Kershaw Is Forcing Us to Confront the Pedro Question

Over the last 365 days, Clayton Kershaw has been baseball’s best pitcher. That isn’t a particularly enlightening sentence given that he’s almost certainly been the league’s best pitcher over the last several years, as well. At this point, the question isn’t really if Kershaw is the best pitcher, but rather if he is the best overall player, Mike Trout included. Kershaw has truly been that phenomenal.

To put some numbers behind it, consider: since May 26, 2015, Kershaw has thrown 253.1 regular-season innings (34 starts) and produced a 39 ERA- and 42 FIP-, thanks in part to a 34.8 K% and 3.3 BB%. By our WAR model, that’s equivalent to an 11-WAR season. It’s closer to 12 WAR if you use runs allowed as the primary input instead of FIP.

We all have our own favorite Kershaw fun fact, but here’s one that’s been bubbling to the surface lately. Full disclosure: I’ve been partially responsible for said bubbling.

Pedro vs. Kershaw
Player Time IP ERA- FIP-
Pedro Martinez 1999-2000 430.1 39 39
Clayton Kershaw Last 365 days 253.1 39 42

Some context: since 1961, there have been just a handful of qualified starters to record less than a 40 ERA- in a single season and the only two qualified seasons under 40 FIP- belong to Pedro in 1999 and Kershaw in 2016. Those Pedro years are often considered the modern gold standard of starting-pitcher dominance. He was 60% better than league average for two full seasons.

It’s important to note that Kershaw is only halfway into the chase. Not only does he need another 200 innings in total to match Pedro, but his 2016 season is fewer than 100-innings old. Lots of pitchers have been great for 100-inning stretches without warranting the Pedro Article, but Kershaw is the game’s best pitcher and he’s somehow improving.

I’m not terribly interested in a clickbait-style “which guy is better” debate. Pitching quality isn’t something we can measure with the kind of precision necessary for that to be a useful goal. Even if we determined which pitcher was better, what good would that even do? Instead, I want to pose a broader and less combative question: is this Kershaw run similar to peak Pedro?

Right off the bat, ERA- and FIP- say yes. Comparing his park-adjusted ERA and FIP to the league average ERA and FIP over the last year, Kershaw is right there with peak Pedro. He hasn’t maintained it as long as Pedro did, but he’s 59% of the way to matching the Peak of Peaks.

On the other hand, there are metrics by which Pedro is clearly ahead. Kershaw is averaging an RE24 of 53.2 per 200 IP while Pedro averaged 68.8 per 200 IP from 1999-2000. While their park- and era-adjusted ERA are the same, Pedro’s run-expectancy numbers are better, indicating that Kershaw is probably getting a little extra help from his bullpen when he leaves the game with men on base (their unearned runs are comparable).

Furthermore, Baseball Prospectus’ Deserved Run Average favors Pedro a little more than ERA- and FIP-:

Pedro vs Kershaw
Player Year DRA-*
Pedro Martinez 1999 31
2000 36
Clayton Kershaw 2015 50
2016 42
*The DRA- values presented here were provided by BP’s Jonathan Judge and are different from the DRA- published at BP which, due to the way they scale the numbers, is not perfectly analogous to ERA-/FIP-.

There are plenty of reasons why FIP and DRA would diverge. Perhaps DRA believes Kershaw’s low BABIP is based more heavily on good defense while Pedro’s is based on Pedro being awesome. Maybe DRA thinks the quality of Kershaw’s opponents during this run is lower than Pedro’s, even once you get inside the era itself. After all, beating the 2016 Padres is different than beating the 2016 Cubs, even though both happened in the same year.

There’s a lot of nuance inside DRA that we can’t unpack without looking under the hood, but I would suspect those would be the two critical areas.

Kershaw’s unadjusted ERA, FIP, and DRA are better than Pedro’s, but in order to compare them we’ve taken to scaling each player’s numbers to that era’s league average. During Pedro’s peak, teams averaged about 5.1 runs per game compared to around 4.3 runs during the Kershaw streak. What we’re saying when we era-adjust the numbers is that it’s more difficult to prevent runs in one era than the other because the overall quality of the league is different.

