David Brown/Spurdynasty on Fastball Velocity and Pitching Performance

by Joey Matschulat on Jul 1, 2009 3:59 PM CDT Comment 21 comments

Ordinarily I'm quite reluctant to post full-fledged FanPosts promoting work from our own site (and as usual I must profusely thank Adam and Brett for allowing me to do so), but David Brown (a.k.a. spurdynasty) might have just published his most informative piece yet -- this one focusing on the relationship between fastball velocity and a variety of performance-related metrics, including ERA, FIP, K/9, BB/9, and a brief rundown of some of the Rangers' most prominent major league and minor league pitchers and their own fastballs. A brief excerpt, which doesn't really do it justice:

* * * * *

Average home run rates and left-on-base percentages for pitchers classified by fastball velocity were plotted together, not because there is a correlation between the two but because they have similar values. The data indicates that pitchers with high velocity fastballs give up fewer home runs and are perhaps slightly more effective at stranding runners who get on base. These data contradict the notion that fastball velocity, at least among pitchers who become major league starters, is not that important.

* * * * *

Check out David's full piece here: http://www.bbtia.com/home/2009/7/1/fastball-velocity-pitching-performance-and-rangers-starting.html

Tangentially related note: Jon Daniels will be on the Hardline at 5:50 p.m. No word on whether Mike will regale him with a second rendition of Rangers play-by-play.

7 recs | Comment 21 comments

Story-email Email Printer Print

Comments

David's best post

yet. Very well thought out and presented.

G G G E-flat_______ F F F D__________....

by t ball on Jul 1, 2009 4:44 PM CDT reply actions 0 recs

After reading that...

I feel like I now have solid data to backup what had only been a gut feeling/common sense type thing before. It’s great that it can be boiled down into easy to remember big points. It’s even cooler that the Rangers seem to draft according to this strategy.

by GhettoBear04 on Jul 1, 2009 4:52 PM CDT reply actions 0 recs

great, great post

very insightful and easy to read, sometimes i find those types of post tedious to read and understand, but that was very well written

"The House That Ruth Built, 85 years old, goes out as The House That Hamilton Knocked Down"

by blalock84 on Jul 1, 2009 4:56 PM CDT reply actions 0 recs

Really good stuff

by Houston27 on Jul 1, 2009 5:38 PM CDT reply actions 0 recs

This is a great post

Before I read it, I was almost going to make a joke about how I’m sure that it’ll somehow try to prove that Wilfredo Boscan is a no-doubt, sure-to-succeed future #1 pitcher. So I was amused that it started with a Boscan anecdote, only then showing exactly why I have trouble seeing Boscan as a top prospect in the organization.

I guess I don’t understand this sentence early on

Contrary to expectations, there was a very strong correlation between fastball velocity and ERA.

given the earlier point about really fast fastballs “only” shaving a few hundred milliseconds off a pitch. A few hundred milliseconds is actually a whole lot to the brain, especially given that these guys have perfected the swing/don’t swing decision making process. Years of training has led these guys to very, very fast reaction times when it comes to judging a pitch. Literally, their brains are optimized for this task in ways that most of us could never hope to achieve. If you gave every major league hitter an extra 200 ms on a pitch, I would completely expect their swing and miss rate would drop and their HR rate would go up.

It is a very interesting collection of data and a neat analysis.

Go Rice Owls!

by JBImaknee on Jul 1, 2009 5:49 PM CDT reply actions 0 recs

I should have written “Contrary to my expectations…” Before comparing starting pitchers, I thought there would be a decent, but not overwhelming correlation between fastball velocity and K-rate. I also suspected that there would be an approximately equal correlation between velocity and walk-rate. My hopethesis was that the two would cancel each other and that there would be very little correlation between velocity and ERA. I was surprised by the results and wanted to share.

The difference in the amount of time that a hitter has to swing at a 95 and 90 MPH fastball is not 200 ms but 20 ms. 20 ms is approximately 1/10th of the time that it takes the average human to blink.

Thanks for the idea on correlating fastball velocity with swing-and-miss rates. I think that I’ll take a look at that. My guess is that there will be a higher rate of swings on 95+ MPH fastballs that are outside the strike zone. If the data are interesting, I’ll toss it out there for discussion.

by spurdynasty on Jul 1, 2009 11:12 PM CDT up reply actions 0 recs

my bad on the 200ms vs 20ms

I misread your sentence in there and didn’t both to think. Kind of embarrassing for me, actually.

The brain can still do a lot with 20 more ms, but obviously not nearly as much as the 200 ms (where it can do a whole, whole lot.).

