Rajai Davis & 80 Steals: Part Two

Part one dissected the claim and found that Davis attempting enough steals to successfully swipe 80 was improbable due to a number of factors. In part two, let’s focus more on the analytical side. I’m going to shamelessly steal an idea from Jeff Sullivan, but first some disclaimers and notes.

This process is going to blanket all steals attempts as being in equal in leverage. This is obviously untrue and affects how baserunners would approach certain scenarios. Certainly Davis is going to be more focused in and less willing to risk a caught stealing in the ninth inning of a tied game instead of the third inning when the A’s are up by four. Context is important, but it’s just not something I can accurately adjust for here.

Let’s assume the league run environment is around 4.5. As Sullivan notes, that makes the run value of a stolen base roughly 0.19 runs and a caught stealing is worth approximately -0.45 runs. One failure is worth two-and-a-third successes in such an environment.

The above chart details the minimum amount of attempts needed at various success rates to reach 80 steals. The best case scenario is that Davis turns into one of the most efficient basestealers in league history and needs fewer than 100 steals to rack up 80. The worst? Davis is caught left and right, but still allowed more than the 120 attempts necessitated by his performance to reach 80. Let’s turn these numbers into run values.

The same overlay is in effect here. The colored cells represent the point in which Davis would have at least 80 steals. The green means positive, the red means negative; or, in other words, not worthwhile to give Davis the green light to such an excessive degree. To surmise both parts: Davis is highly unlikely to get 80, but barring a total decay of his stealing skills, he’d probably make it worth the A’s time by adding a few runs.

The real intriguing part of the chart is look at the contrast between high success rates with moderate attempts and moderate success rates and higher attempts. Stealing bases isn’t always about volume. Perhaps the real goal Henderson should set for Davis isn’t 80 steals, but an 80% success rate.




Print This Post





12 Responses to “Rajai Davis & 80 Steals: Part Two”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. TRE says:

    What is the prognosis for Rajai Davis once Michael Taylor takes over in LF? Does he boot out Coco Crisp?

    Vote -1 Vote +1

  2. Spunky says:

    I love your point at the end that a player’s SB% is often more important than the number of bases he actually steals. (On a side note, this is one of the reasons Tim Raines should be an HOFer and was a such a force on the basepaths–an 85% SB rate. That’s insane for 800+ SB!!)

    Anyway, if he’s borderline in SB% (like say, right around 70%), it might still be worth it for him to try to steal assuming he gets better at it over the course of the season. I don’t know if it usually goes that way though.

    Vote -1 Vote +1

  3. Obligatory Sox Fan says:

    This was a great post, and I almost always dig the sabermetric view on runs and wins, but man, this really takes some of the fun out of stolen bases. Not the post, but the reality of the run values of stolen bases.

    110 attempts at 85% just to get a win worth of runs?! Base stealing is such a fun part of the game, I just don’t want to accept these numbers.

    Vote -1 Vote +1

    • PhD Brian says:

      Shows how over paid are the base stealers.

      Vote -1 Vote +1

      • Joe R says:

        Actually, in some cases, stolen bases are underrated now. We’re so used to the home run and high run environments that we forget that a stolen base can help a lot.

        Look at the uphill battle Raines still faces for the Hall of Fame, for example. If we use straight linear weights found here, Raines added 114.8 runs with steals, and 92.1 runs above average. That is a ton of value added.

        Vote -1 Vote +1

  4. SF 55 for life says:

    nice work R.J. Question for you guys, when do sample sizes for stolen bases become reliable? 50 attempts? 100?

    Vote -1 Vote +1

    • Joe R says:

      Well since SB/CS are totally binary results, it’s pretty easy to put a distribution around it.

      Say a fast guy, over 2 seasons, attempts 200 SB, and gets 160 of them. We can feel 95% confident he’ll steal between 74.46% and 85.54% the next season.

      Vote -1 Vote +1

  5. The Usual SusBeck says:

    Don’t have time to think about this comment, but wanted to get it out.

    Should SBs be looked at somewhat in terms of LI or WPA rather than runs created? More along the lines of how relievers are looked at. I feel like (no time for logic) that a runner who can put himself in scoring position late in a game would be more valuable.

    Maybe that would only make sense for pinch runners. I don’t know.

    Vote -1 Vote +1

  6. Daniel says:

    While I agree that SB% is very important I feel like simple stats overlook many factors. For one, these stats do not measure the benefit of errors on SB attempts that result in extra bases. They also do not measure the effect a base-stealer has on a pitcher, or the effect of middle infielders cheating towards, or forced to cover second base.

    Vote -1 Vote +1

    • JRoth says:

      I was wondering this – it’s not clear to me whether Tango’s original numbers take these peripheral matters into account*. One thing I can’t figure at all is how Tango gets “Pickoff Error” as a negative run value – how exactly is it a bad thing for a runner to advance without an out being recorded? And, actually, how is “Other Advance” a worse outcome than an out?

      I realize that this has been hashed out before, elsewhere.

      * And frankly, I don’t see how it could without being a specialized study – you’d have to look at the likelihood of a run scoring when a good basestealer gets on but doesn’t attempt a steal vs. the likelihood of a run scoring when an equally swift non-basestealer gets on, correcting of course for their relative offenses, the pitcher’s move, the catcher’s arm, etc. I don’t doubt it could be done, but I frankly doubt that linear weights looks at it at all, and further, I’m not sure you could integrate it directly into LW – a guy who steals 5 bases a year isn’t getting/creating the general advantage that a 50-SB player is, so it’s not tied (in a linear fashion) to SBs. You’d have to add it as a factor – 0-10 steals gets 0 P-factor (P for pesky), 10-20 SB get 0.05 runs added, etc. The effects are mostly small, but we’ve all witnessed the disruption caused by a Jose Reyes on the basepaths – it’s not zero, and it’s not mere anecdote. It’s just really hard to measure (mostly, I think, because its hard to extract the speed element – there are plenty of fast guys who don’t steal all that much, but how can you determine who qualifies as fast? IF hits would be a pretty crude measure, and doubles will inevitably be conflated with power, and triples are simply too rare for any given player).

      Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>