On Monday and Tuesday, I identified and discussed a slew of hitters who were underperforming or overperforming their xBABIP marks, a backwards-looking metric I developed to try cutting through the noise to determine what a hitter’s BABIP should be given various underlying skills I deemed important. One of those skills was speed. Obviously, a faster hitter has a better chance to beat out an infield grounder for a hit than a slower batter. The problem is, how do I know how fast a hitter is once he makes contact and exits the box en route to first base?
My current equation uses FanGraphs’ Spd metric, which is called Speed Score. Can we do better, though? Two newer metrics are available via Statcast’s Sprint Speed leaderboard — Sprint Speed (in feet/second) and HP to 1B (in seconds). With three metrics in hand, the plan was to test the correlation of each with BABIP.
I began with a population size of 1,249 hitter seasons that recorded at least 300 plate appearances during any year between 2015 and this one. Both Spd and Sprint Speed data are available for each player season, but HP to 1B only goes back to 2017, and is missing for many player seasons, which reduced that correlation’s population size to 707.
Let’s get to the results:
|Spd||Sprint Speed||HP to 1B|
So much for discovering a definitive answer! These correlations suggest that all three metrics explain BABIP to a similar degree, though perhaps you believe the jump to Sprint Speed is meaningful.
The biggest problem is how much a player’s score versus the average can vary across metrics. A player may score well above average in Spd and Sprint Speed, but below average in HP to 1B. It’s difficult to reconcile, but means that picking the right metric could make a drastic difference.
Let’s dive deeper into the metrics themselves.
The full description of Spd can be found here, though no actual formula is shared. However, the components of the equation are — Stolen Base Percentage, Frequency of Stolen Base Attempts, Percentage of Triples, and Runs Scored Percentage.
Immediately, we find flaws in trying to equate Spd with a batter’s speed. Stolen Base Percentage isn’t a proxy of speed at all, but of basestealing ability. We’ve seen time and time again a speedy hitter be a poor basestealer or a hitter with just average speed steal bases at a highly successful clip. In addition, Runs Scored Percentage is reliant on the hitters behind the batter in question. Regardless of speed, that rate is going to be higher with a better cast of characters hitting behind him and knocking him in. So Spd, despite a similar correlation to BABIP as the other metrics, has serious issues as a proxy for a hitter’s speed.
Next, Sprint Speed is “feet per second in a player’s fastest one-second window”, and the Statcast leaderboard page includes a more complete description. This seems more like what we are looking for, but it’s not perfect either. Why? Because it ignores how long it takes for a player to actually reach that fastest one-second window. If a player is a slow accelerator, but eventually sprints like the wind, he’ll rate well in this metric, but have no chance to beat out grounders given how long it takes to reach that peak one-second window speed.
Last is HP to 1B, which is home plate to first base time in seconds. This is the newest metric available for us to analyze. Unfortunately, there’s no additional description of the metric, like at what point the clock starts, but it seems pretty self-explanatory. In theory, this seems like the exact metric we want as it literally tells us how many seconds it takes for the hitter to reach first base after hitting a ball into play. The faster a hitter could get to first, the better chance he has of beating out that routine ground ball to shortstop.
Oddly, Sprint Speed, and not HP to 1B, had the highest correlation, even though HP to 1B seems like a clearly superior metric for BABIP purposes. Perhaps the smaller player population means that the luck involved in BABIP was more influential, pushing down the correlation. Would a larger sample size of player seasons reveal a truer correlation for HP to 1B and crown it the winner? Sadly, we won’t know the answer to that for years.
Because of what each metric represents, HP to 1B is most certainly the best metric to use for any backwards-looking or predictive BABIP equation. The missing scores for many players during this and past seasons is a major issue though. If the data isn’t all there, it cannot be included in any equation. Sprint Speed would be the logical replacement, but either way, Spd must get the boot from all future xBABIP formulas.
**Update: I asked the Statcast guys on Twitter about the missing HP to 1B data and was informed that since it’s a new process, it hadn’t been built into their daily pipeline yet, but a refresh will be made to fill in the missing records. That’s great news, because now any new xBABIP equation incorporating the metric could be calculated for all players.
Mike Podhorzer is the 2015 Fantasy Sports Writers Association Baseball Writer of the Year. He produces player projections using his own forecasting system and is the author of the eBook Projecting X 2.0: How to Forecast Baseball Player Performance, which teaches you how to project players yourself. His projections helped him win the inaugural 2013 Tout Wars mixed draft league. Follow Mike on Twitter @MikePodhorzer and contact him via email.