Go Rice Owls!

by JBImaknee on Jul 2, 2009 9:30 AM CDT up reply actions 0 recs

Picking the brain of an expert

At BBTiA, the following comment was made by Fai Mao:

The difference in velocity between an 89 mph fastball and a 93 mph fastball may actually be greater than 3/100’s of a second if you include the slower arm motion of the pitcher with the slower fastball. Batters don’t pick up the ball on the way to the plate but out of the pitchers hand. They then extrapolate where the ball will be based upon observed spin and guessed speed. (It is physiologically impossible for them to watch the ball until it meets the bat) There might be enough greater arm speed in a pitcher with a better fast ball to make that task more difficult. Thus it may not be the fastball speed so much as the arm speed.

Based on hitting against batting machines and watching major league pitchers from the stands behind home plate, I was unable to understand the comment. Do you have some insight that will help?

by spurdynasty on Jul 2, 2009 10:34 AM CDT up reply actions 0 recs

Hmm...

It actually makes some sense. You know how they say that the hardest balls for CFers to judge are those hit right at them? If the difference between missing a pitch and hitting it squarely is judging the velocity of a pitch essentially coming at you within a few mph, I’d almost think that it would be impossible to ever do anything other than guess.

On the other hand, the arm motion provides a much bigger sensory input, and by definition the hand and arm have to be traveling at whatever speed the ball is released at. From a sensory point of view, judging velocity off arm angular velocity may be ideal.

Brains are very highly adapted to detect motion – there is a whole region of the cerebral cortex where neurons are motion sensitive, and several subcortical areas that do the same (a more basic detection, but probably faster). Almost certainly there are neurons in the brain that will respond differently to an arm moving at different angular velocities 60 feet away. I’d imagine what happens during training – the mental process of learning to hit – is that your brain is figuring out how to best connect those neurons to the neurons responsible for making decisions about swinging, etc.

As the poster postulates, a reasonable hypothesis would be that arm speed predicts velocity, and the angle the ball comes out and its spin predicts location. Obviously speed + spin is important for separating out changeups from sliders, etc, and thus the decision process is harder. Which is probably why having several consistent pitches can be invaluable.

Go Rice Owls!

by JBImaknee on Jul 2, 2009 11:02 AM CDT up reply actions 0 recs

That helped a lot. Thank you.

I think that the part that was throwing me was the sentence in the middle that stated that it is “…physiologically impossible for them (hitter) to watch the ball until it meets the bat.” I originally read that to mean that a batter can’t track the path of the ball as it approaches the plate but I now think that he is saying that a batter can’t watch the ball from the point that it leaves the pitcher’s hand until it hits his bat.

I think that you are right on with a hitter’s decision process. A major league htter must decide whether to swing and approximately where to swing either upon ball release or soon thereafter. The batter then has a very short period of time to decide specifically where to swing. If the early estimate of pitch velocity and/or location was off, then the hitter winds up flailing at a ball that is not where he expected it at the tim ethat he expected it.

by spurdynasty on Jul 2, 2009 11:19 AM CDT up reply actions 0 recs

arm speed

I agree on the comments about arm speed. Johan Santana has a plus-plus CU because his arm speed is the same when he throws his FB or CU.

A quick search revealed a couple of interesting related articles.

Your Little League coach probably didn’t know it, but every time he sent you to the plate with the instructions “keep your eye on the ball,” he was giving you an impossible task.

curveballs

eyes

Elvis Andrus - 2009 AL Rookie of the Year
Mitch Moreland - 2009 Rangers Minor League POY

by RangerMad on Jul 2, 2009 6:30 PM CDT up reply actions 0 recs

And this makes the recent comments

about Perez’s changeup very exciting. He’s apparently getting a 10-12 mph difference off his FB without changing arm speed.

G G G E-flat_______ F F F D__________....

by t ball on Jul 2, 2009 10:02 PM CDT up reply actions 0 recs

Wilfredo Boscan

In regard to Boscan, he only came up in the article because of the quote from Danny Clark. The article was not intended to disparage Wilfredo in any way and I still believe that he is a very intriguing prospect. One thing that I do not know is whether the 89 MPH fastball that Clark mentioned is a 2-seamer or a 4-seamer. If Boscan is consistently hitting 89 MPH with a 2-seamer, then that is a major league-type pitch that he should be able to complement with a low-90’s fastball in the next year or two.

One of the things that has been discussed here and elsewhere is whether it is better to have a young pitching prospect with great control and secondary pitches but only an average fastball or a prospect with a strong fastball but limited control and/or secondary stuff. In essence, is it more likely that a young pitcher will add velocity or improve control?

