Abstract Baseball: FIP

I like FIP. Thanks to Fangraphs, I can tell you that it’s a stat that shows what a pitchers ERA should look like when looking at the results that a pitcher can control: strikeouts, walks, HBPs and HRs. It is a predictive statistic, and it does a better job illustrating a pitcher’s ability than ERA, which relies on variables that are beyond his control. The results are presented on the same scale as ERA, so it’s read the same way – the lower, the better. For example: a 2.90 FIP is excellent, 4.00 is average, and a FIP of 5.00 is awful.

I was curious to know what happened to a pitchers’ FIP over time. I wanted to see whether or not it would go up after a certain age, if there was a season in recent history that FIP was down across the league, and who had a better overall FIP, starters or relievers.  Again, turning to Fangraphs I pulled the average FIP for Starters, Relievers and Both aged 21 to 45 from 1997 to 2011. I did not include this years’ data because it is only June, and that’s just too small a sample size to mean anything. 
Rather than just putting up spreadsheets (which can be found here), I decided to do something a little more visually appealing, which I have dubbed Abstract Baseball: FIP. Here are the results:

[all pitchers aged 21-45, 1996-2011]

[starters aged 21-45, 1996-2011]

[relievers aged 21-45, 1996-2011]

Each 'canvas' contains 16 columns and 25 rows. The columns (from left to right) represent the years 1996 to 2011, and the rows (from top to bottom) represent pitchers age 21 to 45. As you might imagine, the lighter the square is, the lower the FIP for that age group in that particular year. The absence of colour represents a year where no pitchers in that particular age group made an appearance.
I was hoping to see a definite trend in either age or season, or a major difference between starters and relievers, and I was surprised to see a relatively flat ‘canvas’. In retrospect, there should have been a few more qualifiers in place, as there are a few very dark spots which indicate aberrations in those year’s results.
Next time, I’ll look at xFIP (a regressed version of FIP) and make some changes to see what it looks like.