Defensive Indifference. Why?
March 8th, 2010 by czar

I’ll preface this by saying that I’m relatively well-versed in the physics of baseball (minus the occasional alannathan wisdom nugget that changes the way I think about things), but am certainly not at the forefront when it comes to advanced statistical analysis. As part of my “spring training” I finally got around to reading Baseball Between the Numbers, in particular a chapter about Rickey Henderson and Pete Incavalia; how SB are overrated when taken one-dimensionally and solely quantifiably. But it still didn’t explain one of my most significant pet peeves of all time– defensive indifference.

Defensive indifference is defined as “play in the later stages of a game in which the defensive team, either ahead or behind by a large amount allows a player to advance a base without any attempt to force the runner out.” To simply matters I can’t see any reason why you can’t just look at it as an uncontested stolen base.

Now obviously, defensive indifference is going to carry varying weights in terms of significance depending on whether a team is up or down and by how much. For the sake of this post, I chose to imply a “worst-case scenario.” A team is down 2 runs in the bottom of the 9th. The closer strikes out the first batter but allows a broken bat single to the second. No one holds him on at first and he breaks for second. The catcher doesn’t throw and the MIs don’t cover— our runner takes the base by “defensive indifference.” No one seems to care about this as “his run doesn’t matter; he can score and the home team will still win the game.

Now for what baffles me. As the OBP movement of the past 15 years has shown, the ability to score runs is directly tied into the ability to avoid making outs. For a defense the opposite is true— the key is to record outs as frequently as possible, as outs are effectively the “clock” in baseball. So by just freely giving away 2nd (or 3rd) base you may not be directly giving the team more runs (just moving the potential run around), but you ARE giving away a chance to record an out and therefore bring you one crucial step closer to the end of the game. Let’s assume the CS% remains constant across all batters (not a great assumption, for hypothetical sake we just need one number). In 2009, there were 2970 SB compared to 1133 CS across MLB. This works out to a CS% of 28% (or a SB% of 72%). If we assume this number is constant with inning/game situation (and outside of a slight negative slope, it’s pretty close) it means that for every defensive indifference allowed you had a 28% chance to record an out had you played the defense like you had any other inning, except you are welling to concede a 0% result.

To offset this 0% result and make defensive indifference statistically reasonable, we need to find a mechanism to get 28% of a net gain (defined as some weird “out-making units” I suppose) out of ignoring the potential for the runner to move up. This might seem quite simple for some people: treating the runner like a base-stealer distracts the pitcher, alters the defensive positioning, puts pressure on said defense, causes errors, and may open up holes during situations where the potential SB is occurring (i.e., Pedroia covers 2nd, LHH pulls one through the vacated right side of the infield). From hereon, I will refer to this as the “indirect effect.” If this indirect effect (benefit to the hitter) is greater than 28% “out-making units” then obviously we’d want to pick the lesser of two evils; so ignoring the runner is the prudent thing to do.

Now, I don’t have any readily accessible data of how this would shake out in the ninth inning. However, in BBTN, James Click did a study attempting to examine the effect of having a base-stealer on base on the following hitter. I have copied the analysis here.

To test this theory, we can start by breaking runners on first in steal situations into five groups based on the frequency of their stolen-base attempts. Then we compare the performance of the batters behind them to their expected performance given their overall stat lines. (It’s crucial to adjust for the expected performance of the subsequent batters, as base-stealers tend to be in the highest spots in the lineup so that a disproportionate number of subsequent batters will be #3, #4, and #5 hitters.)

In 2004, batters who came to the plate with the top 20 percent of base-stealers on base- as determined by their stolen-base attempts rate- saw their OPS increase by 34 points over expected performance. On the other hand, batters coming to the plate with the slowest runners on first saw an increase of just 4 points. Over the past five years, batters at the plate with baserunners most likely to steal on first improved their OPS by an average of 24 points, the next group by 27, then 17, 20 and 13 for the slowest. So it’s true that runners on first who are more likely to steal improved the performance of the batter at the plate, but the difference between the most and least aggressive base stealers is marginal: about 11 points of OPS, with data points bouncing around among the five groups of base-stealers.

The overall point of this was to show that base-running is overrated in general; but it also has the cool side effect of implying that this “indirect effect” is more of a myth; or at the very least, it’s hidden within the statistical noise. There is no serious negative effect from treating the bases like they have a runner on versus them being empty.

So, the question I pose: if there is no discernible or significant negative to treating the bases like they actually have a runner on versus utterly ignoring his existence (indirect effect) than why do teams willingly pass up the chance (regardless of significance) to record an out and bring the game that much closer to its conclusion? To me this seems like a very simple concept; so the fact that it hasn’t been hammered in sabermetric circles leads me to believe that I’m missing something—is the “indirect effect” on the same order of magnitude as the WinExp increase from throwing a runner out? Is there some sort of compounding (non-linear?) effect that I’m ignoring? Has this been covered ad nauseum before and I just haven’t read about it?

