Base Stealer Intangibles (Part 1)

You toe the rubber, annoyed. A leadoff walk to Ichiro is not the way to start an inning. Crap! OK, concentrate on Winn now. Right, throw over once or twice to keep him honest. But now focus on Winn. Come to the set, hands at the belt, quick peek over the left shoulder and … (jeez, what’s going on over there?) …

Poised between going on and back, pulled
Both ways taut like a tightrope-walker,
Fingertips pointing the opposites,
Now bouncing tiptoe like a dropped ball
Or a kid skipping rope, come on, come on,
Running a scattering of steps sidewise,
How he teeters, skitters, tingles, teases,
Taunts them, hovers like an ecstatic bird,
He’s only flirting, crowd him, crowd him,
Delicate, delicate, delicate, delicate – now!

“The Base Stealer” – Robert Francis

How can a guy pitch with all that going on? Indeed, it’s often claimed that a dangerous base stealer is effective in disturbing a pitcher’s rhythm, even throwing the whole defense out of whack. In the sabermetric community, the base stealer has lost some of his luster, as many have shown how the outs made by caught stealing largely cancel out the gains of the successful stolen bases. But, does the base stealer help in other, less tangible ways? Does he indeed disturb the defense, and especially the pitcher, merely by the threat of the stolen base, by his “skipping rope” and “scattering of steps”? Many people, I would venture, believe this to be the case; however, I’ve never seen evidence to support that viewpoint. I propose to look for such evidence in play-by-play data available at Retrosheet.

Initial Considerations

If the base stealer at first base is really bothering the pitcher, it should show up in the performance of the batters who come to the plate with the base stealer on first base. So, it’s natural to compare how batters perform when a good base stealer is on first to how batters fare in generic situations. There are two effects that need to be addressed however:

  1. Selection bias in the two samples
  2. Changes in defensive alignment with a runner on first base

By “selection bias,” I mean the quality of the batters represented in the two samples may not be the same. For example, let’s say I look at all plate appearances that occur with a top base stealer on first base, call it Group Man-On-First, and compare it the group of all plate appearances, Group Everybody. Since the top base stealers often bat at or near the top of the order, the batters that follow them are more likely to be middle-of-the-order hitters than the group of all batters. In other words, the batters in Group Man-on-First are just better hitters overall than the batters in Group Everybody. So, a simple comparison of the two groups will lead to the (possibly wrong) conclusion that batters do better with runners a base stealer on first. Selection bias is avoided in the following way: for each batter in Group Man-on-First you record how he did in the man-on-first situations. You compare that to what that batter did in all situations. This “matched batter” technique ensures that the same batters are being compared in the two different situations.

The second effect is caused by changes in defensive positioning with a runner on first. Since the first baseman is generally obliged to hold the runner at first, that opens a hole on the right side that we expect will lead to more hits. Furthermore, if there are less than two outs, the middle infielders will often shade towards second base for a possible double play, resulting in non-optimal positioning for fielding ground balls. These defensive effects need to be quantified before we can say anything about whether pitchers are “rattled” by having a base-stealing threat on first base.

Defense with a runner on first base

I examined the play-by-play data from 2003-2005 to get a handle on how defenses perform differently with a runner (any runner) on first base and second base open. ( Whenever I refer to the situation “runner on first” I always also require that second base be unoccupied.) I focused on ground balls (excluding bunts), specifically the percentage of ground balls fielded by the different fielders in the two cases. The following table shows this percentage for infielders and outfielders.

Percentage of ground balls fielded
+--------------+--------+------+------+------+------+------+
| Situation    | GB     | 1B   | 2B   | 3B   | SS   | OF   |
+--------------+--------+------+------+------+------+------+
| All GB       | 143302 | 0.12 | 0.22 | 0.17 | 0.24 | 0.17 |
| Runner on 1B |  38218 | 0.09 | 0.21 | 0.17 | 0.23 | 0.21 |
+--------------+--------+------+------+------+------+------+

We observe pretty much what we expected: the first baseman fields fewer grounders (33% fewer) and the out fielders field more, i.e. more hits on ground balls. We see the “double-play” effect, whereby the second baseman and shortstop field slightly fewer grounders when there is a runner on first.

We can check if the middle infielders are really fielding fewer ground balls because they are cheating towards second. We don’t expect to see any difference when there are two outs and the double play is not an issue. The following table repeats the previous one, but only considers two out situations:

Percentage of ground balls fielded, two outs
+--------------+-------+------+------+------+------+------+------+
| Situation    | GB    | outs | 1B   | 2B   | 3B   | SS   | OF   |
+--------------+-------+------+------+------+------+------+------+
| All GB       | 43441 |    2 | 0.12 | 0.22 | 0.18 | 0.24 | 0.16 |
| Runner on 1B | 13911 |    2 | 0.09 | 0.23 | 0.18 | 0.24 | 0.18 |
+--------------+-------+------+------+------+------+------+------+

So, with two outs we see that the middle infielders field the same percentage of grounders whether there is a runner on first or not. The data is confirming what we expect: that cheating towards second base in a double-play situation causes the middle-infielders to get to fewer balls.

Note that the positioning of the first baseman has the biggest overall effect on the extra ground balls getting through the infield. The change in the percentage fielded by the second baseman and shortstop is much less. However, many more balls are hit to the middle infielders, so their influence on the outcome is not negligible. I estimate roughly two-thirds of the extra hits go past the first baseman, while the remaining ones get by the middle infielders.

