Skubal's Double Play Decision Debated

[John Bonnes]

Lost in the last night’s Minnesota Twins’ dramatic 4-3 win over the Tigers was a split-second call in an unusual situation that potentially pits baseball traditionalists versus sabermetricians. Or maybe it doesn’t?Holding a 1-0 lead, Tigers starting pitcher Tarik Skubal was in a jam in the bottom of the fifth inning. He had runners on the corners, and no outs, with Eddie Rosario at the plate. Rosario hit a one-hopper back to Skubal, and Skubal had a fraction of a second to decide whether to throw to second and start a double play, which would allow the runner (Jake Cave) to score. Or he could throw home, almost assuredly Cave out, but only get one out. He chose the former, and the game was tied.

 

Immediately Twins broadcasters Dick Bremer and Justin Morneau debated the choice Skubal had made, wondering if he would do the same thing if offered it a second time. On the one hand, the chances for a big inning had been short-circuited, and indeed the inning ended in a tie, and the Tigers regained the lead in the next half-inning. On the other hand, the lead was blown, and the Tigers offense had not gotten a hit, let alone score runs, against Twins starting pitcher Kenta Maeda after he gave up a leadoff home run to the first batter of the game. Actual runs were at a premium.

 

Even after the play, the right answer for Skubal is not obvious, either to traditionalists or (as we’ll find out) to sabermetricians. But sabermetricians attack it in two parts: one strategic and one practical. The strategic is probably the more controversial.

 

Strategically, the fact that this is the tying run is not particularly important since it is only the fifth inning. The choice should be to minimize the expected number of runs scored because there is a lot of game left. Indeed, five more runs were scored after this play. Late in the game, the actual run might mean more, but in the middle of the game, the actual run is not as important as improving your situation as much as possible.

 

Which gets us to the second half of the question: is it preferable to have runners on first and second base with one out or have the bases empty with two outs but a run already on the board? To know that, sabermetricians refer to a Run Expectancy (RE) Matrix. It was developed by authors John Thorn and Peter Palmer in The Hidden Game of Baseball, which was published in 1984. It proved to be one of the foundations on which sabermetrics was built.

 

They loaded decades of baseball games into a computer and crunched basic numbers, including the average runs scored in an inning after a specific situation. There are only 24 situations in an inning. Here they are, along with the average runs scored after that situation in 2019.

 

Download attachment: RE 2020-09-05 empty.JPG

 

Highlighted in blue is the jam where Skubal found himself. Highlighted in red is where things sat after the double play. So the RE dropped from 1.79 to .11 EXCEPT he also gave up a run. So really, it went from 1.79 to 1.11. His actions saved his team .68 runs.

 

Download attachment: RE 2020-09-05 DP.JPG

 

But what if he went home? That depends on how the rundown goes. If both runners stay at first and second, that’s the best possible outcome: a RE of 1.00. If the rundown drags on and they each advance a base, that’s the worst possible outcome: RE of 1.43.

 

Download attachment: RE 2020-09-05 home.JPG

 

Our analysis shows a couple of things. First, almost any decision Skubal made was defensible. He was handed a gift, and he did a solid job, and making a slightly better decision would have only gained him another tenth of a run.

 

It also shows that the answer really is not obvious from any standpoint, even after the play. The default sabrmetric answer defensively is usually “take the outs” but in this case, he might have helped himself more by going home, provided the Tigers were efficient in getting Cave out before the runners advanced. Traditionalists who value the lead, especially those used to watching baseball in decades where runs were more valuable, probably agree.

 

Click here to view the article

[Nine of twelve]

I'm not a sabermetrician, but run expectancy is not the most relevant matter. As the fifth paragraph hinted at, win expectancy is what should be considered.

[Dantes929]

I think most people, even old school would take the double play in that situation.   The sabermetrics analyze averages but only to a certain degree.   Its based on historical data compiled by every kind of team including great offenses, weak offenses, great relievers, and weak relievers on both sides.    No team is exactly average.  Even that doesn't go far enough.   What are the percentages for Cave or the guy following him performing against the relievers the Tigers bring in?    To illustrate my point sabermetrics would say in a tie game with two outs and a guy on third in the bottom of the 9th a routine ground ball to shortstop would be a wonderful thing for the visiting team.

[Seth Stohs]

I would think most would say take the two outs... 

 

I probably would. Justin Morneau certainly suggested he thought going home was the right play. 