These aren’t hypotheticals, but serious questions. What isn’t my sabermetrically-pea-sized brain not comprehending?

Some theories?

Most basestealers have a pretty high success rate, so you probably aren’t going to catch them.  IIRC, the break even points is ~74%?

Last year’s success rate was 72%. The “breakeven” point (as defined by the success percentage where a baserunner increases his teams odds of winning with a SB vs. decreasing with a CS) are a function of inning. But “probably aren’t going to catch them” still gives you a better chance of recording an out than “definitely won’t catch them.”

A pitchout puts the pitcher in a hole (i.e. a ball).  It’s also possible that making a pitch that can be used to catch a runner is a disadvantage in some way (maybe it has to be a fastball).

This is also possible; although if it was this detrimental to the pitcher (and advantageous to the batter) shouldn’t Click’s study pick up on this?

There is some probability of an error, which could lead to a runner on third (and sac fly terratory) or even a run scoring if it’s bad enough.

The error would be sort of double-counted in the SB/CS rates; but even if it isn’t, we can still argue that the run itself doesn’t matter– what we care about is the opportunity to record an out that the other team gives us that would otherwise not be there if they weren’t trying to move up on the basepaths.

It’s possible the distraction effect on the pitcher, effect on batting count and possibility of an error are close to the 28% chance of getting an out?

These are quantifiable enough that if they occurred with any real significance, we’d at least see it in data (i.e., batters who hit with “fast runners” on base have better peripherals than when the bases are empty (or even when “slow runners” are on).

When there is a runner on and the situation calls for holding them close, the pitcher works out of the stretch. If the runner is going to be ignored, then he pitches from a windup position. Some pitchers work better from the windup and this lets him work from the same position. Possibility.

Good point, I did not consider this (weird, because I hated working out of the stretch in HS/College). However, last year in a Buchholz discussion, I remember looking at league splits with runners on vs. bases empty last year and there was no statistical significance between opponent BA/OBP/SLG/OPS.

Their selected strategy will determine whether or not the runner even considers breaking for second – in this situation, he has to be absolutely 100% sure that he won’t be thrown out. So I think the 28% CS number is a bit of a red herring – the runner isn’t going anywhere unless he can take a big lead and the pitcher is going from the wind-up – in which case there’s virtually no chance of him being caught even if the catcher does throw to second.

Huh. Well that’s a pretty big omission. Because that run doesn’t matter, there is really no reason to take the extra base; ergo if a team tightens up and plays SB-prevent defense, the baserunner just will stay put; therefore it’s “lose-lose” for the team (i.e., they don’t gain anything from preventing the runner from stealing, nor do they gain anything (at least significantly) from ignoring him and “focusing” on the batter). I’m a little pissed I simply missed that.

I agree that the GIDP possibility still gives us the same general question, — back of the envelope, I’ll assume 30% of PA end in a “ground ball,” with an out percentage (1 – BABIP) of (1 – .240) (both quick 2009 AL numbers) and we’ll make a very general assumption that 50% of these can be turned into double plays (completely made up) which leaves us about a 10-12% chance of a DP (seems in the ballpark); still relatively high in any given AB without giving away free bases. (I’m aware there must be an easier way to get at this number (odds of hitting into a DP with a runner on 1 or 1/2 and less than 2 outs) but I wasn’t able to split situational hitting by number of outs in the inning on B-R, so the GIDP number would have been skewed down by AB with 2 men already out).

(I’ll admit this is, in a way, a mini-crusade against DI. Not that I hate a Foulkey-bunting level of hatred for it, but I’ve never been able to wrap my head around the notion that a team can just ignore a baserunner that they would not have ignored for the previous 8 innings because their closer needs to “focus” on the batter. Whether that’s rational or not, I don’t know.)

Posted in Message Board Ramblings

(comments are closed).

About The Site

  • Traversing the land that is known as Red Sox Nation, The Czar Who Wears Red Sox is an attempt at compiling a repertoire of my ever-so-sexy forum posts (when I'm too lazy to write my own damn entry) and other random baseball thoughts that strike. For those whose posts serve as the inspiration of my epiphanies and rants, do not be angry, but merely, be honored that you have achieved such status. Names will never be revealed. Feedback appreciated, as this is a work in progress.

    E-mail: webmaster [at] balloflightning [dot] com


March 2010
« Nov   May »





Weather @ Fenway

Vista Themes | Indoor Lighting | Dictionary