Another important consideration is the handedness of the batter. Most ground balls are pulled, and so we would expect left-handed batters to have an advantage over righties in runner-on-first situations. Here is the data:

Percentage of ground balls fielded, left/right splits
+--------------+-------+------+------+------+------+------+------+
| Situation    | GB    | Bats | 1B   | 2B   | 3B   | SS   | OF   |
+--------------+-------+------+------+------+------+------+------+
| All GB       | 41034 | L    | 0.22 | 0.31 | 0.06 | 0.15 | 0.17 |
| Runner on 1B | 10577 | L    | 0.17 | 0.31 | 0.06 | 0.15 | 0.22 |
| All GB       | 54460 | R    | 0.04 | 0.14 | 0.26 | 0.30 | 0.17 |
| Runner on 1B | 14907 | R    | 0.03 | 0.14 | 0.26 | 0.29 | 0.20 |
+--------------+-------+------+------+------+------+------+------+

As expected, the advantage to the hitter of having a runner on first base is greater for left-handed batters, although right-handed batters do benefit, mostly due to the shortstop fielding fewer balls when there is a runner on first base.

To conclude this section, it’s clear that a team’s defensive alignment changes significantly when there is a runner on first base. The need to hold the runner close to first base and the desire to facilitate the double play lead to a significant increase (about 24%) of ground balls getting through the infield. Note that this result was obtained for any runner on first base, not just the base stealers. Since our goal is to determine if the top base stealers disrupt the pitcher merely by occupying first base, we will need to somehow remove the effects of defensive alignment when we evaluate the hitters who bat with the stealers on first.

Stealers and Hitters

I decided to consider the top 10 base stealers of the period in question, 2003-2005. The number 10 is an attempt to find a balance between getting enough statistics and only including true base stealing threats: the guys you expect might rattle a pitcher. I originally planned to use stolen base attempts to rank the runners, but when I used stolen bases instead, I added Carlos Beltran and discarded Alex Sanchez, so I figured that was good. Here are the most prolific base stealers, who will henceforth be referred to as the Stealers.

A Hardball Times Update
Goodbye for now.
Top 10 base stealers, 2003-2005
Name           Att.     SB
Podsednik      218     172
Pierre         228     167
Crawford       193     160
Figgins        146     109
Suzuki         130     103
Roberts, Da    130     101
Beltran        113     100
Furcal         118     100
Abreu          116      93
Reyes          112      92

This seems to be a decent selection of base stealers.

As you might imagine, the sample of batters who hit with these guys on first base is a rather restrictive set. The one with the most plate appearances (274) is Luis Castillo, who hits behind Juan Pierre, and next is Randy Winn (265 plate appearances), hitting behind Ichiro Suzuki. Here are the top 10 hitters behind the Stealers, in terms of place appearances:

Plate Appearances with a Stealer on first base
Name       PA
Castillo  274 
Winn      265 
Giles, M  201 
Thome     191 
Burrell   158 
Floyd     103 
Lugo       88 
LoDuca     76 
Iguchi     72 
Erstad     71 

In all, there are 216 batters in the sample, but the number of plate appearances falls off quickly. Only 16 batters have more than 50 plate appearances, and only 48 have more than 20. As a group, these batters are better than the average MLB batter, which makes sense since they are mostly top/middle of the order hitters. For this reason, the matched batter technique, alluded to earlier, is important for this analysis.

OK, Some Results…

The next table shows a comparison of how our hitters did in Stealer-on-first situations and how they did overall.

             Fly     LD     GB    Pop      K     BB     HR      H     TB     AB
All Opps:    753    523   1236    206    593    381    105    904   1439   3286
S1B Opps:    798    596   1286    204    502    303     94   1007   1504   3355
    Diff:     45     73     50     -2    -91    -78    -11    103     65     69
Diff Pct:    6.0   13.9    4.0   -0.8  -15.4  -20.5  -10.6   11.3    4.5    2.1

Some explanation on the notation: “S1B” means “Stealer on 1B”, so the first line shows the performance of the sample in all situations, and the second line is for when there is a speedster on first. Note that the “All” lines has been scaled to the “S1B” line for easy comparison. The first four columns refer to batted ball types: fly balls, line drives, ground balls and popups.m I include these batted ball trajectories because they will be useful later, when we attempt to take into account defensive alignment in S1B situations. I am not considering bunts in any of this analysis.

There are several interesting things to note about these numbers:

  • Hits – As we expected from our discussion of defense above, there are more hits in the S1B situations.
  • Balls in Play – There are significantly more balls put into play in S1B situation, as seen by the increase in the batted ball type categories and the decrease in strikeouts and walks.
  • Power – While hits increased by 11%, total bases did not keep pace, increasing by only 4.5%. The main reason is a 10% decrease in home runs.

Let’s show the normal offensive line for the two situations, along with the usual production numbers:

         AB      H     2B     3B     HR     BB      K
All:   3286    904    180     20    105    381    593
S1B:   3355   1007    187     14     94    303    502

        AVG    OBP    SLG     RC   OUTS   RC27 
All:  0.275  0.351  0.438    504   2381    5.7
S1B:  0.300  0.358  0.448    539   2348    6.2

I note here that OBP is calculated as (H+BB)/(AB+BB) (i.e. I’ve neglected HBP) and I use the simplest formula for calculating Runs Created: OBP * TB.

The data show a big jump in batting average for S1B situations, although the gain in OBP is modest because batters are taking (or pitchers are giving up) fewer walks in this situation. Slugging goes up a tick, but that is all batting average: isolated power actually is lower in S1B situations. Using RC27 as a measure of overall batting performance, batters in the S1B situation outhit the other guys (themselves, actually, in generic situations) 6.2 to 5.7 runs per game. So, do we credit those extra runs to the Stealers themselves, who are creating the disturbance over at first base? Not so fast, Sparky, we still haven’t taken into account the altered defensive alignment when there is a runner on first. We’ll have a look at that next time.

References & Resources
As usual, thanks go out to the crew at Retrosheet for making major league play-by-play data available to anybody for free.


Comments are closed.