Are fastballs actually getting faster?
I don’t think it will come as a surprise to anyone that I use FanGraphs a lot. It seems that every month, owner David Appelman rolls out some new shiny improvement to the site. Just recently, he gave users the ability to look up splits. But while everybody has been sorting and looking things up, something else has piqued my interest.
I’ve actually been meaning to look into this for a while, but it was the two articles Matthew Carruth posted on Monday that inspired me to finally do this. I want to look at the pitch speeds displayed on the FanGraphs player pages, which are provided by Baseball Info Solutions.
Over the course of this past season, I noticed a change in fastball velocity for a large number of pitchers. Though my evidence was anecdotal, there seemed to be a clear trend of increasing fastball velocity, at least from 2008 to 2009. I decided to look at all pitchers who threw at least 90 innings in a given year, and simply find the average velocity of every pitcher’s fastball. Why 90? Because I said so. The identities of the pitchers in the sample change from year to year, but I don’t think that changes much in this case. Here is the average fastball velocity for pitchers who fit the criteria:
2002: 89.47 mph
2003: 89.31 mph
2004: 89.67 mph
2005: 89.44 mph
2006: 89.88 mph
2007: 89.64 mph
2008: 90.13 mph
2009: 90.67 mph
That’s an increase of one full mile per hour in just two years. I don’t think that there’s some new wave of pitchers who suddenly started throwing really hard the last few years. More likely, it’s a problem with BIS’s data collection. From 2002 (the first year this data is available for) through 2007, the average fastball velocity remained between 89.31 mph and 89.88 mph, a difference of just .57 mph. Said differently, the average fastball velocity from 2002-2007 was 89.57. In just two years, that number increased to 90.67.
I looked at the same thing for other pitches, and there wasn’t much of anything to be found. Curveballs, changeups, and sliders didn’t show any clear trend like the fastballs did. This leads me to believe that it’s not a problem with the radar guns they’re using, or we’d see a similar trend across all pitches. I have another theory, however. It’s possible that pitch f/x data, which became available throughout the league in 2008, is influencing what BIS puts into its database. Maybe BIS saw that pitch f/x was reading pitches at faster speeds, and decided to “catch up.”
I don’t really have an answer for why this is happening, and I only half-believe my pitch f/x theory. If someone has an explanation for why this sudden change is happening, I’d love to hear it.
Is it possible it could be the usage of more young arms? I remember seeing some site/blog show that after a certain number of pitches or innings pitchers would lose x mph off of their velocity. I wish I could remember the numbers, but it’s been too long. You might want to find out the age distributions for the last 3 years and see if controlling for that does anything to the numbers.
There’s 55 feet of measuring space between release and plate with a drop of speed of 10mph. Unless you confirm how BIS is reporting pitch speeds from video monitors, I don’t know that you can actually say anything about the change. Every 6 feet of difference is 1mph of difference. Picking something up at 50 feet or 40 feet will give you enough different results to explain the differences here.
And, Gameday has changed its reporting location on more than one occasion.
Well if that is the reason for the change, it would make perfect sense. But why would they change the point at which they record the speed, though? It just makes it harder to compare data from year to year.
It was in test mode. Their priority is to tell a story right then and there.
Probably nothing, but is it possible some pitches weren’t being classified correctly? How many cutters would it take to drop fastballs that much? Is it possible some 87 mph non-fastballs were getting lumped in? Looks like they’ve had a decline in speed over the same span…but not many are categorized CT so migh be non-issue.
I blame steroids.
Where does the BIS pitch speed information come from since they don’t have anyone in the stadium? Has it been the same source over the years or has it changed over time? And even if BIS is using the same source, that source may be getting its information in a different way. For example, if BIS is using speed numbers from the TV station, the TV station may be reporting the numbers from the stadium scoreboard. At the Summit conference last July we learned that Sportvision has been campaigning to get the teams to use Pitch F/x speeds on the Stadium scoreboards and has been successful in some but not all of the parks. So BIS may think it has been using a consistent data source, but in reality it has not.
Jimbo wins. Cutters have been an increasingly more frequent classification in more recent years. It’s quite possible the teams that they provide data to asked them to make it a priority.
I have to believe data collection is always an ongoing effort to improve accuracy. BIS in the scheme of things hasn’t been at it that long either, and they’ve certainly made improvements since 2002. In my dealings with them, they are seriously committed to providing the best data possible.
Tango and Peter are correct. There is no point in even speculating about whether or not the velocity of fastballs is increasing until we are certain that velocity is being measured in a consistent fashion. The fact is that, although we may speak about measuring the velocity of a pitch, that’s not exactly what we are doing. Whether one uses a radar gun or a video camera, or some mix of the two, one is measuring the velocity of the pitch at some distance, X, from the plate. Unless one is certain that all measurements are being made at that same distance, there’s no way to know whether or not fastballs are faster these days.
Here is what Pitch f/x says:
FA FT FF FS FC
91.08 NULL 92.63 85.06 86.72
90.90 NULL 92.10 84.70 86.41
88.89 88.51 91.62 84.84 88.18
For some reason there were no two-seam fastballs classified in 2007 or 2008, I guess cause it’s really hard to tell them apart with a mass classification system. If you take a look at the percentage of each pitch being classified:
Sorry, my post got cut off
Here are the percentages of each type of fastball out of total fastballs.
FA FT FF FS FC
95.5% 0.0% 0.5% 0.9% 3.2%
96.6% 0.0% 0.4% 0.5% 2.5%
7.2% 6.2% 82.1% 0.1% 4.4%
Mostly fourseam fastballs still, but now adding more sinkers and cutters is decreasing the average fastball velocity. So it looks like Jimbo and David are right.
Did you control for the quality, consistency, and types of hardware and software doing the measuring?
I agree with Jimbo. That was my first thought also. Who classifies which pitch is which?
With the myriad of pitchers, pitching styles, release points, and arm positions I doubt even an experienced catcher should determine with complete 100% accuracy what pitch was which unless he signaled for it.