THE BOOK--Playing The Percentages In Baseball

I looked at it a while back and the key for sending the runner home is the number of outs:

http://www.royalsreview.com/2010/4/20/1429439/is-dave-owen-sending-too-many

Here are the needed success rates to break even considering the number of outs and man on 2nd (used 2009 numbers, running on 2nd and single hit):

0 outs: 78%
1 out: 65%
2 outs: 37%

I figured the success rate for plays with xx3 and a fly out.

0 Outs: 71%
1 Out: 33%

Seems like Russell Carleton did a study on this when that saber-espn blog came out.  If I remember, he said 3rd base coaches should be much more aggressive because 90% or more of runners from 3rd to home are safe.  I think he said 78-80something percent safe was the cut-off so to speak.

Yep - pizzacutter did one also:

http://insider.espn.go.com/mlb/hotstove09/insider/news/story?id=4920847

Not to repeat everything we said back then, but if the breakeven is say 80%, that means the average has to be higher than 80%.  And since there are so many gimmes, the average has to be way higher than 80%.

As for calculating the breakeven, I meant to also include the inning and score, in addition to base/out.

Tom, I agree with your general point, but there are a couple of circumstantial injuries that should have led Bogar to be more conservative, I think:

1. Victor Martinez (who is slow to begin with) has been limping around on a bruised big toe/instep due to a foul ball a few games ago. He’s still not back to playing every day.

2. Darnell McDonald banged his knee on a pickoff attempt at 1st base in the same inning as his play. He was hobbling for a bit, and clearly could not run full steam. Also, the ball he was waved around on was a slow roller into short right; short throw means greater chance of a good throw.

Martinez play:

Attempting to score from 1st on double with no outs.

If runner stops at 3rd, based on the RE from the book, we’re looking at 2.052 runs for the inning.

If runner scores, we’re looking at 1 + 1.189 for 2.189 runs for the inning, for a gain of .147 runs.

If runner is out, we’re looking at .725 runs for the inning, for a loss of 1.327.

I’m not sure that I’m doing this next bit of math right, but that looks to me like you’d need to be successful 89% of the time to break even.

From my perspective, you should only send the runner if you’re clearly ABOVE the break-even point, say a 95% chance of scoring. In real time, I was yelling at the tv to hold the runner, as it seemed like a huge risk to send a half-lame catcher for such a small payoff. With all due respect to Geren, I would have put the odds of scoring at 75-80%, based on the player and the situation.

I didn’t have a strong reaction to the McDonald play, and even with all the factors #5 mentioned, I can’t criticize Bogar for it.

Oops. Above I had the RE as if Youkilis had stopped at 2nd instead of going on to 3rd. That changes the breakeven point down to about 86% or so, if I’m doing the math right. That makes it a closer call, but I still think you have to hold Martinez there.

MAtt, can you redo using the inning + score as well?

Bottom 3, nobody out.

Scenario 1: Runner stops at 3rd, leaving 2nd & 3rd, 0 out, -1, WE is .608.

Scenario 2: Runner thrown out at plate, batter advances to 3rd. 3rd, 1 out, -1, WE is .496. WPA is -.118.

Scenario 3: Runner scores, batter advances to 3rd on throw home. 3rd, 0 out, tie. WE is .652. WPA is .042

Scenario 4: Runner scores, ball cut off, batter stops at 2nd. 2nd, 0 out, tie. WE is .623. WPA is .015.

Which is more likely, Scenario 3 or 4? I don’t have a clue, but that makes a huge difference. In Scenario 4, you have a break-even of about 88%; but in Scenario 3, break-even is only about 65%.

So based on game state, this looks like basically a break-even call. I thought there was about a 75-80% chance Martinez could score and if you split the difference on whether the throw would be cut off in half, that puts break-even right at 75-80% as well. I think I’d still hold the runner, but it’s close. And if he thinks the throw is unlikely to be cut off, Bogar is absolutely right to send the runner.

Those are all terrific, but how does the 3B Coach know if the runner has a 75% chance, and 85% chance or a 65% chance?

Chris (10),

Years of experience. My guess is that in the moment, the coach probably doesn’t have a specific percentage in his head, but has more of a feeling of “I think he can score, but it’ll be close” or something along those lines. It’s less about the specifics of any one call he has to make than about being aware of the percentages and when is a good time to take a chance vs. holding the runner.

The point of this exercise for me has been to make me go from thinking Bogar is a slightly over-aggressive 3B coach to thinking that he knows a lot more about what he’s doing than I do or than I gave him credit for. That’s what I like about this site and sabermetrics in general. I’m not a cutting edge guy who’s going to be able to argue the finer points of UZR or whether there’s selection bias in this statistic or that one; but I can use these tools to figure out things that bother me when I watch games like “why does that crazy 3B coach keep waving them in and getting them cut down at the plate?”

Matt, what you did was exactly what needed to be done.  Thanks for doing it.

Actually, a progressive team should be reviewing video and seeing how often a player of speed X scores against a fielder with arm Y given where the ball was fielded and where the runner was when it was fielded.

From that, they can formulate “rules of thumb.” That is the way I would do it.  Do you think any team is doing that?

From the BTF thread:
Generally, and there are slight, very slight, adjustments to be made on whether or not the runner will make it home before the ball: has the runner reached third base before the outfielder has picked up the ball? ON AVERAGE (ball hit speed, OF depth, runner speed, OF arm, next hitter’s suckiness), if the runner has hit third, he’ll beat the throw.

And this is true for all but the slower runners and the better arms. Now, whether or not to *send* the runner you can adjust more for game situation and hope for a bobble by a fielder (or wet grass, or whatever), but if you watch on TV, and you want to show off whether or not the throw is going to beat the runner, simply watch whether the runner has gotten to third yet when the ball is fielded.

MGL, right, that’s what I would do.  I’d figure out a formula for time to pickup to time to catcher for distance and whether the fielder has a straight line to pick up the ball and using a few arm strength / accuracy reports.  Then, for every one of my runners, I’d know their exact speed from touching 3B to sliding home. 

Then, I’d have to memorize the win expectancy breakeven chart for sending runners home (based on who’s on deck).

I’d be totally on board to do that.  Not the best use of my time, granted, but that’s what I would want to do.