Average Fastball Velocity and Pitcher Performance
The other day, one of the authors of The Book: Playing the Percentages in Baseball, tweeted out something interesting. https://twitter.com/mitchellichtman/status/453026556080693248
The initial reaction is that it can't be that simple. However, with the public release of PitchF/X data we have seen a number of great baseball research projects and perhaps no topic has been covered more thoroughly than average fastball velocity.
In 2010 Mike Fast of The Hardball Times wrote a long post on the affect of a change in average fastball velocity and the trend towards a higher league average fastball velocity. Fast's conclusion was similar to Lichtman's, a starter can expect to see their run average (RA) increase or decrease by about .28 runs per one mile per hour increase or decrease in average fastball velocity. While relievers see a change in their RA of about .45 runs.
So we can see that a simple change in average fastball velocity can have a remarkable affect on a pitcher's effectiveness, but surely just a handful of games into the season it's too early to be concerned. Well, according to research done by Jeff Zimmerman of Fangraphs, after one game we can expect a pitcher's average fastball velocity to have a standard deviation of 0.8 MPH away from that season's average fastball velocity, by three games the standard deviation is just 0.55 MPH. Therefore, a pitcher's average fastball velocity stabilizes quickly. In fact, a follow-up discussion on the blog for the aforementioned The Book found that average fastball velocity after just two fastballs has a correlation factor to seasonal average fastball velocity above r= 0.5. We can be reasonably comfortable in thinking that a player's average fastball velocity in just his first game will be close to his average fastball velocity for the season.
So with these two pieces of knowledge in hand let's see what it means for the Nats. Here's every Nats pitcher who has thrown a fastball for the Nats in 2013 and 2014 and their average velocities as reported by Fangraphs and PitchF/X.
Gonzalez, Detwiler and Clippard all have seen an increase in their average fastball velocity and if this holds should summarily see a bump in their performance this season, that's good to see. The difference in Storen's velocity is too small to make any definitive judgements and considering Zimmermann pitched with the flu we will exclude him from further inquiry for now.
This leaves us with Jordan, Strasburg, Roark and Soriano as pitchers who have seen a drop of over one mile per hour in their average fastball velocity over last season. This doesn't seem to be good news for the Nats' fifth starter competition as both Jordan and Roark are expected to provide the Nationals with quality innings from that spot. Don't be surprised if someone besides Jordan or Roark is occupying the fifth starter spot later in the season if they don't bring their velocities up.
Jordan's drop in particular is worrying as a difference of 2.5 miles per hour is three first game standard deviations away from his average fastball velocity in 2013. This does not inspire confidence, especially in a pitcher who had an RA of 4.70 runs last season and is coming off of a Tommy John surgery in 2011. While it's not a certainty that Jordan won't see his velocity rise as the season continues, it's something to watch.
However, the headliners here are clearly Strasburg and Soriano. Before we run for the hills in fear that the Nats ace has something irreparably wrong with him let's look at the PitchF/X numbers provided by Brooks Baseball. Unlike Fangraphs, which takes their pitch classifications directly from MLB Advanced Media, Brooks Baseball's Harry Pavlidis classifies each pitch by hand himself. Brooks Baseball also back calculates speed to 55 feet from home plate rather than the 50 feet that PitchF/X cameras begin tracking the ball at, leading to slightly more accurate numbers.
For Strasburg the numbers still aren't great. His average fastball velocity in 2013 was 96.23 MPH while in two starts in 2014 it is 94.91 MPH, giving us a difference of -1.3 miles per hour, smaller but still not great. The main problem is that he threw much slower in his first start, just 94.08 MPH. In his second start, Strasburg averaged a more normal 95.79 MPH, which would represent a drop of just 0.4 MPH and is actually higher than his average fastball velocity in September of 2013 (95.75 MPH). So, while we can't just ignore Strasburg's velocity drop out of hand, we shouldn't be too concerned just yet.
The same cannot be said for Soriano. Below are graphs of his average fastball velocity the past four years and his RA for the same time period.
His average fastball velocity has been steadily decreasing and his run average steadily increasing, with 2011 a bit of an outlier likely due to his injury. Last season Soriano saw a drop of 0.9 MPH and saw his RA increase by .99 runs. As we can see from the chart above after two appearances his average fastball velocity is down by about another mile per hour. By Fast's numbers we can expect him to add around another .45 runs to his RA, giving an expected RA of 3.71 runs. Last year, the average RA among relievers with at least five saves was 3.17 runs. Brooks Baseball doesn't paint a rosier picture as by their numbers Soriano has seen his average fastball velocity dip from 92.78 in 2012 to 92.08 in 2013 to 90.84 this year.
If you were looking for a statistical reason to justify concern over Soriano's ability to perform well in the closer role this would be it. While it's certainly not definitive that he will see such a large drop in his average fastball velocity or that he could not find some way to work around it and remain effective, it would be a risky bet to assume so. It's certainly a possibility that the Nats see someone else occupying the closer role by the time October rolls around.