Does Fastball Velocity Influence a Pitcher’s HR/FB Ratio?
On Thursday, I posted an update on three American League rookie pitchers, including Seattle phenom Michael Pineda. One of my criticisms of Pineda was his 56% fly ball rate at the time, which should lead to lots of home runs, despite the fact that he had yet to give up even one long ball. One of the commenters noted Pineda’s fantastic average fastball velocity, currently sitting at 96.1 MPH, and opined that it will be more difficult for hitters to homer off of him, leading to a sustainable depressed HR/FB ratio compared to the league average. Not satisfied with just taking his word for it, I decided to test this hypothesis.
Using what limited knowledge I still retain from my college statistics class, I calculated the correlation between a pitcher’s fastball velocity and his HR/FB ratio. My sample size totaled 226 pitchers from 2001-2010 with a minimum of 500 innings pitched, and included both starters and relievers. Park adjustments or any other such fixes to make the study more accurate were not made. The correlation between the two variables was -.29. Below is a scatter plot of the data with the trendline.
There is a clear positive relationship between higher fastball velocity and a lower HR/FB ratio. This makes intuitive sense, so that is always a good sign. It appears that the commenter may be on to something that may help explain at least some of the differential between an outliers’ HR/FB ratio and the league average.
Over the years, there have been many formulas devised to estimate a pitcher’s ERA based on true talent, defense independence or only factors that are supposedly within the pitcher’s control. However, none of these ERA estimators include anything relating to pitch velocity or pitch type. This data is easily accessible and it does not seem like it would be too difficult to include it in future iterations of these formulas.
Mike Podhorzer is the founder of ProjectingX IQ, an advanced fantasy baseball analytics platform that transforms projection data and in-season performance signals into actionable intelligence. He is the 2015 Fantasy Sports Writers Association Baseball Writer of the Year and three-time Tout Wars champion. He is the author of the eBook Projecting X 2.0: How to Forecast Baseball Player Performance, which teaches you how to project players yourself. Follow Mike on X@MikePodhorzer and contact him via email.
interesting…but the big flaw i see is that you lumped relievers and starters together. relievers seem to have the ability to sustain slightly lower hr/fb rates than starters. i would think that they also tend to throw harder (at least this is true of the same pitcher pitching in relief vs. starting, and i would assume that it is true for the general population of relievers vs. starters). what happens if you redo it with only starters? i bet the correlation decreases significantly.
I wonder if it’s only relievers with exceptionally high velocity who can sustain low HR/FB…
Yes, it would have been better to have only included starters. Completely forgot FG has a “Starters” and “Relievers” tab to easily filter only starters. With only starters from 2001-2010, minimum 500 innings pitched and a total of 182 pitchers in the sample, correlation drops to -0.21. A decline as expected, but still appears to have some significance.
ok…and what’s the p-value? because the -0.21 might be significant…or it might not be
It hasn’t been proven to my satisfaction that some pitchers can’t consistently beat the HR/FB average. I judge established pitchers against their own historical level, not the league’s.
actually it’s been shown that some pitchers can consistently beat (or do worse than) the HR/FB average. it’s just that it takes a long time to separate the talent from noise. i think it’s around the 800 inning mark where about 50% of the observed variance from the league average of ~10% is due to skill and 50% due to luck/random variation. thus if you observed a pitcher with a 7% hr/fb rate in 800 innings, your best guess of their true talent rate is ~8.5% (ignoring park effects of course).
I like the idea of this study but I’d have to agree with the first commenter that starters and relievers should be separated for reasons stated above. I’d also contend that while a pitchers high FB velocity may reduce their HR/FB rate it doesn’t say anything about secondary stuff which can significantly influence their tater rate. I don’t know if it can be done with info here at FG, but it would be interesting to see FB velocity vs HR/FB on fastballs alone.
Yeah, it seems to me like most homers aren’t even off fastballs.
Why is this a rotographs article?
Because this should help project future pitching performance, which is essential to fantasy baseball.
The first comment strengthens the finding noting that RP tend to throw faster. A fly ball can not tell the difference between a relief pitcher and a starting pitcher. In this case it’s about the numbers not pitchers. No need to over complicate the numbers. Basically, the way I took it, if the guy has a heater he might be less prone to giving up the long one.
The extreme FB ratio for Pineda has to go down. It’s down 6% already. The findings are a nice gauge but looking at Mph as a factor in average ratios in isolation is tricky. Not every pitch thrown at 96.1 is equal. For example, in
Pineda’s case a FB% is a much more broad indicator than the ridiculous 70% first pitch strike rate. A lot of things determine where the ball goes once it’s pitched, like quality/type of batter, location of the pitch, break of the ball, type of the pitch, strategy, whatever, etc. But all pitchers control the first pitch. Now, look at the pitchers above 65% they are all useful guys. Pineda is the ONLY guy above 93 mph on the first page! Don’t rip me because I don’t really KNOW for sure what all this means, there is not enough sample, but this is something we have NEVER seen before. Strasburg got all the hype but Pineda might actually claim the status right beside him. Off course, for ESPN to get into that kind of talk he’d have to toss two complete games… But whatever.
