THE BOOK--Playing The Percentages In Baseball

I can’t believe the BE point is less than 10%. I would never have guessed that without looking at any numbers. You need to get an agent to get yourself on the book tour and get on TV.

Yeah, but you have not played, coached, managed, and commentated baseball for 40 years, as I assume Brenley has.  Anything that is “not even close,” as this is, conventional wisdom should get right. Not necessarily exactly right, but pretty much right.  It should be obvious that with 1 out and 1st and 3rd, you go on just about any ground ball (unless, I guess, you are going to be out more than 91% of the time), and with runners on second and third (and 1 out), you go ALL the time.  That a major league manager (or player) does not know that is unconscionable.

Then again, in the “you’ve got to be kidding me,” as in, “No one can be that stupid to make a comment like that,” department, in the Giants game tonight, the regular Giants announcer (the one with the “attaboy” voice) told us that when you get hit by a heavy sinker, you can really feel it.  In other words, according to him, a 92 mph sinker that hits you in the arm or back feels worse than a 92 mph non-sinker.  That is exactly what he said.  My dog looked up when he said that, and went, “Huh?” Or maybe he was just choking on a hair ball.

MGL:  question: do the RE’s you quote include wild pitches/passed balls/balks/botched pickoffs (i.e. ways for a runner at third to score that wouldn’t score a guy from 2nd)?

I think your logic on “always” going is valid, however I think Soriano may have been right to hold, because I don’t think he had even a 9% chance to make it, he would have run right into the pitcher’s glove had he gone.  But, you’re pointing out the inaccuracy of Brenly’s statement, not Soriano’s (or his 3B coach’s) decision.

Also, I’m not sure if your 9% break-even point includes the possibility of the runner being safe at first, leading to bases loaded and 1 out.  I think from Soriano’s perspective, there must have been a lot of doubt as to whether the pitcher could get Lee at first (which he didn’t, due to a bad throw/catch).  Doesn’t really affect Brenly’s statement, but it would absolutely affect Soriano’s choice at third…

Sure, the RE’s I used include everything.  They are empirically derived, i.e. based directly on what happens in games with each base/out state, therefore they include everything that can and does happen in a game.

And of course, you can add in the chance of the throw going to first and him being safe on an error, etc.  No matter how you slice it, the BE point is going to be in the 8-15% range or so.  Even if it is 60% or 80%, that clearly contradicts Brenley.

And sure, I was not questioning Soriano’s decision to stay (although I actually think he would have made it more than 9% of the time, easily, but I’d have to watch the replay, and even then, sometimes, who knows?).  I was questioning Brenley’s statement, which was crystal clear.  He emphatically said that you don’t go unless the ball is an obvious double play ball because you don’t want to get thrown out at home.

I don’t know how many major league managers know that that statement is nonsense. I have no idea.  It could be 90% and it could be 10%.  But shouldn’t they ALL know it?  I DO know that 90% of them would not understand the concept of a “break even point.” For example, how many times have you heard, “Don’t EVER get thrown out at 3rd base with 0 our 2 outs,” as if 5% or 10% is too often?  I don’t think that most managers know that it is OK to get thrown out at 3rd base with 0 our 2 outs 10%-15% of the time, or whatever the BE point is. 

As I wrote in a previous post, Corey Patterson got lambasted by the Cincinnati commentators for getting thrown out at third on a double steal (the BE point is obviously much higher with a double steal) with 2 outs when it was obvious that only a picture perfect throw got him out (and they even remarked on that).  That is also the mindset of the typical manager.  You get thrown out on a play like that, you did something wrong. You are safe and it is OK (well, occasionally they will recoignize a bad play that turns out OK). Never mind how often you WOULD BE safe or out if you attempted the same thing 1000 times (which is the only number that counts of course).

You are safe and it is OK.

I believe the correct term is “heads-up baserunning.”

By definition, if it takes a picture-perfect play to get a runner out, then we can conclude that the runner was absolutely justified in going.

This presumes that the perfect play occurs less than 10% of the time, and the break-even point is at 90% or lower.

No, you do not just run on any groundball. If its a certain out at home plate (and that factors in runners speed, baserunning ability, where the ball is hit, the catcher, etc), then you do not run.

His point about running on a double play was, that in the time the ball goes from the pitcher to 2B to 1B to home, any runner will be able to score.

In your thinking of running on any groundball, if a grounder was hit right down the line and the pitcher fields it and tags the runner trying to socre, he was justified. That’s crap.

If you can’t score, don’t go. A runner at 3rd with 2 outs or less still has a better chance of scoring than a runner at 1st with 2 outs or less.

Brenly could have stated it better, but he’s talking to “mostly” casual fans who wouldn’t know the difference anyhow. People who understand the game know exactly what he’s talking about.

