One of the ongoing debates between myself and my friend Jim is “Why do you keep calling for Hit and Run and Steals in MVP?” Seriously, he had Ichiro on Tampa (don’t ask) and got 30 SB but 40 CS. My point (besides pointing out that CS number) is to drop the “75% success rate … Continue reading “the value of stealing”
One of the ongoing debates between myself and my friend Jim is “Why do you keep calling for Hit and Run and Steals in MVP?” Seriously, he had Ichiro on Tampa (don’t ask) and got 30 SB but 40 CS. My point (besides pointing out that CS number) is to drop the “75% success rate = break even” sabermetric line. Which works until he questions “Why 75%?”
It’s 75% because, well, because other people who’ve looked more into this than me say it is? I’m smart enough not say that out loud, but it’s what I’m thinking. It is a good question, I’ll give him that.
This came up in my mind this morning because I saw an article on the ChicagoTribune article hitting this topic in reference to wacky Ozzie and his wacky tactics. It’s the usual “introduce a sabermetric idea only because we’re having other people shoot it down” stuff, but it got me thinking about this again and wanting to come up with a better case for my reasoning.
What helps is BaseballProspectus ran an article a few weeks ago about run expectancy (in the context of “is there a good time ever to sacrifice and what are they?”), helpfully giving a chart of how much runs people scored in various people on base/outs combinations last year. The important numbers to me are
|
None |
1st |
2nd |
3rd |
None |
.531 |
.919 |
1.177 |
1.380 |
1 out |
.282 |
.535 |
.706 |
1.032 |
2 out |
.109 |
.237 |
.341 |
.384 |
3 out |
0 |
0 |
0 |
0 |
From there, all it takes it is a simple probability formula…
Breakeven point (the runs expected if you had done nothing)
= the probability of succeeding * the runs expected if the steal is successful
+ the probability of failure [in other words: 1 – the probability of succeeding] * the runs expected if the steal fails.
and some high school algebra to solve by the probability of succeeding
probability = (break even run expected – failure run expected) / (success run expected – failure run expected)
Plugging in the numbers:
|
1st to 2nd |
2nd to 3rd |
None |
.712 |
.815 |
1 out |
.714 |
.647 |
2 out |
.695 |
.888 |
In the situation Jim usually steals in, the percentage is actually lower than 75%. Not by much, but enough to make a difference on strategies. I’m quite stunned stealing 3rd base with one out is the best option of all by far, but I’d guess all the man on third/one run strategies inflates the worth of getting a man there with one out. If that number isn’t a particular year variance, teams really ought to be thinking about stealing third more often.
(This doesn’t really affect MVP, since it exists in a different universe. And also, I really do love this post following the last one.)