“The first comment strengthens the finding noting that RP tend to throw faster. A fly ball can not tell the difference between a relief pitcher and a starting pitcher. In this case it’s about the numbers not pitchers. No need to over complicate the numbers. Basically, the way I took it, if the guy has a heater he might be less prone to giving up the long one.”
right, but maybe relief pitchers are able to do something else that leads to lower hr/fb rate other than throwing harder (ie maybe because they throw less pitches per appearance they can throw from a different arm slot that leads to lower hr/fb ratios, etc). starting and relieving are different beasts…it’s rarely a good idea to use a sample containing both when doing a study.
What the significance of the R-squared?
Also, where did you get your database from?
Well, someone should have mentioned this to Verlander yesterday…
it goes beyond the scope of a quick article like this, there may also be some bias here. no idea of this is the case, but if pitchers in hitter’s parks average lower velocity than pitchers in pitcher’s parks, then there’s some bias that make this correlation look stronger than it really is.
Nice study. I see no reason to separate starters and relievers. Of course there may be other factors involved. Matt Cain, for instance, has been able to maintain a low HR/FB despite lower velocity. Of course, I don’t know what those other factors might be, but I’d bet that FB command, which is much harder to measure, has a lot to do with it. A 90 MPH fastball down the middle is more likely to get hit out than a 96 MPH fastball down the middle, but a 90 MPH fastball where you want it is going to stay in the park a lot more often than a 96 MPH heater down the middle.
I meant to add that if someone can come up with a logical factor that would enable relievers to have a low HR/FB, then it might make sense to separate them, but so far all of the suggestions have been straining at gnats.
Josh Beckett gives up a lot of HR thou…
As you can see, there is a lot of scatter so it’s pretty meaningless to pick out one pitcher as a counter-example. I would hypothesize that location/command has a lot to do with it. If you could find a simple way to measure FB command and combine it with velocity, you would get a much tighter graph.
Jono, your point is valid. I actually challenged the FB% as a sharp indicator in general. As it relates to the study going beyond the actual ball in flight, wherever it came from, is overkill. There are too many things to consider before the ball is actually a fly ball (not just RP vs SP). I like the broad impact of the correlation which basically points out there is something worth considering.
Love the study, I think the point is valid. The reason (perhaps) to separate SP and RP is the fact that your SP might be more prone to giving up HRs since the players have seen them more often. If, for example Clayton Kershaw always throws a first pitch curveball to Cody Ross, Cody might learn by the 3rd at bat to swing for that. With RPs, most likely hitters see them once a game, so its much harder for hitters to get a beat on what types of pitches are coming and in what order they may see them, what speeds, etc..
I’d like to see at what point does your heater stop beating the 10% HR/FB avg? Is it 94, 95, 93 mph?
My intuition would have said that harder thrown balls will go farther when hit well than softly thrown balls, so I expected the reverse results.
Thats true, but a homerun is a homerun as long as it clears the fence. You dont count it extra just because it goes farther (due to the pitch being thrown harder or for any other reason)
That’s like saying steroids don’t help hitter hit HRs because who cares how far you hit them. Anything that increases the distances of fly balls will have a positive correlation on HR rate, because besides increasing distances of fly balls that would already be HRs, it also increases distances of other fly balls that become HRs.
That said, I’m pretty sure a hitter’s HR rate has far more to do with his ability to make solid contact and bat speed than the speed of the ball at contact. Higher velocity fastballs tend to induce weaker contact overall (harder to hit fastball itself, plus being able to set up other pitches).
Mike,
I have been looking for a solid program to run some customized stats.
What does FG stand for?
nevermind, I found it.
Interesting analysis.
The pattern doesn’t look like it’s necessarily linear. You need to look at the regression diagnostics and consider other models (second order would by a good first attempt) beyond just simple linear regression.
What was the R^2 from the regression?
one variable linear model. square his correlation…
A few have already requested it, but the significance of the R^2 (i.e. correlation) is a very important value that is missing from the analysis. If the p-value is some outrageous number like 0.5, then this article is completely irrelevant.
With that in mind, this type of analysis is along the lines of something I wanted to work on for quite some time, and might be the kick-start I need. I would like to measure whether or not pitch speed differentials are more (or less) correlative than just absolute speed alone. For example, does someone with a 78mph changeup and 94mph fastball perform any differently than someone with a 97mph fastball? And is there a difference between starters and relievers? Those are just some of the types of questions I would ask myself to help direct the initil iteration of the analysis. I would perform the analysis using WAR as my dependent (i.e. the metric I am trying to estimate).
I don’t have much experience in pitch f/x databases, so if someone could point me in the right direction I would be greatly appreciative. I do, however, have a statistics background and experience in the R statistical package, so once I have the data it will be fairly easy for me to complete my analysis.