Should I Care About Sprint Speed?

Sprint Speed values have been publicly available for a few seasons. While I see it mentioned for this or that, I don’t know how predictive it is or if should I care about it at all. After analyzing the data, Sprint Speed might need to be ignored in favor of Time-to-First. The stopwatch still rules.

The key, in my opinion, is if the ability to run fast can be predictive in any way. No one that I know of is playing in a Sprint Score league, so the speed with have a secondary effect. If a player is running slower, do their stolen bases drop? How about how many infield hits they can leg out? Generally, how will the players change in speed affect their stolen bases and batting average.

One factor to keep in mind is that the aging curve for stolen bases is just a drop with all humans reaching their peak sprinting speed in their early 20’s.  There are going to be a lot of negative speed values coming up but that’s just aging pulling players down.

A second factor to remember is that teams are not allowing hitters to run as much. In 2015, there were over 2500 stolen bases league-wide. Last season, the value was under 2300 for a 9% decline. Again, more negative numbers.

Sprint Speed was first introduced in 2015 at Baseball Savant (links to Time-to-First values) and it is widely cited. Sprint Speed is not the only measured speed metric available. For one fewer season, Baseball Savant has each hitter’s run times to first base which have been the traditional measure of a player’s speed and it’s still used in scouting players. With the two metrics, it’s table time to what conclusions can be drawn.

Table Bonanza

First, I compared the Sprint Speed and Time-to-First values from season-to-season. The in-season value between the two is a nice shiny .823 r-squared. but the season-to-season values aren’t as pretty.

Sprint Speed & Time-to-First Correlation to Next Season
Sprint Speed Year 1 Time-to-First Year 1
Sprint Speed Year 2 0.59 0.50
Time-to-First Year 2 0.42 0.52

A correlation exists but it can’t be assumed players will maintain their ability to run fast.

Next, I moved to stolen bases and their in-season correlations. Here we look at how stolen bases success rate correlates (r-square) to Sprint Speed and Time-to-First base (min 100 PA in each season).

Sprint Speed & Time-to-First Correlation to Stolen Bases
SB/PA SB/(1B+HBP+BB)
Time-to-First 0.37 0.37
Sprint Speed 0.33 0.33

The Time-to-First value is a little more predictive of future stolen bases. It’s a theme that is constant in the results.

Then, I bucketed the changes in Sprint Speeds and Time-to-First to see when hitters may start stealing more or fewer bases. With all the following tables, the stolen base rate significantly drops at the extreme values, but have the smallest number of samples (~20). Also, a few extreme values are influencing the changes.

Change in Sprint Speed & Stolen Bases
Average Median
Change SB/PA SB/(1B+HBP+BB) Per 600 PA Per 200 Times to 1B SB/PA SB/(1B+HBP+BB) Per 600 PA Per 200 Times to 1B
> .8 0.0016 0.0032 0.9 0.6 0.0014 0.0029 0.8 1.7
.7 to .8 0.0008 0.0017 0.5 0.3 0.0000 0.0000 0.0 0.0
.5 to .6 0.0016 0.0031 0.9 0.6 0.0000 0.0000 0.0 0.0
.3 to .4 0.0005 0.0011 0.3 0.2 0.0000 0.0000 0.0 0.0
.1 to .2 -0.0005 -0.0011 -0.3 -0.2 0.0000 0.0000 0.0 0.0
0 -0.0017 -0.0034 -1.0 -0.7 0.0000 0.0000 0.0 0.0
-.1 to -.2 -0.0011 -0.0023 -0.7 -0.5 0.0000 0.0000 0.0 0.0
-.3 to -.4 -0.0021 -0.0043 -1.3 -0.9 0.0000 -0.0001 0.0 -0.1
-.5 to -.6 -0.0026 -0.0052 -1.6 -1.0 -0.0001 -0.0003 -0.1 -0.2
-.7 to -.8 -0.0017 -0.0034 -1.0 -0.7 -0.0014 -0.0028 -0.8 -1.7
< -.8 -0.0024 -0.0048 -1.4 -1.0 -0.0026 -0.0052 -1.6 -3.1
Change in Time-to-First & Stolen Bases
Average Median
Change SB/PA SB/(1B+HBP+BB) Per 600 PA Per 200 Times to 1B SB/PA SB/(1B+HBP+BB) Per 600 PA Per 200 Times to 1B
> .12 -0.0022 -0.0044 -1.3 -0.9 0.0000 0.0000 0.0 0.0
.11 to .12 -0.0044 -0.0089 -2.7 -1.8 -0.0007 -0.0013 -0.4 -0.3
.09 to .10 -0.0007 -0.0014 -0.4 -0.3 0.0000 0.0000 0.0 0.0
.07 to .08 -0.0028 -0.0056 -1.7 -1.1 -0.0008 -0.0017 -0.5 -0.3
.05 to .06 -0.0038 -0.0075 -2.3 -1.5 -0.0023 -0.0044 -1.4 -0.9
.03 to .04 -0.0023 -0.0046 -1.4 -0.9 0.0000 0.0000 0.0 0.0
.01 to .02 -0.0016 -0.0031 -0.9 -0.6 -0.0015 -0.0028 -0.9 -0.6
0 -0.0027 -0.0055 -1.6 -1.1 -0.0039 -0.0078 -2.3 -1.6
-.01 to -.02 -0.0008 -0.0016 -0.5 -0.3 0.0000 0.0000 0.0 0.0
-.03 to -.04 0.0016 0.0032 1.0 0.6 0.0008 0.0016 0.5 0.3
-.05 to -.06 -0.0007 -0.0013 -0.4 -0.3 -0.0009 -0.0018 -0.5 -0.4
-.07 tp -.08 0.0010 0.0020 0.6 0.4 0.0029 0.0060 1.7 1.2
< -.08 0.0013 0.0026 0.8 0.5 0.0000 0.0000 0.0 0.0

The Sprint Speed table is cleaner with small consistent changes. The values from the Time-to-First are a little more dramatic. To clear up the data, here is a simpler table with just hitters going faster or slower (all “no changes” are removed).

