Nationals Baseball: The extra inning question

Thursday, May 19, 2022

The extra inning question

The Nats won last night - after a "fun" ninth full of blown saves and initially blown calls. They won in extras by doing something that is somehow counter-intuitive in the ghost runner era. They bunted the runner over. 

There's a lot of agreement on what you should do and that's not it, but oddly enough there isn't a lot of research into proving that agreement makes sense. And personally I think that agreement is wrong. 

I'll note here that this is a very general talk. A lot of baseball decisions have to the with the EXACT situation on hand. Who's on second, who's up at the plate, who's up after them, who's pitching, how deep and good is your pen, and so on. Blindly following any rule is not the way to go. However, there can be a general guideline. 

The general consensus is you don't bunt the runner over. A man on second with no outs and a team trying to score as many runs as possible will usually score more runs than a team trying to score as many runs as possible with a man on third with one out. That out matters a lot. 

However, I posit that the first run is so much more important than the subsequent runs that the most important thing is that you score at all and you are more likely to score at all man on third, one-out than man on second, no-out. 

Early analysis into ghost runner data showed some things. Generally the winning team scored more than one run. Generally the away team won. Generally though, the teams that bunted men over won more. But there aren't a lot of one-run games so it's a lot of small sample size work. Though each year we do add to it.

There are ways of figuring this out. Getting all the games from 2020 on and analyzing them is probably the best way. Using expected runs to estimate outcomes. Generalizing strategy and running thousands of simulations.  It'd be great if someone did this, but as extra-inning games aren't all that common and the advantages are likely slight in either direction it's not a priority for teams. They weren't even sure it would be back. One of you must have the time though.  Me? I'm busy! I got a job! Kids to feed!

Expected runs would be the easiest but the problem is the numbers we have are not situational.  We know runner on 3rd, one out scores a run more often than runner on 2nd, no out.  But we know that in context of every inning of every game where it's more likely you are always trying to score as many runs as you can. What if you were JUST trying to score 1 run given the first scenario? It's likely that slight advantage that exists now grows. But to what? And if the road team doesn't score the home team almost certainly tries for one, also changing expectancy. We must also consider whatever the road team scores, the home team could play for that number - trying to extend the game more than trying to win it. Do they? I'm sure it happens. But is it the prevalent way the game is played?

Simply analyzing scores doesn't tell you what you need to know.  I can be playing to score as many runs as possible, but my first guy grounds to the right side and my second guy flies out and the last guy strikes out. I get a run but it's no different than if I just tried for one. I can be playing to score one, bunt the guy over, get IBBd to set up a DP and then my batter's attempt to just get one up in the air lands over the fence. Three runs when all you were going for was one. And let's not forget the potential pressure and situation. Teams do hit worse late & close, or in the 9th and extras, or when behind. If you score at all and you can put that pressure on and you can bring on maybe a better pitcher than you would if tied, shouldn't we factor that in?

This is why I think analyzing actual games is the best way to do it. There's assumptions we can make on strategy and then there is the way the game is actually played. And we CAN separate out "bunt first" half innings from others to see if there is success.

But until then right now we're seeing it comes down to feeling. A lot of people want as many runs as possible. I want just one.


elchupinazo said...

A lot to consider, strategy-wise. If you're more likely to score with one out and a man on third, the away team has plenty of incentive to bunt the guy over—make the home team HAVE to get at least one to keep it going. But then you also reduce your chances of putting up a big number that would all but end the game. I'm glad I don't have to make these decisions.

Jon Quimby said...

It matters where you are in your batting order, no? If the first guy up is your worst hitter then don't you just bunt the guy to third. If the first guy up is your best hitter then no chance he's bunting. If the ninth spot didn't lead off the 10th I think there's a good chance it would have been played differently.

Cautiously Pessimistic said...

I was bored, so I went through all the 2020-2021 extra innings games, of which there were 284 according to retrosheet. I then only looked at the games where there was a sacrifice bunt in extra innings, which brought the total down to 85 games (surprisingly high given Harper's intuition about it being a consensus bad idea).

In situations where the away team bunted, they went on to win the game about 2/3 of the time (29/44). In situations where the home team bunted, they went on to win OVER 80% OF THE TIME (34/41). I'd argue teams should be bunting more based on those numbers alone, but I'm also too lazy to dig further into the situations

ocw5000 said...

Also the handedness of the first batter might matter, if a LHB you might be less likely to bunt since a routine groundout likely advances the runner anyway

NG said...

Surprised no one mentioned the true highlight of last night's game, which was Kevin Frandsen's commentary in the bottom of the 8th that Alcides Escobar striking out with the bases loaded and no one out was a "good strikeout" because, essentially, Alcides is so bad that if he put it in play, it probably would've been a double play. Alcides, meet bus.

I'm sure analytics would say Frandsen is totally wrong, and of course Hernandez grounded into a double play immediately after the strikeout. But you know it's gotten bad (and it's time to try literally anyone else) when your own announcer is openly rooting for you to strike out because he thinks it's better for the team.

dc rl said...

Frandsen really said that? That's great!

But when the next batter, Cesar Hernandez, hit into a DP, did he say "See, I toldya so!"

The Ghost of Ole Cole Henry (JDBrew) said...

I think the biggest thing to consider when looking at the sac bunt, when you’re the away team, is what shape your bullpen is in to close out the game. If you have confidence you can shit down the next half inning, then why not? One run is all you need. If you’re the home team and still tied in extra innings then absolutely try and scratch out that run. Bunt that guy to third, then all you need is either the ball in the air, a wild pitch, or a swinging bunt down the first base side.

I firmly believe that a run is MORE likely to score with a man on third and 1 out vs. a man on second with no outs. I would have to do a hard dive into numbers to flesh that theory out. But when you take into account all the ways a run can score from third without a batter actually getting a hit, I just have a hard time buying that a runner in second has a higher chance of crossing the plate.

The Ghost of Ole Cole Henry (JDBrew) said...

Hahaha, I read this while trying to feed my own kids and completely read your conclusion opposite of what you were actually saying. Basically, the whole point of my above post is…I AGREE! Someone please take the time to dig through numbers of the past 5 years of runners being on second, with no outs, in a tie ballgame, 9th inning or later, and compile the outcomes to determine the most prudent course of action. Clearly baseball teams don’t do this right?

The Ghost of Ole Cole Henry (JDBrew) said...

Don’t know the accuracy of the stars. But there is this:

NG said...

I think that article has it basically right. The best source of data I know on this is the Tom Tango run expectancy matrix:

The numbers in this article are close but slightly off, so perhaps more recent data.
But in any event, it is essentially it's a trade-off. The expected total number of runs is higher with man on 2B no outs (1.1 runs) than man on 3B with one out (.950). But the chance of scoring at least one run is higher with man on 3B with one out (.660) than man on 2B with no outs (.614).

And, as the article states, this assumes the sacrifice works 100% of the time, which it doesn't. But on the flip-side, defending teams also sometimes fail to make an out when you try to sacrifice. So those things probably close to cancel each other out.

So if you're the home team and the visiting team has not scored, it probably does make sense to bunt -- you only need one run to win, and a successful sacrifice increases the odds of that. For a visiting team, it probably does depend on state of bullpen, etc. -- is a slightly greater chance of 1 run worth reducing your chance of 2 or more runs by an even greater amount?

Expos 1983 Blog said...

Gotta love the way these Nats are gaming the Pythagorean Formula - scoring crooked numbers once a week to maintain that steady 1 and 6 pace!

Barney said...

About 9 years ago I used Retrosheet stats and some code to find out if bunting was a good idea or not. No reference to inning.
Teams scored more often when having bunted than when they didn't.
I didn't ask any questions about situation, but felt that bunting was incorrectly vilified. Clearly, you don't ask a batter to bunt if he usually can't. And, don't take the bat out of the hands of your best hitter.
But, when a manager chose to use it, it was usually a good idea.