Fastball Velocity and Pitcher Projections
Pitcher projections are an exercise in futility. Pitchers are very hard to predict, and a lot of that probably has to do with not knowing who is going to get hurt, which is a common result when you abuse your arm by throwing a baseball. Beyond the problems of trying to project any pitcher, it seems to me that minor league stats are not quite as reliable as major league stats.
I often read scouts or analysts dismiss some pitchers who have good minor league numbers because they do not have dominant stuff. The theory goes that these pitchers are able to fool bad and inexperienced minor league pitchers, but will be hit hard by major league hitters. Other pitchers supposedly with big league stuff will translate better. Their numbers won't suffer so much in moving from the minors to the majors.
Now that David Appelman has made it easy to find pitch velocity even for those of us who don't download pitch f/x, I combined 2007 pitching results with the CHONE projections and average fastball velocity. I'm looking only at pitchers who had fewer than 50 major league innings going into 2007 (therefore most of their projection is based on pitcher MLE's), and who I actually ran a projection for.
What I want to do is see if, given a pitcher's projection from his MLE, whether fastball velocity tells us any more useful information for his projection. In other words, do fireballers beat their projections? Do soft-tossers fail to live up to theirs?
These inexperienced pitchers had a combined ERA of 4.81 in 2007. The CHONE projections, prorated to their actual innings, had a 4.46 ERA. A big part of that is an error I found in my baseruns formula, which under-calculated the ERA. Also, I may have been too generous to the minor leaguers, not using harsh enough MLE factors or regressing to the wrong mean. The baseruns error probably accounts for most of it, I think it was about .3 runs on average. Anyway, I'll just multiply all the projected ERA's by 4.81/4.46 to make it easier to compare, as I'm interested in relative ERA and not absolute.
I have 142 pitchers who meet the criteria of the study. I grouped them into hard-throwers, with greater than 91.5 average fastball velocity, soft-tossers who throw under 88.5, and a middle group.
The Soft-tossers, 43 of them, pitched a combined total of 1664 innings. Their adjusted projected ERA was 5.08. The actual ERA? 5.03
The middle group projected at 4.66. The Actual ERA was 4.70. There were 61 pitchers here and they threw 2381 innings.
And the Flamethrowers projected to a 4.74 ERA. The actual ERA was 4.76, 38 pitchers and 1790 innings.
Without breaking out the old college stats book, I'm pretty sure these tiny differences are insignificant. The conclusion is, in the words of Officer Barbrady, "Move along, Nothing to see here". Knowing a pitcher's velocity doesn't tell you anything about his chances of success that you didn't already know by looking at his minor league numbers.
This is not to say that we can't improve pitcher projections by quantifying a pitcher's "stuff". There's more to stuff than fastball, though let's be honest, a prospect with a big fastball gets a lot of baseball people excited. Maybe some of the guys who work with pitch f/x will be able to use more detailed data (like quantifying a curveball by its break) and add to pitcher projections. But then again, maybe not.