Introducing Swing Decision Run Value
Drew Haugen makes his Down on the Farm debut with a new hitting metric!
Please welcome Drew Haugen (@Drew_Haugen on Twitter), who is making his first post here at Down on the Farm! Drew’s an analyst who will be covering an exciting new metric that he’s developed here today. Take a look…
A common approach to player evaluation in the public sphere is to use statistics such as wRC+, BB%, and K% for hitters, or FIP, SIERA, and K% for pitchers. Although these metrics are useful in describing a player’s results, they often fail to fully represent a player’s true talent level. Factors such as umpire tendencies, catcher skill, and varying levels of opponent talent are all inherently included in these statistics and yet are all out of a player’s control. To determine a player’s true talent level, process-based statistics are superior to result-based statistics because they more fully represent a player’s actions. However, while valuable, process statistics often do not have as strong a correlation to production as results-based stats.
To address this, I put together a framework for evaluating swing decisions that is both process-based and correlates well to hitter production: SwRV, or Swing decision Run Value. Using expected run value models, I can predict the run value of a swing and a take for each pitch a hitter sees, and debit or credit a hitter based on their decision. If a hitter swings, their SwRV is (expected run value of a swing) - (expected run value of a take), and vice versa if the hitter takes the pitch. This way of looking at plate discipline is valuable because it accurately credits a hitter for how their decision affects run scoring, and because it is based completely on expected results, biases like umpire tendencies are stripped out. This metric is very sticky year over year, far more than BB%.
Additionally, SwRV correlates better to wOBA than any other process-based statistic and is only slightly weaker than BB%, which directly impacts wOBA, even without taking any results into account.
In terms of predicting future wOBA, SwRV is nearly as strong as BB%, and again stronger than any other process-based statistic.
SwRV peaks in both the heart of the plate, where the most damage is done on balls put in-play and where called strike probability is high, and outside, where a swing is not valuable and the probability of a ball is high.
Certain hitters are underrated by their BB%, often because of their tendency to swing at a majority of in-zone pitches, which is valuable but negatively affects BB%. (SD+ is SwRV rescaled on a plus scale, and Correct% is the fraction of pitches where a hitter makes the correct decision.)
Additionally, certain hitters have high walk rates that overrate their true plate discipline, usually because of the high number of pitches that should be swung at that the hitter takes.
A Point in Favor of Swinging Less
There have been many pieces from smart baseball minds such as Eno Sarris and Devan Fink on the values of taking versus swinging, largely due to the fact that hitters produce negative value on swings. Using the SwRV data, just 33% of pitches had a higher context-neutral expected run value on a swing compared to a take, while in reality, the league swing rate was far higher at 48%. Even in-zone pitches should not all be swung at, because a called strike is not as hurtful to as a weakly hit batted ball. Only 64% of in-zone pitches have a higher expected run value on a swing than on a take (again, context-neutral).
Hey there! Bryan’s here with an executive summary on Drew’s insightful piece:
Drew’s new statistic, Swing Decision Run Value (SwRV) is a process-based metric that measures the run value of a hitter’s decision to either swing at a pitch or not.
SwRV is sticky from year to year, as it is process-based and strips out biases and facets that are out of a hitter’s control.
SwRV correlates with same-season and future wOBA more than most other process-based statistics, and almost as well as BB%, which is an input to wOBA.
The run values in SwRV tend to be highest in the middle of the plate and outside the strike zone, as one might assume.
SwRV can help us determine which hitters are being too passive at the plate and being overrated by a high walk rate, as well as help identify lower-walk-rate hitters who are underrated because they are making wise swing decisions.
Generally speaking, SwRV is reinforcing the idea that hitters should be swinging less frequently, even at balls over the plate.
As we’ve said before when rolling out a new metric, we’d like to hear from our readers about where you’d like to see Drew and the staff take this next. What questions do you have about SwRV and how it can be applied to minor-league hitters? Do you want to know more about how SwRV was built, how it can be used, or what you want to see next with this data? Reach out and let us know your feedback!
I read this twice today and then tonight went and made sure my subscription payment method is updated. This is great work and I’m interested to see what comes from the data.
My question is, what is the smallest sample size that would be helpful for this indicator?
If a small sample size is helpful, it would be a good way to gauge how well hitters are doing by month, but if not, we'll have to wait some time.