Related to this, I’m interested to see how Boscan, Font, and NeRa get ranked when Dirkatron gets the LSB prospect threads started. I have all three rated pretty closely and am still trying to work through who I think is the best prospect. One thing in Boscan’s favor is the observation that there are a lot more successful major league starting pitchers with good control, good secondary stuff, and a high 80’s fastball than there are pitchers with below average control and a big fastball.

by spurdynasty on Jul 2, 2009 10:37 AM CDT up reply actions 0 recs

Should have been “…low-90’s 4-seam fastball..” in the last sentence of the first paragraph.

by spurdynasty on Jul 2, 2009 10:38 AM CDT up reply actions 0 recs

Maybe I just can't find it in the article

But what is the actual correlation between fastball velocity and ERA?

r=0.7, r=0.5?

by Requiem on Jul 1, 2009 6:52 PM CDT reply actions 0 recs

Correlation coefficients

I started to include that in the article but decided that most people tend to gloss over those numbers. Below are two coefficients for each of the rolling average values compared to the rolling average pitch velocity. The first coefficient is for all of the pitchers; the second is for all pitchers with average fastball velocities greater than 87.5 MPH. I included the second coefficient to eliminate the group of effective, low velocity veterans who have found alternatives to good fastballs.

ERA: -0.68, -0.84
FIP: -0.80, -0.92
K/9: 0.83, 0.88
BB/9: 0.36, 0.39
HR/9: -0.67, -0.88
BAA: -0.69, -0.81
BABIP: 0.28, 0.12

The correlation coefficients that approach -1 correspond to stats that are strongly inversely correlated with average fastball velocity. Only BB/9 and BABIP are poorly correlated with fastball velocity.

by spurdynasty on Jul 1, 2009 11:23 PM CDT up reply actions 0 recs

Thanks,

but I’m not sure how to interpret correlation coefficients of moving averages.

I think it’s fine to potentially look for the correlation coefficients for pitchers >87.5 mph because the ones who DO stay in the majors with fastballs < 87.5 mph are going to likely have survived in the majors through something else besides fastball speed (like deception, good sinker, great other pitches or just plain luck). But that’s just a layman’s POV of correlations.

Do you have the correlation coefficients for the actual data?

Thanks!
R

by Requiem on Jul 2, 2009 12:20 AM CDT up reply actions 0 recs

Rolling averages are often used in business to spot trends in noisy data sets like daily/weekly/monthly revenues, expenditures, share prices, etc. The benefit of rolling averages is that they smooth out results by reducing the effect that a few outliers can have. As you can see from the graphs in the article, rolling averages reveal that, on average, pitchers with higher velocity fastballs have higher K rates and lower BAA, HR rates, and ERAs.

That does not mean that all pitchers with high velocity fastballs have high K-rates and low BAA, HR rates, and ERAs. There are outliers. Justin Verlander, despite a 94 MPH fastball, had a 4.84 ERA in 2008 due primarily to a 65% LOB-rate. At the other end of the spectrum, Justin Duchsherer’s 86 MPH fastball contributed to a 2.54 ERA due primarily to a .240 BABIP.

Correlation coefficients are great when comparing data from very large populations or when examining the results of well-controlled experiments. Correlation coefficients are less informative when using smaller populations as is the case with the 106 pitchers used for this study because a few significant outliers can greatly influence the outcome of the calculation. Correlation coefficients for rolling averages work the exact same way as they do with data from individuals, though the calculated coefficients for rolling averages are invariably higher because the effects of outliers are suppressed due to the fact that their results are combined with the results of nine other pitchers.

With that explanation, hopefully you will accept the correlation coefficients for the data from the individual pitchers with a grain of salt.

Fastball velocity vs ERA: r = -0.39
Fastball velocity vs FIP: r = -0.49
Fastball velocity vs K/9: r = 0.57
Fastball velocity vs HR/9: r = -0.40
Fastball velocity vs BAA: r = -0.39

by spurdynasty on Jul 2, 2009 10:45 AM CDT up reply actions 0 recs

Thanks!

I can see that rolling averages would be useful to be able to grok the information in an easily viewable form. I was just wondering how to use correlation coefficients of the rolling averages since I’ve never used/seen them before.

Correlation coefficients still have a use even in smaller populations like this. For example, the correlation coefficients you post still show a significant correlation between fastball velocity and the pitcher metrics. Much higher than, say, BABIP shows year over year.

So, there is obviously some connection between fastball velocity and performance stats.

I wonder if we were able to get minor league data whether the correlation would be as strong. Also, maybe the a more interesting thing to focus on over Fastball velocity vs. BAA would be Fastball velocity vs. BABIP since as you noted Fastball velocity vs. BAA is dependent upon K/9 which we already know has a significant correlation.

Thanks for the well presented and interesting article!

by Requiem on Jul 2, 2009 2:53 PM CDT up reply actions 0 recs