How Pitcher Quality Impacts xwOBA

Weighted On Base Average (wOBA) is widely considered one of the top metrics to approximate overall batting ability. wOBA is made by calculating how each hit outcome (singles, doubles, etc.) impact how many runs a team will score. Weights are then applied to each outcome similarly, but more accurately, than slugging percentage does. When it is all said and done, a league average wOBA is always somewhere around .320. The best hitters can get up to around .400, and the worst sit around .280. Team wOBA is far more predictive of team runs than batting average, on-base percentage, or slugging percentage.

Now that we’ve established that wOBA is important, you can understand why people have spent so much time trying to predict it. The first metric, and possibly most used is expected weighted on-base average (xwOBA). Like other expected stats, xwOBA uses batted ball data to find the probability of each batted ball being an out, single, double, triple, or home run. These possible outcomes combine to give an xwOBA value for that plate appearance. Naturally, xwOBA is more predictive of future wOBA than wOBA because a player can get extraordinarily lucky with balls dropping in, bad defense, or a number of other reasons. There are many factors that people combine with xwOBA to get an even more predictive model, such as an aging curve, but one that many people do not talk about is pitcher quality.

One thing to note before I go on, the dataset I used for this analysis included every at bat from 2016-2020, then it filtered out batters who did not have 80 at bats. To determine pitcher quality, I collected the data for every pitcher over the last five years, and recorded what their xwOBA was each of those years. Next I matched pitcher xwOBA with each at bat and took the average xwOBA against for each hitter.

The first thing you have to question is how random pitcher quality is. It makes sense that better hitters would face tougher pitching, but to what extent is that true? As you can see by the following graph, not that much. The correlation coefficient is less than .08, so we know that pitcher quality is at least somewhat random. It is also important to understand that players from the same teams will a lot of times face similar pitching for obvious reasons. This was amplified during the 2020 season where all of the games were between division teams and one division in the other league. This clearly showed up in the 2020 data, where nine of the thirteen hitters who faced the worst pitching were on the Braves. You could also argue that because players who stay on the same team are likely to play a similar schedule and face similar pitchers the next year, pitcher quality is less random.

The fact that pitcher quality seems to be mostly random is quite useful. It means that there are hitters who get lucky and unlucky every year. Being able to identify players that consistently squared off against better pitching more often, can help find positive regression candidates.

To find if any of this meant anything in terms of predicting future season xwOBA, I decided to run a linear regression. In an attempt to keep it simple and isolate my measure for pitcher quality my regression equation was just:

**NextSeasonwOBA = PreviousSeasonwOBA + PreviousSeasonPitcherQuality**

I compared this equation to just using the previous season to predict wOBA for the next year. The model that included pitcher quality had a higher adjusted R-Squared value, and it turns out that my measure for it was negative, and statistically significant at the 10% level. This means that as pitcher quality increases for the previous season, next season’s wOBA should also increase.

Yandy Diaz is a good example of the impact that this can have on a player. In 2019, he got very lucky with what pitchers he faced. The average xwOBA against of the pitchers was an absurd .384. Diaz faced the easiest pitchers by far; the next closest was Hunter Pence at .355. In 2019, Diaz had a very good season, and ended the year with an xwOBA of .362. In 2020 however, the quality of pitching against him was much more normal (xwOBA of .305), and so was his xwOBA, which fell to around league average (.322). Of course, there are many other factors that played a role in his xwOBA decreasing, but it is interesting that the hitter who faced the easiest pitchers in 2019, then much more normal pitchers in 2020 would have his xwOBA drop by .40 points.

This same process can apply to pitchers as well, and once again quality of hitters faced will make a difference in year to year xwOBA. The difference that it makes isn’t huge, and it definitely isn’t as important as other factors, but it still shouldn’t be forgotten. The simple hitter model I made estimates that a difference in .6 difference in pitcher xwOBA, which is what the range was this year, impacts next season xwOBA by 11.5 points. That might not sound like a lot, but in 2020 that’s the difference between Francisco Lindor (.333) and Mitch Moreland (.322). If that is a difference that you think is important, adjusting for pitcher quality might be for you.

- Yakkertech and PitchCom Join Forces In Exclusive Deal to Quantify Pitch Command
- Introduction to bcWAR: SEC Position Player Rankings
- The Jordan Hicks Dilemma: A Quantitative Analysis of Pitch Tunneling, Arsenal Coherence, and Stuff-Based Evaluation
- Believe the Breakout? (Pitcher Edition)
- BCTeam Update: Pitch Outcomes Tab

- February 2023
- October 2022
- September 2022
- August 2022
- July 2022
- June 2022
- May 2022
- April 2022
- March 2022
- February 2022
- January 2022
- December 2021
- November 2021
- October 2021
- September 2021
- August 2021
- July 2021
- June 2021
- May 2021
- April 2021
- March 2021
- February 2021
- January 2021
- December 2020
- November 2020
- October 2020
- September 2020
- August 2020
- July 2020
- June 2020
- May 2020
- May 2017
- April 2017

All Rights Reserved. 2023

%d bloggers like this: