Thanks to our friends at Inside the Book, we can see the math behind the decision, and in Jerry Manuel's case, he failed when he had GMJ bunt Luis Castillo over in the 9th.
The situation was man on second, no outs, and GMJ had worked himself into a hitter's count. In that situation the run expectancy, per the chart, is 1.228. That's right, the Mets should have expected to not only score that run, but they had a near 23% chance of tacking on another one. The correct call was to remove the bunt sign (which never should have been given in the first place).
When Jerry bunted Castillo over, he lowered the run expectancy to 0.980.
I'll also add that with the players that were due up, there are actually more reasons not to bunt. This was not your 7-8-9 hitters, but 3-4-5, the heart of the order.
David Wright did fail, that is a given, but the manager failed as well.
Jerry wants his team to win with 26 outs.
Here are the runs expected based on men-on-base and the number of outs in the inning:
| Situation | No Outs | One Out | Two Out |
| Empty | 0.551 | 0.295 | 0.111 |
| 1st | 0.972 | 0.576 | 0.252 |
| 2nd | 1.228 | 0.746 | 0.337 |
| 3rd | 1.424 | 0.980 | 0.394 |
| 1st/2nd | 1.649 | 1.027 | 0.478 |
| 1st/3rd | 1.845 | 1.261 | 0.535 |
| 2nd/3rd | 2.101 | 1.431 | 0.620 |
| Loaded | 2.522 | 1.717 | 0.761 |
![Reblog this post [with Zemanta]](http://img.zemanta.com/reblog_e.png?x-id=a423101e-e201-420b-a97d-c0bc978bcbab)
11 comments:
Spot on!
That pic is hilarious BTW.
Would it have been better if Matthews struck out? The numbers disagree.
What if he hit a ground ball to shortstop? The numbers disagree.
How about a shallow fly ball? Or liner?
Ultimately, if Matthews does not advance the runner, then the numbers say that the team is worse off than if he bunts the runner along -- 0.746 vs. 0.980 for a successful bunt.
Right now, it's far more likely Matthews strikes out than any other outcome (he's striking out 34% of the time). Second most likely outcome is that he is out; unless he hits to the first-base side (Matthews is right handed, so the chances of that are reduced), the runner does not advance.
The chances of advancing the runner on a bunt toward first are far greater than the chances of advancing him with anything else Matthews can do.
So the run expectancy of 0.980 is an improvement over not bunting.
Math isn't that hard, but you have to pay attention to what the numbers actually say, and factor in what the batter is likely to do.
Did you factor in the 3-1 count into your equation?
I would normally agree with you... but this was GMJ at the plate. Really the best we could have hoped for was him moving the runner over (what are the % of GMJ getting a pitch hit... like 1 in 15?), so asking him to bunt is trying to up the odds. Plus, they only needed that one run.
Dave,
These are interesting data, but they aren't relevant to the situation last night, these are (Tables 4 and 5):
http://baseballanalysts.com/archives/2006/07/empirical_analy_1.php
The Mets were not looking to maximize the runs they scored, they were looking to score one run. The probability of scoring 1 run is higher with a man on 3rd and 1 out than it is with a man on second an 0 outs.
I was about to point out obvious fallacies of your argument but I see some of them (and perhaps most important ones) were already addressed by Greg and Anonymous. The use of expectancy matrix is really tricky here because it only gives you expected values, not standard deviations. And, as a matter of fact, the second situation has a much lower standard deviation. Consequently, while bunting decreases the expected number of runs, it increases both the probability of scoring at least one run and the probability of winning. If you do not believe me, believe fangraphs. You can check there that the bunting actually increased the probability of the Mets win.
To my mind, if you're going to pursue this strategy, though-- trading outs for a much-surer-shot at one run (you can argue the wisdom of this, too)... you pursue it to its logical endpoint.
As long as you're trading outs for run-probability... squeeze the damn run in.
As already pointed out, your run expectancy is wrong, because odds of a single run with a man on third and one out. Having said that, Mathews was in a pitcher's count, and strikeout guy is up next, perhaps changing the odds. The bigger sin is Jerry's constant use of castillo to bunt Reyes from first to second in close games. The run expectancy is about equal between runner on first and no outs and runner on second with one, and Reyes has a great chance to steal.
Great discussion...yes, there are a few different ways t look at this, and I chose to look at run probability based on situation.
When the wild pitch occurred and GMJ got the count in his favor at 3-1, you don't bunt.
I personally would not have bunted anyway, but at the point in the game where he did sacrifice successfully, I felt that the math was far more in our favor with a hit attempt there.
Had he not been bunting, I doubt he would have swung at ball four (the high pitch he bjnted), and then we have an entirely new set of math to deal with.
Good stuff guys and appreciated.
For you people suggesting "but it was Gary Matthews Junior bunting", Manuel brought in GMJ to bunt. He could have used Tatis instead.
@Chuck Rothman: You completely butchered the math. You assume there is a 0% chance of GMJ either gets a hit or advances the runner on some sort of fielder's choice, error, wild pitch or some other crazy way. Of course a successful sacrifice bunt is better than an unproductive out but that doesn't mean you get to ignore the non-zero probability of a productive out or better yet, safe passage to first.
That's right Moose...as a hitter there are a myriad of ways that GMJ could have moved Castillo over with an out...ground ball, fly ball, etc, not to mention the chance he scores Luis on an actual hit (and god forbid GMJ hit a HR).
Bunting was a losing call, by a losing team, with a manager that has one tool in his toolbox.
Post a Comment