New pitch tunneling metrics and Luis Severino's offseason focus

There are all sorts of adjustments that players make in the offseason. Often times, we hear the trite "I'm in the best shape of my life". In other instances, a player attempts to improve something more measurable or readily apparent, such as Aaron Judge trying to cut down on strikeouts. Luis Severino's winter plan touches on both of those categories, though the latter is far more interesting (to me).

Brian Cashman asked Severino to improve his flexibility this offseason. That's fine and dandy, but it's next to impossible for us to determine how much that could potentially help. On the other hand, we can get an idea of the upside of another part of his winter plan: to become more consistent with his release points. In talking to NJ.com's Brendan Kuty, Severino highlighted his trouble with his arm slot in 2016:

"My fastball was all the way over here," Severino told NJ Advance Media, showing wider-than-normal release point.
"But my changeup was over here," he said, his arm dropping even lower. "My slider was over here and then sometimes over here."

Severino wasn't speculating about his release points, as there is data that backs up this problem. Thanks to Baseball Prospectus' latest fascinating method of analysis, pitch tunneling, we can compare Severino's slider and changeup against his fastball when it comes to a couple of key metrics: release point (out of the pitcher's hand) and tunnel point (23.8 ft from the batter, the decision point for a swing). Release point data has been available for a while with PITCHf/x, but the folks at BP have taken further steps in using it in comparative pitch pairs while also layering on tunnel point data.

The ability to compare and contrast two pitches can give us a glimpse into how opposing batters are able to decipher what pitch is coming. And, if we know how easy (or difficult) it is to read a pitch out the pitcher's hand and/or at the decision point, we can get a better understanding of how effective certain pitches are. If Severino grades poorly per these metrics, then it would appear that he's taken the right strategy this offseason. Here's a snippet of the data I pulled from BP comparing the Yankees' 22 year-old hurler to his counterparts:

Per Baseball Prospectus, this stat measures the average variation of a pitcher's release point (in inches) of a pitch pair. The smaller the better, as it is tougher to decipher the difference between pitches with smaller variation. (FA=Fastball, CH=Changeup, SL=Slider)

Per Baseball Prospectus, this stat measures the average variation of a pitcher's release point (in inches) of a pitch pair. The smaller the better, as it is tougher to decipher the difference between pitches with smaller variation. (FA=Fastball, CH=Changeup, SL=Slider)

Per Baseball Prospectus, this stat measures the average variation of a pitcher's tunnel point (in inches) of a pitch pair. The tunnel point is when the batter must decide whether or not to swing. The closer the better, particularly if there is significant break differential after the tunnel point.

Per Baseball Prospectus, this stat measures the average variation of a pitcher's tunnel point (in inches) of a pitch pair. The tunnel point is when the batter must decide whether or not to swing. The closer the better, particularly if there is significant break differential after the tunnel point.

First, a look at Severino's changeup. Conventional wisdom says that in order to have a good changeup, a pitcher must have arm speed and a release similar to his fastball. That way, the pitcher can trick his counterpart into thinking he's throwing a fastball, only to be baited into swinging too soon. Some movement helps, too. Without similar arm speed or an inconsistent release point, a changeup isn't deceptive. In other words, as a drastic example for illustrative purposes, if a pitcher throws a 95 MPH fastball over the top and throws an 80 MPH changeup from three-quarters, the batter shouldn't struggle with the velocity difference because the release point is apparent.

Per BP, Severino's release point variation between a pair of a fastball and changeup was around five inches in 2016, about two inches more than league average. While five inches of separation seems difficult to decipher at game speed, we have to remember that we're mere mortals when it comes to Major League hitters. They can pick up on it. Despite the tunnel point variation being better than the average fastball-changeup combination, hitters have the opportunity to identify one vs. the other far sooner (i.e. at the release point).

