We all know the obvious results. He is safe and the RE or WE (we’ll only talk about run expectancy from now on) changes to reflect a runner on second rather than first (although now we are in middle of a PA, so it is not quite so simple). Or he is out and the RE reflects a base runner removed, although again, we are in the middle of the PA with some kind of a ball and strike count.
But, what happens when the base runner is attempting a steal and the batter puts the ball in play? That is the hidden value (presumably) of stolen base attempts, not withstanding the effect it might have on the pitcher and the defense, as per conventional wisdom. I will look into that in Part II.
Where does that extra value come from? In the extra bases that the runner takes on a single or double, staying out of the GDP, removing the force play even when a GDP was not in order (but the force still was), and occasionally forcing a FC no out (when they tried to get the force on the runner at second and he was safe) or a hit, when the only play would have been a force at second had the runner not been in motion. The one downside is the occasional DP on a line drive or short fly ball.
Unfortunately, there is no database that I have access to that tells me whether a runner was in motion or not. I don’t know why. This is a basic piece of information that is necessary for all kinds of important research. I am pretty sure that most of the database companies track this information (it is certainly easy to do so – just one more click of the mouse on occasion), but for some reason they don’t include it in the information that is available to me, and that includes retrosheet.
So, I had to figure out a way to infer when a runner might be running and the batter puts the ball in play. Here is how I did that:
First, I looked only at base runners on first that had a high stolen base attempt per opportunity for that year (.20 per opp or higher). Then I split the pitchers into 2 categories – those who allow very few stolen base attempts per 9 innings (<.35) and those that allowed quite a bit (> 1.80). The assumption is that even these high frequency base stealers would attempt a stolen base much less often against the first group of pitchers than against the second group. I also only included RH pitchers, otherwise the “low attempt pitchers” would contain too many lefties and the batted ball results would be biased.
I only looked at situations where there was a runner on first base, and no other base runners.
What we want to look at is the rate that all that good stuff I mentioned above happens when the runner at first is either likely to be running on the pitch or not.
So let’s look at some data.
Remember that all of the data are with high frequency base stealers (HFBS) on first base, and no runners on second or third. A high frequency base stealer is any player who had at least 50 base stealing opportunities (runner on first, no one on second) and a 20% attempt rate, in that season.
Then I looked at what happened with those runners on base with two groups of RH pitchers on the mound – one, those that allowed very few SB attempts (less than .35 per 9 innings) and those that allowed a lot (> 1.8 per 9). The presumption is that against the first group even these prolific base stealers attempted a SB infrequently, and against the second group they ran a lot. This assumption turned out to be true.
Table I: SB attempt rates (among prolific base stealers), when on first base and no one else on base.
Against low frequency pitchers (Group I) | Against high frequency pitchers (Group II) |
7.6% | 41.3% |
…
That is a lot of base stealing against the second group of pitchers! These are attempts that resulted in a SB or CS only. They don’t include pickoffs, balks, etc.
These numbers do not include when the runner was in motion and the batter put the ball into play. That number has to be inferred. You will see in a second how I did that.
Now let’s look at the GDP and Force out (and FC) rates when the batter hit a ground ball with less than 2 outs. Presumably, some percentage of the time these high frequency base stealers were running on the pitch and the ball was put into play (therefore no SB or CS was recorded). We can assume that they did so much more frequently against the Group II pitchers than against Group I pitchers since the SB attempt rate was so much higher against Group II than Group I. In situations where the runner attempt to steal a lot, he will also necessarily be in motion quite often when the ball is put into play. If in a certain situation there are very few steal attempts, then the runner will likely not be running much on a ball in play.
Table II: GDP and Force out (and FC) percentages against Group I and II pitchers
Against low frequency pitchers (Group I) | Against high frequency pitchers (Group II) | |
Batter hits a ground ball out or FC w/ runner on first, less than 2 outs | 12.3% | 12.8% |
GDP | 39.3% | 30.6% |
Force out at second, batter safe | 20.9% | 21.7% |
Fielder’s Choice, no out | .5% | .6% |
Runner safe at second, batter out | 38.3% | 47.2% |
…
From these numbers, we can infer, at least approximately, how often the runner at first is running on the pitch (attempting to steal or perhaps a hit and run).
