Yeah yeah, we all know how WAR(P) works: You take the number, you take the team wins, you add ‘em together, and then you’ve got your thing. The Rangers won 88 games last year. Yu Darvish, over 150 innings, projects to be worth 2.6 WARP. If everything else washes out—A full season of Cole Hamels, a healthy Derek Holland, the loss of Yovani Gallardo, a little regression in luck, Adrian Beltre’s a little older, but Rougned Odor is, too, and Ian Desmond plays well, while Jurickson Profar contributes somehow, but a little more regression in luck, offset by the improved back end of the bullpen—and you add Darvish’s 2.6 to that 88 and now Bill over in the Standings Department has to invest in a new 9.

But that’s not how WAR(P) actually works, because the (P) is never the same from one situation to the next. The Rangers have as little pitching depth as any team in the game—indeed, Jeff Long just covered this territory a couple days ago:

The Rangers have a lot of somewhat intriguing options for the sixth/seventh starter competition. The only problem is that PECOTA hates all of them.

PECOTA hated all of them last year, too, when a large number of them were called upon to pitch in the absence of Yu Darvish. We know that Yu Darvish improves the Rangers rotation. How much he improves it is our question today.

I took last year’s Rangers starters and added a Yu Darvish start every fifth day. (In a couple instances when the Rangers skipped a starter, Darvish would be inserted on the fourth day, just as rotation stalwarts Yovani Gallardo and Colby Lewis did.) I then removed the starter who would, in my estimation, have been the least likely to be in the rotation at that time. Generally, to stay consistent, I simply followed the principle of last-one-in-first-one-out. I excised pitchers based as much as possible on the Rangers’ signaled intent. I devised a formula for making exceptions, involving average game score over the previous three starts divided by years of experience for each pitcher, but I didn’t need to use it. LOIFOO (and, correspondingly, LOOFOI) ended up working perfectly throughout.

So, for instance, in my estimation, a Rangers rotation that had Yu Darvish to start the season would have gone Darvish-Gallardo-Lewis-Martinez-Holland, rather than Gallardo-Lewis-Detwiler-Martinez-Holland. Then Holland got hurt, so for the second trip through the rotation I assumed a Darvish-Gallardo-Lewis-Martinez-Detwiler rotation, rather than Gallardo-Lewis-Detwiler-Martinez-Ranaudo. So we now have one fewer Detwiler start, one fewer Ranaudo start, and two more Darvish starts.

The Rangers went through a lot of pitchers last year. Not just Darvish but Holland, Martin Perez and Matt Harrison missed long stretches with injuries. Harrison—well, Harrison probably had no business being in the rotation regardless, but the other injuries compounded the damage of Darvish’s absence. Had he been available all season, and started 33 games*, he would have displaced the following pitchers:

Displaced Pitcher

Displaced Starts

Anthony (?) Ranaudo


Wandy Rodriguez


Klein. Joe? Or… Phil? I think Phil? Not sure.


Chi Chi Gonzalez


Matt Harrison


Martin Perez


Ross Detwiler


Okie doke, that’s a lousyish group. How lousyish, you wonder? I took each start that would never have happened (and, for simplicity’s sake, I’m ignoring the butterfly-flap implications of all this), and created an aggregate Not-Darvish Frankenline**:

  • 33 starts, 174 innings (5.3 innings/start)
  • 111 runs (100 earned)
  • 5.74 runs per nine innings
  • 14-19 team record

That guy sucks! Darvish projected last year to have a 3.20 FRA, so over those same 174 innings lets say he would have allowed only 62 runs, or 49 fewer.

*I'm cheating slightly. Cole Hamels pitched the final game of the season, because the Rangers needed it to clinch the division. He threw a complete game and won, allowing two runs. I didn't replace that Hamels start, and it's not reflected in the aggregate line. I'm assuming that the Darvish-fueled Rangers wouldn't have needed that win, because they would have been many games ahead of the Astros, and so Darvish would have been held back to get ready for the first game of the ALDS.

**Yes, I know that Frankenline is the doctor, and Frankenline's Monster is technically the line itself.

But that’s not all. A starter who leaves with one out in the sixth leaves a lot of work for his bullpen; Darvish, who have averaged 6.6 innings per start, takes care of four extra outs each time he starts. The Rangers bullpen averaged 4.34 runs per nine innings allowed. Yes, bullpen innings can be leveraged based on the situation, but we also have to assume that many of these innings, because of the nature of the situation (starter knocked out early) are going to be eaten not by the closer or setup man, but assorted muck in the middle. So let’s call that 4.34 a fair estimate for the non-Darvish relievers who took over for the non-Darvish starters. That’s another five runs that Darvish would have “saved” by his presence. We're getting close to a six-win improvement, relative to Darvish's replacements—and that's assuming only an average Darvish season, not some out-of-this-world career year.

Finally, there’s this: The guys Darvish displaced from the rotation were bad starters. But that’s not to say they would have all been bad relievers. Darvish starting would have bumped one of them at each given time down to the bullpen, where that guy might have actually helped prevent runs.

And finally finally there’s this: With a healthy Yu Darvish, there’s a chance that Chi Chi Gonzalez never gets promoted, and the Rangers never start his service clock. I don’t think that’s all that likely, but it’s plausible, which means that Darvish’s injury might have cost them one full season of club control for their best pitching prospect.

And I guess finally finally finally there’s this: The Rangers lost the ALDS to the Toronto Blue Jays. Game 3 starter Martin Perez likely never takes the hill if there’s a healthy Darvish available; the Rangers lost Game 3. Game 1/5 starter Cole Hamels is likely bumped to Game 2, with a healthy Darvish starting Games 1/5. The Rangers lost Game 5. Would Darvish have made a difference, even in the extremely simplistic no-butterfly-ramifications way of looking at things? Maybe not. The Rangers only scored one run in Game 3, so they probably lose that game regardless, and the difference between Hamels and Darvish is miniscule. But we’re exhausting our finallies, and that should do it.

That was last year, and there’s no guarantee that Darvish will be that good or the rest of the Rangers will be that bad this year. Darvish no longer projects to be a 3.20 pitcher; by DRA, PECOTA sees him as more like a 3.60 pitcher now. He also won’t be ready on Opening Day this year. He might not even be ready until June. But there’s one pretty safe bet regarding his return: When it happens, the Rangers rotation will be ready for him.

You need to be logged in to comment. Login or Subscribe
This is kind of analysis I appreciate. Assuming health anybody who thinks Darvish is only worth 2.6 wins is unable to recognize everything that this article makes abundantly clear. There is a striking similarity between this situation and the Red Sox signing of Price and how it affects each member of the pitching staff.