Numbers say pull Kershaw after six

Lefty has 3.31 ERA through sixth inning in playoffs, 18.25 in late frames

October 12th, 2017

There's an alternate version of the postseason story, one in which he has allowed a 3.31 ERA in October, not a bloated 4.63 mark. In that storyline, he has more successful starts than forgettable ones, and the ongoing question of "Can he pitch in the playoffs?" isn't ever-present.

NLCS Game 1: Saturday, 5 p.m. PT on TBS

In that other universe, Kershaw gave up two runs to the Cardinals in Game 1 of the 2014 National League Division Series, not eight. He allowed two homers to the D-backs in this season's NLDS Game 1, not four. He walked only one Met and allowed a single run to score in the first game of the 2015 NLDS, not four and three, respectively. He allowed two Nationals to score in Game 4 of last season's NLDS, not five.

Dress for NLCS: Get Dodgers postseason gear

This may sound like completely rewriting history, but all it would have taken is one simple change. In this version of Kershaw's postseason story, he is never asked to pitch into the seventh inning; the Dodgers have stronger bullpens and deeper rotations, and so Kershaw is never extended past his limits. He's out when it's time to be out.

:: NLCS schedule and coverage ::

That's obviously not what has happened in reality, of course. But it does present an interesting question for manager Dave Roberts and the Dodgers as they prepare to send Kershaw to the mound in the first game of the NL Championship Series presented by Camping World on Saturday night. Can a second-year manager tell baseball's best pitcher that enough is enough, even when Kershaw doesn't agree? We've seen Roberts yank Rich Hill in the midst of a perfect game; we've seen him pull with a no-hitter going. Neither pitcher has the stature of Kershaw.

There's evidence that Roberts is thinking about it, anyway.

"With the strength of our 'pen," Roberts said after Game 1 of the NLDS, "there's nothing in me that says Clayton needs to be pushed, and he needs to stay in until the game goes the other way."

So can Roberts do it? Here are two compelling reasons he ought to strongly consider it.

Because the 7th really has been a trouble spot

Kershaw, as we said, has put up a 3.31 ERA in the first six innings of postseason games in his career. After that, it's 18.25. Some of that isn't entirely his fault -- , for example, was on the mound in 2015 when knocked in two runs that were charged to Kershaw -- but there's no getting around that the issues are real.

For example, let's look at our most advanced Statcast™ metric, Expected wOBA, which tells a very strong story. It combines quality of contact, based on exit velocity and launch angle, along with real-world strikeout and walk totals, to give a good estimation of a pitcher's skill. In this measurement, a pitcher gets more credit for a poorly hit bloop single that fell in front of a slow defender than he does for a crushed line drive that is caught for an out. (wOBA is like OBP, except more credit is given for extra-base hits, rather than all times on base being treated equally.)

From 2015-17, the three years for which we have data, the Kershaw story is clear. The seventh inning (and beyond) in October is trouble.

Kershaw Expected wOBA, since 2015

Regular season, innings 1-6:   .233

Regular season, innings 7+:    .275

Postseason, innings 1-6:         .244

Postseason, innings 7+:          .471

For some context, the 2017 Major League average wOBA is .327. had the season's lowest starting pitcher wOBA, at .244, so Kershaw is as good as they come in the first six innings, in the regular season and postseason. In the seventh inning and later during the season, he declines slightly, to a still very good .275. In October, Kershaw's .471 number is so massive that it basically means that every hitter he's faced is equivalent to (.472). 

Why? While we're focusing on the seventh here, it's similar to the "third time through the order" effect we discuss so often. As pitchers work deeper into games, they fatigue, which can mean either a drop in velocity or missing their spots, and opposing offenses gain familiarity and begin to pinch-hit, increasing platoon advantages. Plus, obviously, playoff lineups are more dangerous than most regular-season versions.

In Game 1 of the NLDS against Arizona, it was clear that Kershaw's velocity was on a downward trend.

When Kershaw appeared in the seventh inning, he allowed each of the three hitters he faced to make contact. All three (Adam Rosales, 102.1 mph lineout; , 110.9 mph homer; Jeff Mathis, 97.6 mph homer) hit the ball extremely hard. It was the third time through the lineup. Two batters before, J.D. Martinez had taken him deep, as well.

While a fair retort would be that some elite postseason starters like can pitch effectively without such issues, that isn't really Roberts' concern here. The game has changed to move away from long postseason appearances from starters, and Kershaw has missed time in both 2016 and '17 with back injuries. By the time he gets that far into a game, it's not fair to expect him -- or any starter -- to be as effective as they were in the first.

Because the bullpen is more equipped to handle it

As Roberts noted, the strength of the bullpen has been an issue in past postseasons, aside from . In 2013, the Dodgers' bullpen featured and Brian Wilson. In '14, it was Baez, Brandon League and . That's not exactly the '17 Yankees of , , , Tommy Kahnle and .

But it's different now. Jansen has pitched more than one inning in a postseason six times across 2016 and '17, after having done so just once before. has emerged as a late-inning weapon, striking out 50 in 43 2/3 innings in the regular season; Tony Cingrani turned his 5.40 ERA with Cincinnati into a 2.79 mark with the Dodgers, in part because he went from throwing his slider three percent of the time to 25 percent of the time.

Former starter has emerged as an unexpected asset against righties, striking out four of the six D-backs he saw in the NLDS. By upping usage of his cutter and slider from 32 percent at 84.9 mph in the regular season, to 57 percent at 86.3 mph in the postseason, he's suddenly given Roberts a very appealing option out of the bullpen, and that's the point. Roberts has options, in a way that Don Mattingly and Joe Torre before him often did not.

It's not that Kershaw can't do it, of course. Just last year, he threw seven scoreless innings in a 1-0 win over the Cubs in Game 2 of the NLCS. The season before, his second look at the Mets ended with eight strikeouts and one run allowed over seven innings in a 3-1 NLDS Game 4 win. Kershaw can do it, because he's the best pitcher of a generation. It's just not the way the game works these days -- and it's not the way Roberts or Kershaw need it to work.