Against Group I pitchers, the runner is out at second (via a force or a GDP) 60.2% of the time. With Group II pitchers on the mound, they are out at second only 52.3% of the time. That suggests that at least 7.9% of the time that a ground ball out is made, the runner at first is on the move. I say, “At least,” because some small percentage of time the runner is on the move against Group I pitchers as well (remember they still allow a 7.6% SB attempt rate) and occasionally when a runner on first is off with the pitch, he is still forced at second (e.g., on a hard hit ball right near the bag).
So let’s do some fairly simple algebra. If the SB attempt rate against Group I pitchers is 8%, and 41% against Group II pitches, we have this equation:
.41 * P – .08 * P = .08
where P is the ratio of runners on the move when the ball is in play to runner on the move when the ball is not in play (an actual SB attempt). Solving for P, we get .24. That means that for every SB attempt by a base runner, there is an addition .24 times when the batter puts the ball in play and the runner is on the move. This creates an advantage for the batting team by staying out the DP and rarely being forced at second, and advancing the extra base more often on a single or double.
By the way, the percentage of fly ball or line drive double plays for both groups was almost the same: 1.5% when running a lot (against Group II pitchers) and .9% when not running a lot. While that suggests that the runner in motion gets doubled up slightly more often than the runner not in motion, it is a rare occurrence in either case. So the disadvantage to the runner in motion by virtue of the occasional extra air ball DP can probably be ignored (it is actually worth – .0065 runs per SB attempt, or around -.26 runs per year for a 40 SB attempt player).
Let’s compute the average RE on a ground ball out against both groups of pitchers, to see the actual advantage in runs on an attempted steal of second base with less than 2 outs. We will assume that with 2 outs, there is not much of an advantage, although that is not exactly true.
Table III: RE after a GB out
Against low frequency pitchers | Against high frequency pitchers | |
Average RE after a GB out | .317 (0 and 1 out combined) | .364 (0 and 1 out combined) |
…
As you can see, there is a .048 run gain when a batter hits into a ground ball out against a Group II pitcher. This is because the runner is more often running on the pitch and ending up safe at second. From Table II above, a ground ball out happens around 13% of the time (with a runner on first only and less than 2 outs) against Group II pitchers. So the overall gain is .048 * .13 or .0062 runs with a 41% steal rate. So, per steal, that is a .0152 run gain (.0062 / .41). So being able to avoid a force out at second ball on a ground out by the batter is worth an extra .0152 runs for every stolen base attempted by a runner. If a prolific base stealer attempts 40 SB per year, that is an extra gain of .61 runs, nothing to write home about.
Finally let’s look at advancing on hits. Presumably, the runner in motion will be able to take the extra base more often that one who is not on the move. Here is the data:
Table IV: Advancing the extra base on a hit
Against low frequency pitchers | Against high frequency pitchers | |
Extra base on a single | 41.2% | 42.3% |
Runner out trying to advance | .8% | 1.4% |
Batter moves up a base | 6.4% | 9.9% |
Extra base on a double | 52.0% | 51.5% |
Runner out trying to advance | 3.9% | 2.9% |
Batter moves up a base | 2.8% | 2.6% |
…
Although I only looked at outfield singles, and I assume that virtually all doubles are to the OF, there is obviously no guarantee that a runner on the move will be able to take the extra base. If I had to guess, I would say that rather than a 40-50% advance rate when the runner is likely not on the move, there will be a 75-80% rate when the runner is going. Let’s see if this estimate fits the data.
