The Saber's Edge: Analyzing Fastball Velocity

At some point during spring training, I planned on writing about how preseason velocities compared to the regular-season values. What I didn't know was Major League Baseball would change their velocity baseline, thereby increasing all the new readings.

The change is important to understand. Once understanding the change, I could calculate how much pitchers maintained their velocity from spring training to the regular season.

I have reported spring training fastball velocity changes for years. Velocity stabilizes almost immediately, with a 1 mph drop equaling a ~0.25 increasein ERA. With this fast stabilization rate, velocity is an early season stat owners can utilize.

At the regular season's start, FanGraph's Dave Cameron wrote about pitchers throwing harder. Almost immediately, MLB Advanced Media's Tom Tango posted an article about the adjustments needed to correct fastball velocities.

The new values reported would be calculated at 55 feet from home plate instead of 50 feet. This difference worked out to a 107 percent increase (~0.6 mph on 90 mph pitch) compared to previously reported values.

With the change, fantasy owners need to put into perspective historical and current fastball speeds. I think most websites will eventually bump up previous values to match up with current values sometime this season or during the next offseason. The current reporting system isn't going to revert back to the slower values, so it's time to embrace the abrupt change. The confusion level will be high this season as everyone adjusts.

Once 2018 begins, the comparison will be between two equally scaled values. The old, lower values most likely will be ignored since people likely will only compare the most recent season's values. With that mess out of the way, it's time to see how spring training velocities hold up.

Spring Training Velocities

Since spring training velocities have been reported, fans have assumed pitchers ramp up their velocity. This assumption is wrong. Pitchers increase their velocity before spring training and continue the increase into the early bullpen sessions. Instead of velocity, pitchers are building up top velocity endurance from a couple dozen pitches to more than 100 for the season's start.

To show this idea, I compared velocity from 44 pitchers from an early spring training appearance to one later on. On average, there were 15.5 days between the appearances and pitchers saw a drop of -0.064 mph (median = -0.1 mph). Basically, unchanged. A small drop is not surprising with the starters going all out in their short early starts but holding back in longer later starts.

With pitchers not picking up velocity during spring, maybe they pick it up to start the season? I compared the values I collected this spring to the pitchers' regular-season values thus far.

After adjusting some spring training values because the parks used the old Pitchf/x (slower) values, I calculated the difference. In all, I compared the last reported velocity for 68 pitchers to their season average. Their velocity dropped -0.23 mph on average (median = -0.25 mph). Again, spring training velocities decreased, not increased, as the pitcher got closer to the season.

While the average change wasn't much, the standard deviation between the values was 1.11 mph. So ~32 percent of the pitchers will see their velocity change more than 1 mph than their spring training value. Pitchers can go into the season with significantly different velocity, but we have no idea which way. The velocity could drop just as much as it could rise.

Knowing the above information, here are some pitchers who've seen significant velocity changes this spring training and regular season.

Shelby Miller: (2016 velo: 93.0 mph, 2017 spring training velo: 96.3 mph, 2017 regular season velo: 95.7 mph)

I knew from tracking the spring training velocities that Miller was throwing harder. With just that information, I took a chance he could get back to his 2013 production level and picked him up in Tout Wars. Most owners have kept their distance from Miller after his 17-loss season in 2015 and his demotion-filled 6.15 ERA in 2016. The start to his 2017 season hasn't been better with both his ERA and walk rate at 5.06.

Hope exists because the added velocity has bumped his K/9 to 11.8 (career high) and his SwStr% to 11.4 percent (2nd). I have not had the nerve to start him yet in Tout Wars, but I actually had to think about it last week.

Sam Dyson ('16 velo: 95.3, '17 ST velo: 94.2, '17 RS velo: 94.3 mph)

Dyson's a mess. His ERA is a 36.00. His K/BB ratio is 0.33. His HR/9 and K/9 are equal at 4.50. These are the signs of a lost pitcher. Dyson has never thrown in the 94 mph range, so his talent level is a little unknown. I'm not sure he will keep his job if he melts down again.

Jake Arrieta ('16 velo: 93.7, '17 ST velo: 91.8, '17 RS velo: 91.5)

I wonder how much longer Arrieta can effectively pitch with the velocity drop. Here's his average velocity the last seven regular season months (thanks to BrooksBaseball.net).

April '16: 94.9
May '16: 94.3
June '16: 95.2
July '16: 94.2
August '16: 94.7
September '16: 93.2
April '17: 91.6

He's down more than 3 mph from last April. No one is worried about the decline because Arrieta's ERA is at a 2.08 and his strikeout rate is at 11.1 K/9 (career high). The strikeout rate will drop with the slower fastball. When his fastball velocity averaged 92.4 mph in 2011, his strikeout rate was only 7.0 K/9. I think this is the perfect time to sell Arrieta at or above his preseason value.

Tyler Chatwood ('16 velo: 92.2, '17 ST velo: 94.2, '17 RS velo: 94.5)

I don't like recommending Colorado pitchers but don't forget about Chatwood's road starts while streaming pitchers. The extra velocity has helped push his early season swinging-strike rate (9.7 percent) and strikeouts (9.5 K/9) to career highs. He's not an every week play, but he has sneakingly good skills in deeper leagues.

Brandon Finnegan ('16 velo: 91.7, '17 ST velo: 94.1, '17 RS velo: 94.1)

Finnegan seemed to reinvent himself this offseason. Owners need to ignore what they previously thought of him because nothing is the same. First, he's dropped his release point a quarter of a foot. His fastballs are coming in 2.5 mph faster. He's throwing his four-seamer more (historically an extreme flyball pitch) and two-seamer less (not a good sinker, no groundballs). He's finally using his change (17 percent SwStr%, 54 percent GB%) more than his slider (14 percent SwStr%, 37 percent GB%).

All these improvements point to a potential breakout candidate. Be careful of old reports. Just recently, ESPN's Keith Law was asked about Finnegan and said, "I think he ends up in the bullpen." I like Keith's work, but he is going off old reports and looks. Every season a pitcher has the potential to breakout, and Finnegan could be that guy this year.