What Does A Pitch’s Run Value Actually Mean?

Baseball Savant is full of funny colors and statistics, and the website gets a little more crowded every year as more and more niche information becomes publicly accessible. Some information is self-explanatory, while other information is anything but. For example, when we look at Freddy Peralta’s page, we see this:

On the left, you see his high-level numbers from the past few seasons, a good overview of his recent performance. These figures can easily be found on other sites like Baseball Reference and FanGraphs, so it’s not typically what Savant is used for, but it still provides a necessary baseline for analyzing the rest of his profile.

On the right is a relatively new addition that illustrates the movement and usage for every weapon in his arsenal, along with his arm angle. Interpreting it can take a little getting used to, but once you’re familiar with general ranges for pitch movement, you can more easily spot which fastballs have more ride, which sliders are more like sweepers, and so on and so forth.

In the middle are the sliders that all armchair GMs (myself included) know and love. They contain percentile numbers for key qualities like average exit velocity, strikeout rate, and ground-ball rate. This gives users a normalized view of how a particular metric fits in with the rest of the league. While the bottom section covers more granular details, the top section labeled Value contains four sliders, each measuring something called “run value.” It’s pretty easy to understand that more run value is “better,” but how is that even calculated—and is it really that simple?

The website’s own definition of the metric does little to clear up the uncertainty, but it also links to an old blog post by Tom Tango, a pseudonym for a sabermetrician whose real name isn’t publicly known despite him having an outsized impact on the analytics community. In this post, he explains that run value is essentially the cumulative sum of every pitch’s value from the run expectancy matrix. These values from the run expectancy matrix can be interpreted as the average number of runs expected to score from a given base-out state. For hitters, it looks something like this (the matrix for pitchers is the same but inverted):

Based on these broad-strokes descriptions, we have evidence to support our original conclusion that more run value is “better”—akin to other counting stats, like strikeouts. However, keen-eyed Savant users will notice some discrepancy between the run value of a specific pitch and its other numbers.

Despite Peralta's slider having significantly less run value than his fastball, it has a lower batting average, slugging percentage, and wOBA while having better whiff and strikeout rates. How can both stories be true?

It’s important to remember that while many numbers may seem related, they’re calculated very differently. A simple example of this is batting average and on-base percentage. One might assume both stats to have the same denominator, but they don’t, since there’s a slight difference between at-bats and plate appearances. In this case, batting average and slugging percentage, strikeout rate and wOBA only consider the plate appearances that end on that pitch. We want to know the value of each individual pitch, even if it isn’t the last one thrown in a given plate appearance, which is where run value comes in.

What we do know is that when his sliders are good, they’re really good, often resulting in swings and misses, punchouts, or soft contact, but what about the other kind? The only way for Peralta’s slider to impact his run value negatively without also dragging down the other numbers listed above would be throwing a lot of balls. If you look at the run expectancy matrix again, you can see that getting into hitter-friendly counts like 2-0 and 3-0 increases the probability that a run will score that inning and thus adversely affects the metric for the pitcher. This makes sense, as getting behind in counts gives pitchers less wiggle room and results in more walks and hittable pitches. 

Consider this hypothetical possibility. Against the first batter of the season, Freddy Peralta throws three consecutive sliders, each resulting in a ball. Now down 3-0, he throws yet another slider that ends up as a groundout. Now let’s assume he repeats this several more times with a few walks mixed in, and you can see how he could have a perfect batting average of .000, while tanking the run value on his slider. Along with the other data, run value gives us a more complete understanding of a given pitch’s true effectiveness. 

This real-life example of Freddy Peralta’s slider is fairly accurate. It’s great when commanded well, but too often, it ends up as a waste pitch, reducing its overall contributions to his game. Part of this is the result of how much he toys with the shape of the pitch, and part of it is the simple difficulty of being a big-league pitcher. Either way, it’s a clear roadblock to him reaching the peak of his potential.

With our new knowledge, we can attempt to ascertain what’s going on with the rest of his arsenal, as well. His fastball sometimes gets hit hard and isn’t missing as many bats as he may like, but it’s still getting him more strikes than anything else in the repertoire. His changeup gives him the most bang for his buck, as it combines the best qualities of both his fastball and his slider: infrequent waste pitches and strong results on balls in play. His usage ticked up a bit from 2023 to 2024, and we may see it even more frequently next year. 

Every statistic in baseball was created to answer some specific question. Unfortunately, the path to answer one question often opens up the door to two more, so never over-index on just one fact or figure. Instead, acknowledge that every number is incomplete and has its own strengths and weaknesses. Run value may seem like a way to quickly judge the effectiveness of a pitch, but it’s far more layered than that. It’s a useful tool on the ol’ sabermetric belt, but it shouldn’t be the only one you bring to work.