Tuesday, May 29, 2012

Tweaking xFIP

FIP is an ERA estimator.  You can read about it more here and here

FIP = (13*HR+3*(BB+HBP)-2*K)/IP + Constant

The constant brings FIP on the same scale as ERA.  The constant so far for 2012 is 3.00. 

The FIP for the 5 Tigers’ starters so far this year is:

Justin Verlander – 2.38
Doug Fister – 3.82
Max Scherzer – 3.96
Drew Smyly – 4.03
Rick Porcello – 4.37
This gives a pretty good indicator of how these pitchers have truly pitched this year, at least better than what ERA would tell you.  However, there is a further step that can be taken to give you xFIP.  To quote FanGraphs:

Home run rates are generally unstable over time and fluctuate around league-average, so by estimating a pitcher’s home run total, xFIP attempts to isolate a player’s ability level. A pitcher may allow home runs on 12% of their flyballs one year, then turn around and only allow 7% the next year. HR/FB ratios can be very difficult to predict, so xFIP attempts to correct for that.
xFIP = (13*xHR+3*(BB+HBP)-2*K)/IP + Constant

The only difference between FIP and xFIP is the HR rate.  FIP uses true HR while xFIP adjusts it to the league average HR/FB ratio.  The league average HR/FB ratio so far this year is 10.9%.  The HR/FB ratio so far for the 5 Tigers’ starters in 2012 is:

Max Scherzer – 16.7%
Doug Fister – 15.4%
Rick Porcello – 13.0%
Drew Smyly – 12.5%
Justin Verlander – 5.2%

Except for Verlander, the Tigers’ starters have been very homer happy.  Scherzer, Fister, Porcello and Smyly have allowed HR at a higher rate than the league average.  This should average out as the season progresses and according to xFIP, it should be closer to the league average. For example, Max Scherzer has allowed 10 HR so far this year on 60 fly balls, (10/60 = 16.7%).  However, the league average HR/FB ratio is 10.9%, so he should have only allowed about 6.54 HR (10.9%*60).  I know it’s impossible to allow a fraction of a HR, but that’s how the calculation works.  Substitute 6.54 in for the 10 HR and his FIP is reduced.  The xFIP for the 5 

Tigers’ starters so far this year is:

Max Scherzer – 3.13
Justin Verlander – 3.14
Doug Fister – 3.37
Drew Smyly – 3.79
Rick Porcello – 4.12

There you have it, Max Scherzer has been just as good as Justin Verlander when you adjust for their HR rates.  Er…eh, no.  I’m not buying it.  There is one major flaw when adjusting for the league average HR rates.  To quote FanGraphs again:

While HR/FB ratios are generally unstable over time, some pitchers are still more prone to allowing home runs than others. If a pitcher has a long history of over- or under-performing the league average with their HR/FB rate, then you can reasonably expect them to perform closer to their career average than the league-average. In cases like this, xFIP may overestimate or underestimate a player’s true talent level by assuming a league average HR/FB ratio.
Over their careers, Justin Verlander has a significant lower HR/FB ratio than Max Scherzer.  

Career HR/FB ratios:
Doug Fister – 7.5%
Justin Verlander – 7.7%
Max Scherzer – 11.4%
Rick Porcello – 11.4%
Drew Smyly – 12.5%

So instead of using the league average 10.9%, what would be the harm in using their career rates?  I think this gives a better idea on what to expect instead of using the league average.  Verlander and Fister have shown to give up less HR than the league average, so expecting that to continue is not out question.  

Justin Verlander – 2.71
Doug Fister – 3.04
Max Scherzer – 3.20
Drew Smyly – 4.03
Rick Porcello – 4.18

Now, those numbers I believe.  

1 comment:

  1. Good post. While I agree that career numbers make more sense, how to account for a rookie? At what point in a career would you start using career numbers vs. league averages, etc?