To Throw, or Not to Throw

With a runner on first, a pitcher’s approach to a batter changes in many ways. The most obvious is the delivery out of the stretch. More subtly, the pitcher can be distracted by the runner potentially stealing a base. To mitigate the runner’s chances of stealing a base, the pitcher has the option to throw over to first, thereby keeping the runner closer to the bag.

After a few throw overs, I often hear announcers remarking that the pitcher is “distracted” by the base runner, implying that the pitcher is compromising his effort to get the batter out. By expending his mental, and maybe also his physical, energy on the base runner, he commits less towards the batter. In this article, we attempt to demonstrate this proposed effect with a statistical analysis.

To study the effect of throw overs on a pitcher’s success with a batter, we look at the pitcher’s change in Win Expectancy (ΔWE) for all the batting events over four seasons, from 2007 to 2010. During this period, the average ΔWE for each batter is -2.3 bps*. Equivalently, batters had an average ΔWE of +2.3 bps for each plate appearance (PA). In other words, for the period observed, offense/hitters were stronger than defense/pitchers.

*Definition of bps: 1 bps is 1/100th of a percentage point. 100 bps = 1%.

In this period, there was a total 773,123 plate appearances (PAs). To observe trends related to the effects of pitcher throw overs (TOs), we look at the specific configuration where there is a runner on first and no other base. The data shows that there is not much difference in ΔWE when a pitcher TOs or does not TO.

Average ΔWE
TO & No TO: -6.745
No TO: -6.892
TO: -6.380

In fact, the results are counterintuitive: this view of the data indicates that throwing over results in a better ΔWE. In order to see what’s really happening, we dig deeper. The first breakout of data we perform is effect on lefty and righty pitchers. A lefty has the distinct advantage of being able to see the first-base runner, whereas a righty has his back to the runner. Mechanically, one expects a righty to be more out of his comfort zone than a lefty. In the time period observed, both lefties and righties TO 29% of the time.

Average ΔWE, by Pitcher Handedness
Lefty, Total +2.95
Lefty, No TO -0.18
Lefty TO +10.70
Righty, Total -10.61
Righty, No TO -9.56
Righty, TO -13.21

The lefty’s total average ΔWE is +2.95, whereas a righty has -10.61, a very stark difference. This is an indication of how much stronger a lefty is in this situation. In all base configurations (not just runner on first), the difference is much smaller: a lefty’s ΔWE is -1.70 and a righty’s ΔWE is -2.51. Hence, we see there is something unique about the runner-on-first situation. I find it more interesting comparing the pitchers’ differential with respect to TO or not. When a lefty throws over to first, his ΔWE increases dramatically (-0.18 vs 10.70), whereas righties have the opposite effect (-9.56 vs -13.21).

Intuitively, baseball fans assume that lefties are less distracted by the runner on first than righties. This difference in WE can serve as a pillar of statistical evidence.

A last point: Is ΔWE an appropriate metric to measure the pitcher’s level of distraction? I believe so, because ΔWE weights higher-pressure configurations (configuration = run differential, inning, base-out). This is precisely the situations that I believe the pitcher is more inclined to fall victim to being distracted by a runner on first. In non-pressure configurations, the pitcher typically ignores the runner. By using other metrics, we are implicitly saying a pitcher’s success when he is up by 10 runs and doesn’t TO is equivalent to his success when the game is tied and doesn’t TO.