Change in Stolen Bases With Speed Changing
Average Median
SB/PA SB/(1B+HBP+BB) Per 600 PA Per 200 Times to 1B SB/PA SB/(1B+HBP+BB) Per 600 PA Per 200 Times to 1B
Time-to-First Slower -0.0024 -0.0048 -1.5 -1.0 -0.0009 -0.0017 -0.5 -0.3
Faster 0.0003 0.0007 0.2 0.1 0.0000 0.0000 0.0 0.0
Sprint Speed Slower -0.0019 -0.0038 -1.1 -0.8 -0.0003 -0.0005 -0.2 -0.1
Faster 0.0003 0.0005 0.2 0.1 0.0000 0.0000 0.0 0.0

Interesting. Hitters are running less based on their slower Times-to-First than their Sprint Speeds. Also, running faster seems to have almost no boost in stolen bases which is influenced by the league-wide stolen base decline.

With the stolen base data out of the way, it’s time to move to BABIP. One item to remember is that the league-wide BABIP is down just a bit over the time frame mainly because home runs aren’t a BABIP component. And home runs are way up.

Change in BABIP & Speed Score
Change in Sprint Speed Average BABIP Change Median BABIP Change
> .8 0.003 0.002
.7 to .8 -0.004 -0.012
.5 to .6 -0.008 -0.009
.3 to .4 -0.003 -0.002
.1 to .2 -0.005 -0.005
0 -0.003 -0.005
-.1 to -.2 -0.007 -0.006
-.3 to -.4 -0.005 -0.008
-.5 to -.6 -0.009 -0.009
-.7 to -.8 -0.010 -0.006
-.9 to -1 -0.002 0.006
< -1 -0.028 -0.023
Change in BABIP & Time-to-First
Change in Time-to-First Average BABIP Change Median BABIP Change
> .14 -0.020 -0.017
.13 to .14 -0.029 -0.025
.11 to .12 -0.005 -0.008
.09 to .10 -0.025 -0.024
.07 to .08 -0.020 -0.017
.05 to .06 -0.010 -0.010
.03 to .04 -0.002 -0.003
.01 to .02 0.001 -0.003
0 -0.005 -0.003
-.01 to -.02 -0.006 -0.003
-.03 to -.04 0.002 -0.002
-.05 to -.06 0.000 -0.003
-.07 tp -.08 -0.005 -0.001
< -.08 0.000 -0.002

With Sprint Speed, the change in BABIP is gradual but Time-to-First drops significantly with hitters who take 0.07 sec longer to get to first base. The slugs see a drop in their BABIP and therefore their batting average declines.

At this point, I’m about done with Sprint Speed and for reference, here is the BABIP change for hitters going faster or slower to first base.

Change in BABIP With Change in Time-to-First
Average BABIP Change Median BABIP Change
Faster to 1B -0.002 -0.002
Slower to 1B -0.019 -0.019

Those who slow down, really take a pounding.

Pulling it all together

After going through the data, I’m not going to use Sprint Score again. Time-to-First seems to more indicative of a hitter’s future performance. And for several reasons this makes sense.

First, the Time-to-First is the time it takes a hitter to get to first (rocket science) and the faster that time is the more hits the player will accumulate. Sprint Speed is just looking at the top end speed which may take players longer to reach.

Second, teams seem to be acting on changes to Time-to-First when deciding if hitters should run more. The value has been around for decades and coaches can use it to help determine who will be running based on that day’s pitcher-catcher combos. One front-office executive I talked to stated that their coach still uses the runner’s Time-to-First, along with the pitcher time to plate & catcher pop time to determine who should steal bases. Coaches seem to not care about Sprint Speed when making decisions on the field.

Another interesting bit is that it takes almost a change of 0.1 seconds for the manager to stop the steals. A 0.1 change in Time-to-First means a drop in the scouting scale by one full grade.

Whenever a hitter changes a grade, bad things begin to happen. For example, the players who saw a 0.1 increase in their Time-to-First from 2018 to 2019 lost 1.7 stolen bases and 23 points from their BABIP.

So, as the season starts up, I’m going to closely be watching the Time-to-First values to find hitters who are likely to struggle. And if anyone feels ambitious, they can get a jump on everyone and breakout the stopwatch to get a jump on their competition.

 
While some beat your favorite author leagues will be starting soon, feel free to go to the NFBC and sign up for their Draft or Online Championships. Also, for those with deeper pocketbooks, the NFBC Main Events are quickly filling up.





Jeff, one of the authors of the fantasy baseball guide,The Process, writes for RotoGraphs, The Hardball Times, Rotowire, Baseball America, and BaseballHQ. He has been nominated for two SABR Analytics Research Award for Contemporary Analysis and won it in 2013 in tandem with Bill Petti. He has won four FSWA Awards including on for his Mining the News series. He's won Tout Wars three times, LABR twice, and got his first NFBC Main Event win in 2021. Follow him on Twitter @jeffwzimmerman.

12 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Rollie's Mustachemember
4 years ago

Where can one find Time-to-First stats? Is it publicly available?

pilldommember
4 years ago

Baseball Savant has 0-90 times which is more or less time to first.

zurzlesmember
4 years ago

Statcast Leaderboards -> Sprint Speed has HP to 1B numbers