Luis Severino 2016 Changeup Location Heat Map - via Baseball Savant

Luis Severino 2016 Changeup Location Heat Map - via Baseball Savant

Interestingly enough, his command of the pitch isn't dreadful. In fact, it looks pretty good. He's staying away from the heart of the plate, which is important. However, because his changeup is easier to differentiate than most and thus not particularly deceptive, hitters aren't overwhelmed by the pitch. League average whiff per swing on changeups last season was 31.4%, whereas Severino's mark was merely 20.0%.

Next, the slider. Severino's fastball and slider came out of his right hand closer than the previous pair, at about 3.4 inches of release differential. That was a tad worse league average, but not as awful as his changeup release. Still, though, it was nothing particularly special. That said, unlike a changeup, a slider isn't predominantly reliant on deception. Movement plays a big part in its effectiveness.

The best time to have significant movement on a pitch is after the tunnel point, especially if there is a small tunnel point variation between a pair of pitches (in this case, fastball-slider). I've already pointed out that Severino's fastball vs. slider release point wasn't anything to brag about, but the positive is that his tunnel point differentiation was better than most. Not strikingly better than average, but above average nonetheless. From there, the post-tunnel break (variation in movement in inches from tunnel point to home plate between two pitches) between Severino's fastball and slider was approximately five inches. That's almost a full inch more than the average combination of the two offerings.

To this point, nothing sounds disconcerting about Severino's slider. It would be nice if his release point was tighter to his fastball, but it's seemingly fine where it is. However, moving on to his slider location heatmap, and welp...

Luis Severino 2016 Slider Location Heat Map - via Baseball Savant

Luis Severino 2016 Slider Location Heat Map - via Baseball Savant

That's ugly. It didn't matter how good or bad Severino's release or tunnel differentials were for this pitch. With it constantly over the heart of the plate last season, it wasn't particularly effective. Severino garnered only 26.9% whiffs per swing, well below the league's 34.2% standard for the pitch type.

Thanks to Brooks Baseball, we can compare each of Severino's offerings to others in terms of pitch outcomes. The combination of Severino's pitch tunneling and location should at least partially explain some of the good and bad results shown above.

As it's painfully clear, the young righty's changeup and slider were well below average (index: 100) in whiffs per swing last season. In other words, batters weren't missing the pitch when they choose to hack at it. That's not a surprise for the changeup given Severino's poor pitch tunneling (read: lack of deception). It's not a shock for the slider either, as his command of the pitch was poor.

Severino generated more foul balls than average on those two pitches too, and that makes sense: he's not getting the desired swing and miss from what are supposed to be put away pitches. Again, some of the reason could be that pitch tunnels and location aren't strong for Severino's changeup and slider, respectively. Other bad indicators include his line drives per balls in play, quite high for his changeup and slightly above average compared to the league.

Despite a blazing fastball, Severino's inability to put away hitters with his secondary offerings has hurt his four-seamer. Even if batters can't do significant damage against his slider or changeup (opponents had a wRC+ of 78 against both last year), they are managing to fend off the two offerings until a heater comes. Hence the low whiff per swing and high foul per swing numbers on the changeup and slider. Once Severino turned to the fastball, opposing hitters feasted, posting a remarkable 164 wRC+ last year.

The metrics paint an ugly picture of Severino's secondary pitches for now. The good news is that he and the Yankees' staff are keenly aware of it. Perhaps data similar to this was available to them, which spurred them to address the problem. Regardless of how it was discovered, it's pleasing to see the data support that his release point is a potential opportunity for improvement. How much such adjustments can help Severino, let alone if he's capable of making said adjustments, remains to be seen. It certainly seems plausible that closing the gap between the release of his fastball and changeup could help, though I don't think it's clear how much his slider stands to gain from a release point adjustment. Perhaps it could help his command of the pitch, but it's not an obvious fix. From the sounds of it, Severino is making good progress:

"I'm doing very well," he said. "I've been throwing my bullpen and my changeup is way better than last year. My fastball location is better, too. So hopefully in spring training it'll be good."

Let's hope so.