Wednesday, August 28, 2013

Clayton Kershaw's Historic Season

Pitching is very difficult to write about as pitcher use has changed so much over time. The number of pitchers used, how many games they started, how long they stayed in the game and numerous other factors have evolved over time, making it tricky to truly make comparisons across eras. It's very difficult to compare Roger Clemens with Walter Johnson because their career numbers are so different (click on any image to see it clearer):

There are extreme differences between these two pitchers, both of whom are on any list of top 10 pitchers of all time (I'll leave aside any extracurricular activity for Clemens):
1. Johnson had almost 100 more games and around 60 more wins
2. Clemens had about 100 fewer losses
3. Johnson threw almost exactly 1,000 more innings than Clemens and had almost five times as many complete games
4. Clemens struck out many more batters both in absolute terms and per nine innings

FanGraphs WAR values for Clemens is 139.9 and Johnson 125.9. Baseball-Reference uses a different set of calculations which has Cy Young at #1 (170.3), Johnson #2 (152.3) and Clemens #3 (139.4). Someday I'll write a post discussing what I THINK are the reasons behind the difference between FanGraphs and B-R WAR values, but that difference becomes less important than the fact that both values have the same three pitchers as the top 3--ignore the number and place greater emphasis on the rank.

Ranking modern pitchers is very difficult when even going back to the 1960s and 1970s when it wasn't unusual for a pitcher to throw 250 innings or more. Therefore, as I compare the season Clayton Kershaw is having, I'm only going to compare him to pitchers from 1990-2013. This first chart shows how he compares with those pitches in his WAR value per batter faced:

This may not even be a real statistic but I use it to normalize WAR performance. In 2000 Pedro Martinez had one of the most dominant pitching seasons ever, totally eclipsed by an "ordinary" 18-6 record. He had an 1.74 ERA right in the middle of a hitter's era and his ERA+ (normalized to compare his ERA to the league's) was a ridiculous 291, the best in this time span by a very healthy margin (Maddux's 1994 strike-shortened year is next at 271). This gives a clue as to how rare this season was--this chart shows any pitcher with at least 20 starts and an ERA+ of 291 or greater:

Pedro Martinez 291 29 2000 28 BOS 29 7 4 18 6 .750 217.0 128 44 42 32 284 1.74 817 .167 .213 .259 .473 18
Provided by View Play Index Tool Used
Generated 8/27/2013.

That's a small list. In a hitter's era in the American League, teams batted .167 against him. The only mystery to me is how he lost ANY games.

Look at Kershaw and his "average" 13-8 record. I'm down in print stating that not only should Kershaw win the NL Cy Young, it shouldn't even be close but wouldn't be surprised if somehow Francisco Liriano makes a run with a strong finish. I'm not ready to go the MVP route simply because I think the Cy Young should be enough but I don't reject it out of hand--Kershaw will face around 1,000 batters whereas position players bat around 650-700 times. It will be intriguing since I don't consider the NL MVP race to be cut and dried. No, I'm more surprised that Kershaw has lost ANY games, but here's how he did it:

Games in yellow are either losses (8) or no decision (7)--pay particular attention to the number of runs the Dodgers scored in those games in yellow--he's in the bottom third in baseball in run support. Granted, Kershaw did have a rough streak at the end of May with four games where he gave up...5 or more runs! 28 starts with only 6 games with 5+ runs--that's one way to achieve an 1.72 ERA. This chart shows the pitchers with lower run support than Kershaw--keep in mind the average runs scored by team per game is around 4.2:

If you read my post on Miguel Cabrera's historic season, one metric I used was weighted runs created plus (wRC+), a normalized statistic that shows how much better than the rest of the league a player's season was (or in Cabrera's case, IS), and ERA+ is another normalized stat showing how much more dominant a pitcher was than the league. Kershaw's 207 ERA+ is very special--this chart shows every season in baseball history with an ERA+ of 200 or greater (minimum 20 starts in a year):
Go back to the Dead Ball Era and notice some of those numbers--Christy Mathewson's 1909 ERA of 1.14 led to an ERA+ of 222, meaning the league ERA was around 2.59. To expect significant numbers of ERA+ above 200 is very difficult in a pitcher's era, which also explains the extreme paucity during the mini-Dead Ball 1960s. The same names start popping up in the 1990s and I'll save my comments on Pedro Martinez and the Hall of Fame for a different day. I will guarantee that NO ONE would have expected to see Rich Harden on this list--his problem that year was staying healthy.

If you haven't, read my posts on wins (found here and here)--in brief, wins haven't been an  accurate marker of pitcher success since at least the 1940s when pitchers stopped pitching complete games. Especially as we reach the modern era from around 1990 the 20-game win season becomes very rare and will continue to be so (absent drastic changes in pitcher utilization) because if the typical pitcher gets a decision in only around 65-70% of his 35 starts a year, he'll have to be lucky like Max Scherzer (highest run support in baseball) to go 20-2. Therefore, new metrics and methods of pitcher evaluation need to be incorporated to reflect seasons like Kershaw's that aren't just great, but historically dominant--there are only 37 seasons with an ERA+ of 200 or greater--out of over 10,000 possible seasons. Anytime someone enters the domain of seasons that have occurred only .3% of the time should be noted.

No comments :

Post a Comment