a value standpoint.

There’s a simple rebuttal to this: It's correct! There’s no arguing with the claim that six years of Torres, McKinney, and Crawford will likely produce more surplus value than three-plus months of Aroldis Chapman at a prorated $4.3 million and change.

There’s a problem, however. The Cubs are not acquiring Chapman for his ability to accumulate WARP, nor are they acquiring him to cover their asses should their best efforts go un-rewarded come November. The Cubs are acquiring Chapman because World Series Championships occur in the real world, and the real world largely doesn’t give a damn about WARP.


In an effort to quantify in a rational way the impact a player makes, significant strides have been taken to remove context as much as possible. This allows us to isolate performance and produce valuable statistics like TAv or WARP or any other of the dozens of stats you can pull from this very website. These stats tell us who the best players are, and by how much. They are the basis of player evaluation. In some cases—DRA for example—we’ve done well to incorporate both context-neutral and context-specific data, providing a clearer picture of performance and value in the real world.

Some effort to address this is incorporated of course. A WAR model might incorporate leverage, or a writer might pair WARP and leverage to explain an MVP vote, but at the extremes—and these elite relievers do represent the extremes—value can be missed.

Pitchers like Chapman, Giles, Kimbrel, and Miller are tasked with locking down the late innings of games. Often these situations are extremely close, and therefore high-leverage. Here is where the aggregation of statistics and the real world come crashing together.


On June 5th, Aroldis Chapman was called on to pitch the eighth inning of a game against the Baltimore Orioles. At the time, the Yankees were a few games back of the Orioles, Red Sox, and Blue Jays in the AL East, making this intra-division matchup an important one.

Chapman inherited runners on first and second with just one out from the usually dominant Dellin Betances. Chapman would make easy work of the first batter he faced in Jonathan Schoop, striking him out.

  • +0.48 runs are credited to Chapman’s ledger.

The next batter was Francisco Pena, who has 42 career major-league PAs. He would single to right, loading the bases.

  • -0.34 runs are debited from Chapman.

Matt Wieters would pinch-hit for Ryan Flaherty and promptly single up the middle. Because of an error from Jacoby Ellsbury, Wieters’ hit would clear the bases.

  • -2.59 runs are debited from Chapman.

The next batter, Adam Jones, would go down swinging to end the inning.

  • +0.37 runs are credited to Chapman.

Overall, Chapman got hit for about two runs, according to RE24, a stat that simply tracks the changes in base-out state during a pitcher’s time on the mound. That’s a blow, but only a small portion of a total win, the currency baseball value is so often traded in.

Let’s review that same portion of the eighth inning, this time using WPA, a measure that tries to assess the importance or leverage of a particular event in terms of game outcome:

  • Schoop strikes out. +.12 WPA to Chapman.
  • Pena singles, loads bases. -0.076 WPA to Chapman.
  • Wieters singles, scores 3. -.576 WPA to Chapman.
  • Jones strikes out to end the inning. +0.012 WPA to Chapman.

Overall, Chapman gets hit with -.52 WPA, roughly equivalent to five runs, or half a win. That’s a much bigger hit on his value.

The reality, of course, is that the Yankees lost the game. The whole game was lost. Not half a game, or 20 percent of a game. The whole thing. They were four outs away from winning a whole game, but instead they lost a whole game.***

The repetition seems silly, but it’s important. Managers and major-league front offices can’t always traffic in parts or shares of wins. They can’t always look at aggregated totals and claim to have done a good job. In the end they either win the game or lose it. This, of course, is doubly true in the playoffs, where the ultimate measurement is the whole series—which can, similarly, be lost on Wieters single, scoring three.

Theo Epstein and the Chicago Cubs are in the business of real world wins and losses. Most of the time they can gleefully work in aggregate performance, applying seemingly patented Ivy League rigor to this child’s game. Sometimes though, they have to face the reality that the game imposes: At the end of the day they’re charged with winning a World Series for a franchise that hasn’t done so in more than 100 years.

It’s times like this the Cubs need to work in hypotheticals. Would Hector Rondon close out all the same games as Chapman down the stretch and in the playoffs? Maybe. Did the Cubs lose the surplus value battle in trading a handful of talented, cost-controlled assets for half a season of a relief pitcher? Probably. At the end of the day though, the Cubs have spent hundreds of millions of dollars constructing a team with one sole purpose: To win the last baseball game this season.

When you put it that way, relying on a pitcher who will ‘maybe’ be as good as Chapman when it really counts seems kind of crazy doesn’t it?

Thank you for reading

This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.

Subscribe now
You need to be logged in to comment. Login or Subscribe
Maybe WARP and other artificial-win-based stats aren't accurately measuring real-world-wins in later game situations.

Maybe leverage in a generic early game situation - where data is simply aggregated leaguewide and relationships of events to wins is linear - is not the same as specific late game or late season situations. Where much of the bullpen/roster isn't available cuz they've been used. Where some teams are predictably playing the 162 and others are aiming higher. And where a real-world win for some teams can have, generically, exponential value and not just linear value.

Context does matter in baseball. And it doesn't seem to me to be a coincidence that the two situations where it seems to matter most - fielding and late-game - are the two areas where saber struggles the most
I thought it was interesting that Theo acknowledged giving up giving up a ton for Chapman, but then said if not now when?...
Well presented. No disagreement here.

Just to play devil's advocate, though, might it also be considered kind of crazy to give up several years of cost-controlled goodness for a guy who will 'maybe' be better than Rondon when it really counts? Especially when the acquisition provides no marginal benefit for the next two months (and will be of value only for this one post-season)?
It might!