## Running Into an Out as a Strategy

I tried to come up with a witty preamble to this but all I could come up with was a lame story about playing RBI Baseball 4 against my older brother. And unless you have mistakenly come to FanGraphs while trying to get to Farmers Almanac (no judgments, Google auto-complete can be weird sometimes) then you probably don’t care about that. So let’s dispense with the amusing introduction and get right to the question. (Or did I just subversively come up with a witty preamble by explaining how I did not have a witty preamble?!)

**Scenario:**

Runner on first with two out. 0-2 count.

Now anyone who is even slightly familiar with baseball will tell you that this is not a good situation for the offence. Those who are very familiar with baseball to the point that they read things like this post will probably even quote the run expectancy matrix to demonstrate how bad of a situation this is for the offence.

So, yeah, not looking good for the offence. The chance of scoring a run from that base/out state is 0.127. And that is without even accounting for the 0-2 count which obviously makes things worse. MLB as a whole slashed .155/.187/.237 with a 47.6 K% and a 10 wRC+ last year through two-out, 0-2 situations with runners on. In other words, the batter made the third out ~80% of the time. Even Mike Trout, who is Baseball Jesus, strikes out over half the time in 0-2 counts and is running a tOPS+ that is almost single digits. For all intents and purposes, the inning is likely over when it hits that situation.

But the team at the plate is not totally powerless. It can still decide **how** to end the inning, and they could do it in a way that gives them a more favourable outcome. Which brings me to the crux of this argument;

*Why not have the guy on first just take off running?*

Before the pitcher even comes set, just take off for second. Worst case, they tag him out and the inning ends (which was the most likely outcome anyway), but now the guy at the plate leads off the next inning in a fresh count, which is obviously a much more favourable scenario for a hitter. And best case, the defence screws up and the runner is now on second. Granted, that is an extreme outcome, and even two out and runner on second is still not a great scoring scenario. But referring back to the run expectancy matrix, it’s ~50% higher than when he was standing on first.

If the outcome of the scenario is almost overwhelmingly going to be an out, then you are not really giving away an out as much as you are just deciding who takes the out. If you have a good hitter at the plate, why have him continue to hit in what is a pretty futile situation, and waste one of his limited PAs, when you can reset the situation and give him what amounts to an extra PA by having the runner take the out instead?

Let’s look at Mike Trout’s career as an example since, well, since it’s fun to look at Mike Trout’s numbers.

No surprise, Mike Trout is a much, much, much better hitter overall than he is in 0-2 counts. Every hitter is. Now let’s also check back in with our friend, the run expectancy matrix.

So right off the bat (no pun intended), we see that the chances of scoring a run at the start of any inning are considerably better than scoring a run with two outs and a runner at first. Add in the fact that you have a very good hitter leading off in Trout and things have seemingly changed significantly for the better, simply by having your base-runner act like an 11-year-old exchange student on the base paths.

If Trout does *anything* to get on first (single, walk, HBP, dropped third strike, coming to the plate and performing a stand-up routine that is so good the opposing team just awards him first as a thank you, etc etc), now all of a sudden the chances of scoring a run in the inning have gone up to 0.416. Given that Mike Trout got on base nearly 45% of the time last year and is around 40% for his career, it seems like a fairly reasonable outcome. So by having your base-runner deliberately make an out to end the previous inning and saving Trout from doing so, you have gone from a situation where you had a .127 (or lower given the fact that the 0-2 count is not accounted for in the matrix) chance of scoring a run and your best hitter producing an out to a situation where you very likely have a 0.416 chance of scoring a run. And that does not even account for all the other things Trout might do new in this new PA. If he hits a lead-off double, your chances of scoring a run in the inning are now 0.614. If he hits a lead-off home run, your chances of scoring a run are….hold on, where is my calculator? Plus, you have also avoided what was highly likely an out for your best hitter and having to wait two or three innings for him to bat again.

Last year, MLB teams averaged 219 PAs where they had runners on and an 0-2 count. As stated above, in that situation the hitter wound up making the third out ~80% of the time. So that is ~200 innings that could have started with a different guy at the plate and ~200 outs at the plate that could theoretically have been something other than an out. How many innings would have been different by simply giving up the runner for the third out and letting the hitter lead off the next inning in a more favourable count? If you have a good hitter at the plate and he is down 0-2, it might be worthwhile strategy to just tell your base-runner to take off and let your hitter try again the next inning.

Or maybe I have had too much coffee today.

Print This Post

Very interesting concept marred by less interesting attempts at humor.

If your batter makes an out, there is still a .268 chance of scoring the next inning. That means that you’d need to compare the chances of scoring in NEITHER inning to the chances of scoring in one inning. (This should also be weighted for possibility of multiple runs, but that’s a bigger discussion)

The math would work out to:

(1- (chance of not scoring in 1st inning AND chance of not scoring in either inning)= player’s OBP * chance of scoring with man on 1st and no outs.

(1-(1-0.127)*(1-0.268))=OBP*.416

This would require a player with an OBP of .867 to be mathematically viable.

I think your math is a bit flawed because the RE matrix is just the average outcome across all situations. You can’t really apply it to a specific hitter. Mike Trout is a much better than average hitter so it would follow that the RE with him leading off is likely significantly higher than the average .268 which is something the article touches on.

The real question, IMHO, should be what is the difference between the guy hitting with 2 out and the guy hitting behind him in the order (because he would lead off the next inning if the previous hitter makes the 3rd out.) In the specific case of Trout it would be Kole Calhoun most of the time.

Trout has a career wRC+ of 168 and Calhoun is at 115. So by the laws of wRC+ Trout is 53% better at the plate than Calhoun which is a pretty massive difference. It’s probably safe to assume that the RE with Trout leading off is significantly higher than Calhoun (and that logic would likely apply to any teams best hitter).

Now is that difference big enough to justify the runner sacrificing himself once the count gets to 0-2 in order to let Trout have another PA leading off the next inning? I don’t know but its certainly an interesting question.

In order to solve this fully, perhaps you could create a player+lineup specific run expectancy chart. Once you have that, you will want to add together the RE’s of this inning and the one after to see which strategy is more efficient.

RE “Run into out” = SB%*(2 outs, 0-2 Exp)P1 + (0 outs, 0-0)P1

RE “Play normally” = (2 outs, 0-2 Exp)P1 + (0 outs, 0-0)P2

If P1 (player one) is a much better hitter than player 2, then run into out is more likely to be worthwhile. However, for this RE chart, player 2 will still be hitting in inning 2 no matter what, so you are just potentially reducing his leverage.

This is really complicated as there are a couple issues.

The math depends more on the profile of the batter than the quality of the batter. Guys with worse splits 0-2 (is that even a skill? Probably high-K guys would be susceptible) are better candidates to run with. Also, guys with good OBP and low power are better candidates to run with (because the value of a single or walk with a runner on first with two outs is comparably low, but when leading off an inning the value of singles and walks are comparatively high. Conversely, guys with low OBP and high power are terrible candidates to run with, because a double/homer is gonna be more productive with that runner on first and two outs than it would be leading off an inning.

So you might need to recalculate linear weights by base-out state (e.g. a walk leading off is worth ~.5 runs, but a walk with two outs and one on is only worth ~.2 runs) and then try to tease out how counts affect the distribution of outcomes. And THEN I guess you can compare whether you’ll make score more often by letting him hit and then playing the next inning or by punting the inning and playing the next inning.