The Statistical King
You guys seemed to enjoy the breakdowns I did on Gil Meche recently, so let’s writeup King Felix from a similar perspective. After another good outing last night, everyone seems to agree that the King is back on track and pitching like we expected from the outset. So let’s break him down by month, and see where the improvement has come from.
Month IP BB% K% GB% FB% LD% HR/FB FIP xFIP ERA April 26.2 10% 23% 56% 25% 19% 25% 4.89 3.52 5.06 May 35.2 7% 22% 53% 25% 22% 21% 4.35 3.29 6.31 June 34.2 4% 20% 59% 23% 18% 13% 3.17 3.04 3.37 July 26.1 12% 19% 59% 27% 15% 15% 4.79 4.40 3.42 August 13 10% 17% 70% 16% 14% 0% 2.97 3.63 1.38
Lots of numbers there, so, here’s a quick overview, by component, and what it’s meant to his run prevention by month:
BB% – He threw a lot of strikes in June, but other than that, he’s still struggled with his command. 10% walk rates aren’t death, but they’re still something that can be improved upon. He’s not really cutting down on his walks lately, so this isn’t the key to his improvement.
K% – He’s actually missing less bats as the year goes on, as he’s struck out fewer batters each month than the previous. His 18% strikeout rate the last 5 weeks is actually just a tick above average. This certainly isn’t the cause for the improvement.
GB%/FB%/LD% – He’s getting more groundballs, less fly balls, and less line drives lately, and that has certainly helped. This is definitely part of the improvement, but the difference isn’t big enough to explain the rebound entirely. The LD% is probably due for a spike going forward, as 14% line drives isn’t sustainable.
HR/FB% – And here’s 95% of the improvement, right here. A league average starting pitcher will allow 11-12% HR/FB rate, and significant variance from that is in almost every case unsustainable. Felix’s HR/FB rate was an absurd 25% in April, a still awful 21% in May, then an almost normal 13% in June, slightly high 15% in July, and he has yet to give up a home run in either start in August. The August number isn’t an abberation, as you might think from the percentage, since he’s only given up six fly balls total in his last two starts.
His Fielding Independent ERA basically tracks the difference in run prevention if his HR/FB rate was indeed a true skill, and you can see how big of a difference home runs make. xFIP translates HR/FB rate into a league average number, eliminating that “skill” from the analysis, and gives a pretty clear picture that Felix hasn’t really changed much at all. Outside of July, he’s posted consistent xFIPs in the 3.00-3.60 range, which is outstanding. For comparison, Johan Santana’s xFIP for the season is 3.42. Roy Halladay’s is 3.57. That’s the class of pitcher that Felix has been in most of the year, by xFIP.
This is about as clear a picture as one could paint for why ERA is not a good tool for predicting pitcher performance going forward. The only thing that has significantly changed for Felix from April to August is the amount of his flyballs that go over the wall. That entire change, mostly out of his control, has led to the mirage of improvement.
Felix has been pitching mostly well all year. ERA just didn’t do a good job of telling people that.