A few years back, Mike Fast wrote an article for The Hardball Times called “Lose a Tick, Gain a Tick”. It basically looked at everything to deal with gaining or losing fastball velocity. Recently, I wrote a RotoWire article where I basically had to recreate Mike’s data, and I thought I would share the results.
The part of the article I looked into was the change in pitcher results, as their fastball changed velocity. Originally, Mike found starters saw their RA go up by 0.28 RA for each mph in fastball velocity they lost. For relievers, it was .45 RA. While I didn’t use RA values, I did get ERA, FIP, xFIP and K% changes between the 2013 to 2014 season. Here is what I found:
The numbers for starters are identical. The numbers for relievers have increased a bit. I am guessing hard throwing relievers, without the ability to “pitch,” are more common than they were five years ago. As these pitchers lose their speed, they lose their effectiveness faster.
Another interesting number is the xFIP value, which doesn’t change as much as the other two, which are almost move in tandem. There seems to be a study someone could do here.
My theory is that a pitcher’s ability to induce fly balls seemingly stays relatively constant over time. As they throw softer, they see their xFIP only increase because of the decline in their strikeouts, while FIP increases because of the loss of said strikeouts and home runs hitter are able to square up and take them deep easier. For example, here are Tim Lincecum’s yearly ERA, xFIP and FIP values plotted against his average fastball velocities.
With this sample size of one, his ERA and FIP were lower than his xFIP over 92 mph. As his velocity declined, his xFIP has not increased as much as his FIP and ERA.
Looking back at Mike Fast’s five-year-old article, most of the information still holds up today. The only change I found was relievers now will see more of an increase in ERA and FIP as their velocity declines. Even though he didn’t look at it, xFIP doesn’t move in tangent with ERA like FIP does. This discrepancy is one corner of the sabermetric world someone can explore.
Jeff, one of the authors of the fantasy baseball guide,The Process, writes for RotoGraphs, The Hardball Times, Rotowire, Baseball America, and BaseballHQ. He has been nominated for two SABR Analytics Research Award for Contemporary Analysis and won it in 2013 in tandem with Bill Petti. He has won three FSWA Awards including on for his MASH series. In his first two seasons in Tout Wars, he's won the H2H league and mixed auction league. Follow him on Twitter @jeffwzimmerman.
A simpler explanation:
Because xFIP is the most heavily regressed of the three stats (regressing BABIP, HR/FB, and LOB%), it simply doesn’t fluctuate as much as the other two
We are talking about fluctuation over a complete sample though, that being all starters and all relievers. Given that ERA and FIP both change so similarly, and this holds over a complete sample, it seems unlikely that it’s just random variation.
The variable that affects ERA and FIP and not xFIP is HR/FB rate. We also know from research that went into the development of SIERA that high-strikeout pitchers do actually tend to have better HR/FB rates overall. This causes me to want to agree with Jeff, that velocity is related to HR/FB rate in a real way.