Stolen bases: a topic I haven’t talked much about this year. This is partially because of their unpredictability. Stolen bases, unlike any other stat in fantasy baseball, can be controlled by the simple will of the player and/or his manager.
We’ll talk more about him later in the article, but last year, Shane Victorino stole just four bases and attempted only seven (in 462 plate appearances). This year, he’s stolen 37 and attempted 40 (in 484 plate appearances). That is a huge difference and—despite the fact that he stole 45 bases twice in the minors—one that would have been difficult to predict from stats alone.
Stolen base stats
While stolen bases are largely a function of willingness (which is pretty unpredictable), let’s try to find some stats that can help quantify base stealing ability.
To quantify willingness, let’s use Stolen Base Attempt Percentage (SBA%). This measures how often a player will attempt a steal once he is on base. To calculate it, we’ll use (Stolen Base Attempts)/(1B+BB+HBP).
Willingess doesn’t mean anything, though, if a player can’t get himself in position to steal in the first place. To quantify opportunity, we’ll use Stolen Base Opportunity Average (SBO%). To calculate this, we’ll use (1B+BB+HBP)/(TPA). This will resemble batting average or on-base percentage because it’s basically on-base percentage without including extra-base hits. It’s a measure of how often a player reaches first base. As a side note, I am in no way endorsing a strategy that maximizes steals, but because they are a part of fantasy baseball, we need to see which players do this (either intentionally or by default).
You can get yourself to first in every at-bat and try to steal every single time, but if you have the speed of Jason Phillips, it isn’t going to do you any good. To quantify speed, we’ll use the Speed Score formula originally proposed by Bill James in 1987. His formula took six elements and put them on a scale of 1-10, then took the average of the six elements. These elements included stolen base percentage, stolen base attempts, triples, runs scored, grounded-into double plays and range factor. We’re going to exclude the Range factor element. If you’re interested in the exact formulas, here they are:
Stolen base percentage: ((SB+3)/(SB+CS+7)-0.4)*20 Stolen base attempts: SQRT((SB+CS)/((H-2B-3B-HR)+BB+HBP))/0.07 Triples: 3B/(AB-HR-K)/0.0016 Runs scored: ((R-HR)/(H+BB-HR+HBP)-0.1)/0.04 Grounded-into double plays: (0.063-GIDP/(AB-HR-K))/0.007
For those of you curious, Jason Phillips’s Speed Score this year was 1.74.
Of course, we can’t forget a simple measure of Stolen Base Efficiency (SB%). For this, we’ll use stolen base success rate, calculated by (stolen bases)/(stolen base attempts).
Ideas for using these stats
I was playing around with these stats a little and found something interesting. So far this year, 17 players who have attempted at least 10 stolen bases have a stolen base success rate of at least 88%. Here’s the list, with all of the stats we talked about before included.
When I saw this, I was curious just how repeatable this skill is… whether a player who is that successful is doing it as a result of skill or a little luck. So, I looked at how these players did last year. The table is below.
Only two of these guys attempted fewer than 10 stolen bases last year (Victorino!), and just four had success rates above 88%. This is by no means a complex, scientific study, but I think it does provide a good starting point for some research. What is the year-to-year correlation? How much of a player’s stolen base success rate is luck-related? Can it be predicted with a reasonable degree of accuracy beforehand?
One thing I will look into is how Speed Score plays into success rate. Perhaps break it down into sections (7.00 to 7.50, 7.50 to 8.00, etc.) and find the success rate for each group. Let’s say, for example, that the success rate for players in the 4.00 to 4.50 range is 65% (I made that number up, but bear with me). That would mean Shawn Green and Edgar Renteria are in for a decrease in their success rate and, subsequently, their stolen base totals.
This could also indirectly affect their attempts. Again, it’s something we need to look into, but it would make sense that a player who is no longer succeeding at such a high rate would stop trying so much, which would further decrease his total number of steals.
We could also run a study to tell us how much of the difference is skill related. David Wright‘s Speed Score is 5.24, but his success rate is 88% this year. It was a quite solid 80% last year as well. He has long been said to have great instincts on the base paths to make up for his only slightly above-average speed. It would be interesting to find out just how good those instincts are when we remove speed and luck from the equation. If the regression for guys in the 5.00-5.50 is to, say, 70%, we wouldn’t want to regress Wright that far if he has some innate base stealing skill that doesn’t involve speed.
Where did he come from?
Hard as we may try to predict things, there will always be cases that come out of nowhere. This year, the biggest example is Victorino. He attempted just seven steals last year and was successful just 57% of the time. Check out his numbers below.
Victorino was considered a stolen base sleeper preseason because he worked with first base coach Davey Lopes in spring training on his base stealing, but looking at last year’s numbers it would have been very difficult to see this coming. How could we have done it without using this sort of qualitative data? The answer is that we really can’t, and that qualitative data like this can’t always be relied upon.
We can put an asterisk next to the player on our draft sheets, but it would be nigh impossible to try to quantify how working with Lopes would improve Victorino’s numbers this year. I mean, his Speed Score wasn’t great, his attempt percentage was very low, and his success rate was very low. I’m not sure anyone could be confident in saying that a guy like this will steal 40 bases.
These types of things will happen with steals, so we need to come to terms with this fact and strive to find guys who are more consistent from year to year.
We’ll talk about this more in the future. I wanted to introduce some of these stats throw out some ideas that might help us better use them. As always, if you have ideas or suggestions, feel free to either comment or email me.