Mixing Fantasy & Reality: Pitch Velocity Changes & Notes
Pitch Velocity Effects on Groundballs, Exit Velocity, and Swinging Strikes
Last week, I examined Danny Duffy and several other writers have examined at him also. If you want to read up on various theories on my he is performing great and why that may change, go ahead. Instead, today I am going to concentrate on his fastball velocity changes and how the results change as the velocity changes.
Danny Duffy is starting to get some Cy Young consideration after spending part of the season in the bullpen. One cause for the turnaround is his fastball velocity increasing from 93.8 mph to 95 mph. The average velocity was even higher earlier in the season but it has been steadily dropping.
So what difference does it make if he is throwing 96 mph or 94 mph? Today, I am going to lay the groundwork for finding such an answer.
Simply, I looked at three different factors, exit (or batted ball) velocity (EV), groundball rate (GB%), and swinging-strike rate (SwStr%) and how each compared to a 1 mph velocity block. To help smooth out the results, I looked velocities between whole values like 90 mph to 91 mph and labeled them 90.5 mph. Also, I looked at values between 90.5 mph and 91.5 mph and put them in the 90 mph bin. I know there is overlap, but I hoped the higher number of samples would help smooth at the final results, especially with a limited number of samples at both ends of the data range.
To start off, here are Duffy’s average exit velocities for a given range of fastball velocities.
Duffy sees his average exit velocity dropped steadily as his velocity increases. So the harder he throws, batters will weakly hit his fastballs.
Here is his velocity compared to his groundball rate. Just so everyone knows, Duffy is one of the most extreme flyball pitchers in the league. He should be attempting to keep his pitches at this high flyball rate to help limit his BABIP.
I am not 100% sure how to read this information, but it looks like Duffy wants to increase the number of easy pop-outs, he can lower his velocity. It is a tough trade off.
Now, moving onto his swinging-strike rate versus his velocity.
The slope of the graph is pretty steep with almost a tripling of his SwStr% from 93 mph to 96 mph. Duffy’s fastball definitely performs better at higher velocities.
So how do this match up with results?
| Dates | Fastball Velo | SwStr% | FB% | Exit Velocity |
| 5/15/16 to 7/7/16 | 95.4 | 12.9% | 46.5% | 89.4 |
| 7/16/16 to 8/16/16 | 94.0 | 12.6% | 48.5% | 90.2 |
The results are all heading in the expected direction, but I thought there would be a little bit bigger change in his SwStr%. Duffy’s swinging strike rate will likely to continue to drop has his fastball velocity drops. Additionally, he is going to allow more and more hard hit flyballs (i.e. home runs) but will likely see a lower BABIP.
I am just beginning to analyze data with this method, so I have no idea on league-wide averages, stabilization rates, or year-to-year correlation. What I do know, is that I look the results so far and plan on using it with future velocity changing pitchers like Dylan Bundy and Michael Pineda. Let me know if you have any questions and pitcher samples you would like to see.
Notes (These got a little out of hand in length. I will try to keep them shorter next time):
On Wednesday, I warned people about using Tyler Skaggs in his start because he would need to work with a new stretch move. The Angels talked about it some.
It was a matter of importance to the organization. Manager Mike Scioscia promised it would be immediately addressed. Asked if it had been fixed Wednesday afternoon, before Skaggs’ follow-up start against Seattle, Scioscia described the solution as simple and himself as confident the 25-year-old would “do a better job of it.”
“It’s just experience,” Scioscia said. “Sometimes a blessing comes out of when you have a day like he did in Cleveland.”
Scioscia said he and his staff had recognized that Skaggs’ difficulties condensing his delivery could cause challenges. His two-year recovery from Tommy John surgery, he said, simply took precedence over more minor aspects.
But now that the problem has presented itself, Scioscia said it should not again. “In fact,” he said Wednesday, “I know it won’t be an issue. It’ll be a quick fix.”
The fix didn’t seem to happen last night with Skaggs giving up four runs in only 3.1 innings of work. I would still stay away from owning him for a bit longer.
• There seems to be some disagreement on how to value Jon Gray with some pundits just looking to stay away like they previously did with all Colorado pitchers. This approach is not the preferred method in my opinion. The best way is to come up with educated projection, find similar players, and slot the player around these similarly valued players.