From the GDP data, we estimated that the runners were moving when the ball was put into play around 8% more against Group II pitchers than versus Group I pitchers. If we expect a 35% increase in the advance rate when the runner is attempting a steal, then the overall advance rate for the Group II pitchers should be around 2.8% higher than with the Group I pitchers. For singles, we see only a 1.1% increase and for doubles it is actually a .5% decrease (unfortunately these are very small samples – 371 and 127 opportunities, singles and doubles against Group I pitchers, respectively). If we combine them (weighted according to number of opportunities, of course), we get a .67% increase with Group II pitchers, less than expected (2.8%). Of course the 35% increase (75-80% rate of advance) in advance percentage with the runner off on the pitch was a wild guess on my part. It could be only 25% more or it could be 45% more. In addition our sample sizes of singles and doubles are so small, that the differences between the two groups is not very meaningful. So I think that our 8% or .24 runners on the move per stolen base attempt, garnered from the GB data is still a reasonable estimate.
As I did with the GDP numbers, let’s see how taking the extra base adds to the value of a SB attempt, if at all.
Table V: RE after a single or double
Against low frequency pitchers | Against high frequency pitchers | |
Average RE after a S or D | 1.186 (1 out) | 1.190 (1 out) |
…
Using the numbers in Table IV above, there is a only tiny difference in the resultant RE after a single or double with the runner on first, .004 runs. These situations occur around 20% of the time, which results in an overall advantage of .002 runs per SB attempt. Suffice it to say that advancing the extra base more often when on the move is probably not worth very much overall. Even if we assumed a 2.8% extra advancement rate, rather than the modest .67% in our small sample, that would be worth .008 runs per SB and not .002.
So, there does not seem to be much of an extra advantage from the stolen base attempt, beyond the traditional numbers gleaned from the success percentage (and catcher errors on a SB, pickoffs, pickoff errors, and balks). Staying out of the GDP and taking the extra base on a single or double are likely very small advantages, and hitting into the occasional air ball DP appears to be a tiny disadvantage. if we add up all the numbers, we get .0152 for the GB situations, .008 runs (we’ll assume the more optimistic number) for the single and double advances, and – .0065 runs for the air ball DP. That is a total gain of .017 runs per SB attempt or around an extra .7 runs per season for a high frequency base stealer.
About the only thing we have really accomplished is to estimate that a base stealer is on the move when a ball is put into play about .24 times for every time he records either a SB or a CS. In other words, around 19% of the time (.24 / 1.24) that a runner takes off, the batter puts the ball in play. Unfortunately, that extra few steps isn’t worth much to the offensive team, at least terms of what I looked at so far, staying out of the GDP and force outs, and advancing extra bases on hits.
In Part II, I’ll look at how the hitter’s results are affected by the runner attempting a steal of second (or not). Can we expect more hits when the runner is in motion because the infield is moving? Is the batter distracted by the runner going? Is the pitcher distracted with a prolific base stealer on first? Does the batter take some pitches that he might ordinarily swing at? Are some of these “runner going” situations hit and runs such that the batter is forced to swing at anything? Stay tuned for Part II…
It seems to me that all of the attempts through baseball history to create more offense, such as the hit and run, stealing, bunting all of that nonsense is just a waste of time, when in reality the best strategy at any level is as simple as dont get out. All of these created strategies just increase your teams odds of creating outs which reduce the odds of creating runs. Of course this is a blanket statement and doesnt always apply but on the whole, managers stop getting in the offenses way!
In general I agree with you. Most of these strategies on the whole are pretty darn close to break even. That being said, obviously there is an advantage to stealing 23 bases and getting thrown out only 4 times. Or for a pitcher who can’t hit a lick to execute a sac bunt.
The problem is this: Because managers and players do not understand the mathematics behind these strategies, they will end up using them in such a sub-optimal manner that they take otherwise profitable strategies if used correctly, and turn them into break even strategies at best. It is like the poker player. The expert poker player will use his bluffs, calls, raises, folds, etc.,in a near optimal manner against his opponents, while the average or poor player will use the same strategies, but in a sub-optimal fashion, allowing the expert player to exploit them.