 

I think that it's not really bad either. If you go after the guy at third, and he stays in a run-down, you suddenly have 2nd and 3rd with one out... but no runs scored yet. 

 

Tough decision. I don't think that Skubal made the wrong decision, but it's possible there might have been a better one, but i don't think it's 100% 

[AlwaysinModeration]

Interesting analysis. I actually think he made the right choice. While the RE shows he erred 1.11 to 1.00, that 1.11 means there are many scenarios where the inning turns into a big crooked number, even after preventing the first run. They would almost certainly score one of the remaining runners on base anyway, and additionally they have a decent chance of scoring 2 or more runs in the inning.

 

Taking the double play in a low-scoring game, and almost assuring that you limit the damage to one run by clearing the bases with two outs, is the right call, in my view. Basically you are saying, once men got on first and third, ok, you got a run, but we are limiting it to just one.

.

[USAFChief]

I'm not a sabermetrician, but run expectancy is not the most relevant matter. As the fifth paragraph hinted at, win expectancy is what should be considered.

I would almost never pay any attention to "run expectancy."

 

No two situations are equal, and using a composite average to evaluate specific situations doesn't appeal to me.

 

In this specific situation, I was glad to get the game tied over 1st and 2nd, 1 out, no runs in, and the 2nd half of the lineup due up against a pitcher throwing well.

 

I dont think there was much chance of the two runners advancing to 2nd and 3rd, either. The most likely outcome is Cave keeps running and is tagged out on a slide at home. Even if he pulls up, the catcher can just run him back to third, make one throw, and no other runner will risk advancing.

 

BTW, its not by accident or mistake that Cave WAS running on that play. You teach runners in that situation to go on a ground ball, forcing the defense to make a decision. Worst case scenario is Cave retreats to 3rd, and they turn a DP anyway.

[ande4148]

I don't think it's appropriate to compare the prior RE with the post RE in this situation.  I think you should compare the post RE for both events.  1st and 2nd ,1 out: RE = 1 run; 0 on ,2 outs: RE = 0.11 runs.  After the event that stopped the run scoring, you expect 1 run.  After the event that allowed the run to score, you expect 0.11 runs.  Therefor, it is 1 run versus 0.11 + 1 (the run actually scored).  The right choice is to take the out at home, ignoring the time in the game.

 

Using the prior RE of 1.79 and only adding the 1 run that actually scored to the 0.11 post event RE, ignores all the possibilities where more than 1 scored before the post event state.  Using 1.79 for comparison requires including all those possibilities.

[bighat]

 

In this specific situation, I was glad to get the game tied over 1st and 2nd, 1 out, no runs in, and the 2nd half of the lineup due up against a pitcher throwing well.

 

 

Same here.

 

This was a 1-0 game in the 5th, runs were clearly hard to come by. Sacrificing 2 outs for a run -- in this particular game -- seemed like a good deal for Minnesota.

 

By his body language after the play was complete, I think Skubal regretted his decision as well.

 

[tarheeltwinsfan]

 Here's my North Carolina formula for this question: 1 + 4 +3 = 2   > 1+4+3=1

[Trov]

One thing the article does not address is situation.  I get the average of runs scored in a particular situation over the years will try to mitigate the good lineups vs bad line ups.  I am sure bases loaded with a HOF hitter up has a much higher run average than with a pitcher hitting.  Baseball has, and always will, have to take each individual situation to worry about.  Point in game, who is coming up next, who are the runners on, how good is your team doing, ect.  

 

I agree either decision is debatable as the right one.  A good manager would have communicated to team the plan, and since it would seem middle infield played back the plan was to turn two.  The pitcher knew that and did the plan.  Sure, he could have changed it up if he saw Cave heading home, but possible he did not see it, or he was doing what his manager told him to do, and possibly what the players on the field were telling him to do.  

 

Personally, I am a fan of stopping runs from scoring.  You know by going for two you give a run, by going for out at plate you cut down run, keep the lead and go get the next hitters.  Sure, in different situation you let that run score and get the two outs, but to me this was not the time.  

 

Good read on pointing out the runs expected, but feel more info on situation could affect how Tigers should have played it.  Twins clearly have gone with the run prevention plan all season bringing IF in early in games.  They must have crunched numbers and found that playing back does not increase the likelihood that you will get an out and let run score more enough to risk letting the run score.  That comes down to how often a player playing back would make an out that a player playing in would not.  My guess they found it was very limited compared to alternative.