For next season, I would think 190 innings of 9 K/9 and 3 BB/9 baseball is reasonable starting spot for Gray. I think he should be able to throw a full 200 IP workload, but his innings may get cut short because of some rough starts at home. Looking at his current WHIP of 1.25 and .299 BABIP, I could see his BABIP increase and WHIP go to 1.30.
Finally, his ERA. All his current ERA estimators are in the high threes. 3.83 FIP, 3.70 xFIP, 3.79 SIERA, 3.73 kwERA, but his ERA his sits at 4.69. Colorado pitchers are usually going to have an inflated ERA because of the extra hits which fall in Coors, but probably not to the 4.69 level. I will go with a 4.25 ERA for now.
Note: I have started collecting all my preseason projections in a spreadsheet for reference. I know there is only two players on it, but I hope to fill it up before next season.
Some pitchers who are putting up similar numbers this season are Kevin Gausman, Gio Gonzalez, and Scott Kazmir. If these three pitchers are owned/valued in your league, so should be Gray. If they aren’t, maybe stay away from Gray.
• Yasmany Tomas is slugging .289/.319/.867 this August. I went looking through his data and found his batted ball distance to be the same, but he is pulling the ball over short left field fences more and more.
Month: Pull%
Apr: 33%
May: 44%
June: 44%
July: 48%
August: 54%
This increase in pull rate, and a small drop in his K%, makes him an interesting value pick for next season.
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 four FSWA Awards including on for his Mining the News series. He's won Tout Wars three times, LABR twice, and got his first NFBC Main Event win in 2021. Follow him on Twitter @jeffwzimmerman.
Nice analysis on Duffy, but there’s only one hole in it – situational stats. It would be great to see a cross-application where you only compare changes in velocity while he was a starter. When you’re only pitching 1-2 innings at a time, you can easily overpower hitters for a few minutes. But when you’re trying to preserve yourself to throw 90-110 pitches, completely different approach. Unfortunately, his 96 MPH starts were few, so not sure if sample size becomes a problem if we slice by role. What do you think?
Yes and no. Pitchers ramp up velocity in the bullpen, so the higher velo’s would only exist then. Maybe, the best option is two lines.
Jeff, I’m curious about the actual velocity data. The last Duffy article had me curious about his velo because the numbers on Fangraphs really don’t seem to jive with what I’ve seen watching his starts. That led me to Brooks Baseball, where it looks like there is quite a bit of variation in average velo for Duffy relative to Fangraphs, in some cases over 1 full MPH difference from game to game. I understand Brooks has a system in place to correct errors in pitchfx data, but does that mean the velocity readings there are more accurate or just that they’re measured differently (release point vs crossing the plate)?
I ask because if you use Brooks’ velo data for Duffy, his average velocity from 7/16 through today is 94.82, maxing out at 97.56. Obviously not a huge difference from 94.0, but if I am reading the graphs right, there is a huge spike in swinging strike rate right at that point. From 5% at 94.0mph to over 10% at 94.8mph. If Duffy’s actual velocity is sitting in the 94-97 region instead of 92-95 as Fangraphs’ data seems to suggest, it seems like the data may be skewed. The velocity decline is still there, but a decline to 94-97 is a lot different than a decline to 92-95.
I also think Duffy’s velocity decline is less important than it is for other pitchers. He has an 11-12mph difference from fastball to change, a difference he’s maintained even as his fastball velo dipped. As the whiff rate on his fastball went down, the whiff rate on his changeup and slider have increased. Some of that is skewed by the absurd 16 K start, but I watch Duffy and see a guy making a conscious effort to be more efficient and go deeper in games, which likely means less max effort pitches. He’s averaging over 7 IP per start in that 7-16 to 8-16 time frame after averaging 5 2/3 over his first 11 starts.
He’s still “dialing it up” to 97+ when he really needs to miss bats but he looks like he’s mainly trying to cruise and let the defense work. He’s probably in a better position than most to get away with that as a fly ball pitcher in a home park that suppresses homers and an elite outfield defense behind him. Of the 7 starts he’s made from 7-16 through 8-16, 4 were at home, 2 were in Detroit and the other at Tampa Bay (16 K). My working theory is that his velocity decline is more intentional than it is fatigue, recognizing the parks and defense.
Purely conjecture on my part, but it is something I’ve been mulling over.
Glad you brought this up. Fangraphs itself has stated that Brooks is more accurate yet no one seems to use it because it is not an easy export to a spreadsheet? The data is so skewed it is actually pretty crazy. Thankfully they appear to move in the same direction though. Jeff can you help us on this?