Tuesday, July 23, 2024

An Introduction to pwOBACON+ (Hitters)

I’m proud to introduce a new predictive hitting stat: pwOBACON+.

Some of you may remember back in April when I wrote up a stat called pDRC+.

pwOBACON+ is constructed in a similar manner.

pwOBACON+ variables

  • Non-bunt wOBACON+
  • Barrels/BBE%
  • Average exit velocity
  • Sprint speed

pDRC+ variables

  • HBP%
  • BB%
  • K%
  • Barrels/PA%
  • % of BBEs that are ground balls or pop ups that leave the player’s bat at an EV of less than 90 mph
  • Average exit velocity
  • Sprint speed

You may be wondering why pwOBACON+ does not consider strikeout rate, walk rate, and hit-by-pitch rate. The reason for this is simple: pwOBACON+ is only seeking to predict how successful a hitter is when they make contact. pwOBACON can be used in place of wOBACON (combined with K%, BB%, and HBP%) to arrive at a predictive weighted on-base average (pwOBA).

Some notes

  • I do my best to accurately exclude bunts for wOBACON+ values
  • Average exit velocity for years 2015 to 2019 may include a couple of bunts for some players (the vast majority of them are left out because Statcast failed to track those exit velocities; the 2020 ones should have no bunts included, as it appears the custom leaderboards on Baseball Savant report non-bunt average exit velocities)
  • The BBE denominator for Barrels/BBE% is all non-bunt batted ball events (for pDRC+, it was as a percentage of plate appearances)

pwOBACON+ was created using a linear model with non-bunt wOBACON+ in season n+1 as the y-variable to maximize its predictiveness. I prefer a bigger range of outputs, so then I regressed the fitted values against non-bunt wOBACON+ in season n (the ordering of all pwOBACON+ values is not affected by this process).

The way pwOBACON+ works is easy to comprehend.

If a player has a league average exit velocity, league average sprint speed, league average barrel rate, and league average wOBACON+, that player’s pwOBACON+ will be league average (100).

What happens to that player’s pwOBACON+ if I change one thing?

  • his sprint speed is now 30 ft/s >>> increases to 105
  • his average exit velocity is now 20% above average (back to average sprint speed) >>> increases to 130
  • his barrel rate is now 75% above average (back to average exit velocity and average sprint speed) >>> increases to 110

In short, pwOBACON+ adjusts a player’s wOBACON based on his sprint speed, average exit velocity, and barrel rate in order to best predict future offensive performance.

Looking at pwOBACON+ next to TrueHit, Connor Kurcon’s hitting metric, allows one to best predict a hitter’s wOBACON+ in season n+1. The same thing can be said about Steamer projections.

Here’s how my metric did compared to xwOBACON (which is descriptive) and wOBACON (real stat) for consecutive player-seasons of at least 300 plate appearances (2018-2019 | out-of-sample testing)…

R^2 to non-bunt wOBACON+ in 2019

  1. pwOBACON+ (0.4518)
  2. xwOBACON+ (0.3821)
  3. wOBACON (0.351)

pwOBACON+ is the most reliable of the three metrics as well.

R^2 to itself

  1. pwOBACON+ (0.6222)
  2. xwOBACON+ (0.5237)
  3. wOBACON+ (0.351)

Hitters w/ the highest pwOBACON+ last year

  1. Aaron Judge (142)
  2. Miguel Sano (141)
  3. Nelson Cruz (135)
  4. Mike Trout (133)
  5. Christian Yelich (132)
  6. Yordan Alvarez (131)
  7. Brandon Lowe (128)
  8. Jorge Soler (127)
  9. Fernando Tatis Jr. (125)
  10. Gary Sanchez (125)

The full 2019 leaderboard can be found here.

I hope you enjoyed reading this article. If you have any questions, feel free to comment below or direct message me on Twitter (@MaxSportsStudio).

Featured Image: Wikipedia