Introducing Speed Unit

I previously looked at how aggressive or not the Brewers have been this year at taking the extra base. But that doesn't tell the whole story, as it does not account for stolen bases. And I have a few more questions about some of these baserunning statistics.

As mentioned before, the baserunning values listed on Baseball Savant are:

Quote

"Based on inputs including runner speed, outfielder throwing arm, runner position on the base paths, and outfielder distance from both ball and bases, an estimated success probability can be created for each opportunity. With that information available for each play, the player’s actual success rate can be compared to the estimated rate and the cumulative metric can be created, accounting both for bases taken (prevented, for outfielders) and chances taken or not. This does account for extra bases taken by batters or runners on batted balls; it does not include stolen bases, as it’s about taking extra bases against fielders."

More on the stats at Tom Tango's blog.

Fangraphs has their own version of baserunning value which is calculated using weighted stolen base runs (wSB), Ultimate Base Running (UBR), and weighted GDP (wGDP). The UBR portion of Fangraphs baserunning, created by Mitchel Lichtman, is most closely related to the Baseball Savant calculation, but it does have a different method.

Quote

"To calculate UBR you need to know the average run expectancy for a given batting event given the base-out state and then the actual run expectancy change that occurred during the play."

So while Baseball Savant uses Statcast metrics of runner speed, arm strength, and runner position, UBR uses play-by-play information of changes in run expectancy due to the runner advancing.

While both of these statistics can be useful for comparing modern baserunners, what is an analyst to do when neither Statcast nor play-by-play data is available? What if you only have basic statistics like PA, 1B, 2B, 3B, HR, SB, and CS?

Well fortunately, our ancestors had to deal with such a problem and created a solution. One of those people was Bill James, who invented Speed Score back in 1987. It takes six factors--stolen base percentage, stolen base attempts, triples, runs scored, grounded into double plays, and defensive position and range to calculate an estimate of a players speed. That value can also be found on Fangraphs as "Speed."

In 2005, Bill James' Speed Score was revamped by 'Patriot' into Speed Unit.

Quote

"Speed Unit uses four categories: Stolen Base Frequency(SBFrq), Weighted Stolen Base Percentage(WSB%), Runs/Time on Base(R/TOB), and Triples/Balls in Play(T/BIP). Each category is evaluated against the league average and the standard deviation, in an attempt to convert it into a z-score. Then the z-scores in each category are summed and converted into a total score."

2024 Brewers

We can take all of the various implementations of speed to get a sense of the current Brewers roster:

Data from Fangraphs here and here and Baseball Savant.

One thing to note is that Fangraphs Base Running Runs are counting stats, while the Bill James and Patriot Speed Scores are rate stats. So where Oliver Dunn has only been worth 0.8 runs on the bases this year, he has performed well when given the opportunity.

There is quite a bit of disagreement between UBR and Statcast on Yelich, Frelick, and Perkins. From the previous analysis, there were differences in aggression between these players and perhaps they need to still find a happy medium. They might be good at knowing when to go, but getting caught at the wrong time and taking away win probability from the team on the bases. Three of the young speedsters--Chourio, Frelick, and Perkins--have lower stolen base value than expected. Joey Ortiz has the sprint speed to be a good baserunner (28.3 ft/s, 78th percentile), but none of his numbers are all that great.

Comparison

If we zoom back out to the entire Major Leagues, how do these stats compare to each other? In particular, how well does Speed Unit estimate a statistic closer to true baserunning value like Fangraphs Base Running? If we look at the statistics for the 395 players with at least 500 PA from from 2021 through 2023, we get this relationship, with a correlation coefficient of 0.84:

The obvious discrepancy between the two statistics was already mentioned. Speed Unit is a rate stat and Base Running is a counting stat. Players with fewer plate appearances will have fewer opportunities to add or subtract runs by nature. Another big limitation for Speed Unit comes from extreme pinch runners. For most players, their normal statistics at the plate will balance pinch running appearances out, but there are some outliers. Terrance Gore and Billy Hamilton are recent examples of specialty pinch-runners. The speed unit calculations all compare their stolen base and runs scored values with their hitting numbers, but these players' ratios are off.

Still, when setting a high enough value on plate appearances and looking over a large enough sample, the simply calculated Speed Unit can describe upwards of 85% of the variation of Fangraphs Base Running. Not bad for scratches on an ancient tablet.

Historical Brewers

After finding the value and limitations of an estimator like Speed Unit, we can do the fun stuff. Now that we're no longer limited to the modern Statcast era, play-by-play results, or even the Major Leagues, who are some historical players in the Brewers organization who have had notable speed seasons? Here are the top 10 in the Major Leagues since 1978, led by Scott Podsednik's 70 stolen bases and 7 triples in 2004.

But we're not limited to the Major Leagues, how about the best Speed Unit seasons for the Brewers in the Minor Leagues?

Finally, we can look at the same two tables but find the top 5 Speed Seasons by decade since 1978 for both the Major and Minor Leagues:

What do you think, any interesting names? Any notable seasons missing?