But maybe it’s worth unpacking what that really means. For instances, are all 4.3 R/G environments the same? The league is currently scoring runs at a rate similar to 1991, but the game is quite different. Hits are down (8.69 to 8.51 per game), but home runs are up (0.80 to 1.07 per game). Strikeouts are way up (5.80 to 7.97), but stolen bases (0.74 to 0.51) and sacrifice bunts are way down (0.39 to 0.21). Should we expect pitching to be equally difficult in 1991 and 2016 because the overall environment is the same or should the difficulty vary on more than that one dimension?

Additionally, the run environment can change for a variety of reasons. As discussed by Sam Miller and Ben Lindbergh on Episode 875 of Effectively Wild, the run environment could change because the pitchers change, the hitters change, or some combination of both. If the run environment changes because the hitters improve, pitching gets more difficult across the board but a pitcher’s relative performance shouldn’t change. If the the run environment changes because all of the other pitchers improve, pitching doesn’t get any harder, but the same pitcher’s relative performance will change.

The goal of era adjustments is to provide a rough estimate of how a given player would perform in another context, but the actual mechanics of era adjustments aren’t able to do that. We can compare a player’s unadjusted line to the average line during his era, but the differences in eras extend beyond the overall run environment.

This is different than the “would Babe Ruth be able to hit today’s pitchers?” question. Obviously, modern athletes are bigger and stronger than their predecessors. I have no doubt that if you placed peak Ruth and peak Barry Bonds into 2016, Bonds would out-hit Ruth by a substantial margin. Rather, I’m asking a question about the way the game is played, controlling for the talent level of the league. Put another way, is it easier for Kershaw to post numbers 60% better than league average in 2016 than it was for Pedro to post them in 2000 because of the way the game is played?

As the game has skewed toward the three true outcomes, has it become easier (or harder?!) for pitchers to stand out? Is it harder to be 60% better than average when the league scores fewer runs overall, simply because you’re starting closer to the runs allowed floor? Did Pedro have a harder time because bullpens weren’t as good and he was asked to get outs he wouldn’t have to get in 2016? Technology has led umpires to be more consistent, how has that influenced pitching? I don’t know that I have good answers, but I do know that I’m very interested in the questions.

I’m open to the idea that our era adjustments are too simplistic to truly parse the differences between distinct eras. I’m fairly confident that FIP- is an effective tool for comparing 2013 and 2015, but can it handle the differences between 1941, 1968, 1999, and 2015? Is what Kershaw’s doing really as impressive as peak Pedro or are we missing something by collapsing all of the game’s differences into a single number?

But it might also be the case that we can simply compare a player’s overall numbers to the overall run environment and call it a day. If that’s the case, Kershaw is making a compelling argument whose strength varies slightly depending on your metric of choice. At the very least, this method puts peak Kershaw in the vicinity of peak Pedro and forces us to have this conversation. Even if it turns out that peak Kershaw isn’t peak Pedro, the fact that we’re taking the comparison seriously is awfully impressive accomplishment.





Neil Weinberg is the Site Educator at FanGraphs and can be found writing enthusiastically about the Detroit Tigers at New English D. Follow and interact with him on Twitter @NeilWeinberg44.

124 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
FrancoLuvHateMets
6 years ago

Who is better is becoming a tougher question. But, I’m pretty comfortable saying that Kershaw is one of the best pitchers of all time. His name looks right next to Pedro, Big Unit, Maddux, Clemens, Koufax, Seaver, Carlton, Gibson.

adcjordan
6 years ago

Isn’t it a little early to say that? I mean, Kershaw has about 50 WAR to his career…Clemens has 133. That’s an absurd difference.

nenrightmember
6 years ago
Reply to  adcjordan

kershaw is 28

Kyle Sharamitaromember
6 years ago
Reply to  adcjordan

You DO know he did it, right?
#EW

Bipmember
6 years ago

Upvoting for the joke, but half a downvote in my head for messing it up; it’s “you know why?”

redsoxu571
6 years ago
Reply to  adcjordan

WAR is a counting stat. At some point, the peak of players needs to be weighted more than the unadjusted tally. Also, a lot of what Clemens did was fake, so that needs to be factored in.

paqza
6 years ago

He’s still only about half way to solve of those guys, though he’ll definitely close that gap assuming health.