Sometimes you need to put down the slide rule and just let the game play out. And I’ve played, coached, and umpired for 37 years, so I understand the game. And before you get all mad about me “defiling the game by not using a computer to anayzlye every aspect of it”, think about why I come to this site. And why I link to it from my site.

Just becasue a print out doesn’t back it up doesn’t mean its wrong.

Not only does the “conventional wisdom” in this case contradict the math but it also (i belive) contradicts itself.

Because as I understand it the typical fielding instruction (and that seems to be also how they play) is to NOT go after the lead runner unless you are almost sure to get him and instead take the easy out at first.

Common sense and/or a minimum of logic thinking should draw the conclusion that if a certain play in the field needs to be close to 100% to be worth even attempting that the corresponding baserunning play does NOT need to have a very high chance of success to be worth it.

On another note I think one must also take into account in these calculations that by forcing the defense to try to make these “picture-perfect” plays they will also presumably create more errors, some of which actually give you more than the initial desired result.

MGL,

While Brenly’s inane comment is a good topic starter, keep in mind he does this all the time! For some reason it’s part of his analyst shtick.

Of course, it could be worse. He could be Darrin Jackson, his South Side counterpart.

Let’s go to the numbers, shall we?

Here is the situation: bottom of 7th, 1 out, runners at the corners, home team down by 1.

As luck would have it, I have a limited win expectancy (WE) chart on my site, so let’s use that one:
http://tangotiger.net/welist.html

WE: .504

But, that’s before the PA starts.  Once the ball is hit and is fielded for what is an easy out at 1B, the WE for runners at 2B, 3B, 2 outs is:

WE: .382

That is our starting WE against which we evaluate the alternative for the runner at 3B.

If you go for the runner at 3B, the possible results:
.341: runner out at home, runners at 1b, 2b
.648: runner safe at home, runners at 1b, 2b
.733: runner safe at home, error, runners at 2b, 3b

Let’s say the error happens 5% of the time.  So, going for the runner at home means that the fielding team is going for a .341 hometeam WE against a .652 hometeam WE.  Their starting point is .382.  So, runner scores, WE for hometeam goes up +.270.  Runner out, WE is -.041.

That basically makes it a 13% breakeven play.  If the runner thinks he has at least a 13% chance of scoring, he should go for it.

***

Maybe my Markov is wrong.  As more luck would have it, I have the historical data online as well (1957-2006):
http://tangotiger.net/wins.html

The WE is .531 at the start of the PA.  But, at the start of the decision making for the runner at 3B, it’s .421.

(Note, we expect all the empirical numbers to be higher, since my Markov doesn’t have the home field advantage.)

Anyway, going for the runner at 3B:
.3350: runner out at home, runners at 1b, 2b
.6636: runner safe at home, runners at 1b, 2b
.774: runner safe at home, error, runners at 2b, 3b

The choice is .335 v .669, with a starting point of .421.  That’s +.248 vs. -.086

Here, the breakeven point is much higher: 26%.

***

There are good reasons and bad reasons to prefer the empirical.

The good reason for empirical: it uses actual data for that precise state of affairs.

The bad reason: limited sample size.  You need 1 million plays in that state to get it significant at the 3rd decimal place.

The good reason for my Markov: it looks at what has really transpired in that base/out situation, and knows exactly how often each resultant state will be entered.

The bad reason for my Markov: it does not look at the inning/score, and therefore, doesn’t realize that being down by 1 in the 7th is not the same as being up by 4 in the 3rd.

***

So, you have to ask the runner on 3B the chance that he thinks that he had to score.  I don’t want to hear “I didn’t think I could make it”.  I want to know: If you tried that play 10 times, how often do you think you could make it. 
- Any answer that is 4 or above means he needs to run. 
- If the answer is 0 then he should hold (duh, any time you know you are dead in the water, of course you don’t go for it… we are always presuming in our discussions that the chance of success is between .01 and .99… it is never 0 and never 1… if it is, then the answer is self-evident, and we’re out of business.)
- If the answer is 1 or 2, then it’s a gut feel play
- If the answer is 3, then it might be ok to not run, depending what the breakeven point really is

By the way, the longer he takes to make his decision, then the lower his chances are of scoring, natch.  If the thought only entered his mind when the ball is in the glove of the pitcher, then the chance of him scoring is close to zero, so of course he doesn’t run, as the breakeven point is now 99%.

If you decide, as a matter of practice, to *always* run upon the ball contacting the ground, then the decision is again outside the runner’s mind, and then it’s just a matter of counting up all GB hit by teams that have this mindset to see whether they scored at least 13% (or 26%) of the time.  If it is, then the “always” run motto is justified.

So, somewhere between ball hits ground (breakeven point of 13%) and ball fielded by player (breakeven point of 99%) is where the decision making lies, and the longer it takes, the more it goes from 13% to 99%. 

If the true chances of the runner scoring is say 30%, then if you let him think about it too long, he may price himself out of the attempt.

I think it’s fair to say that in this kind of situation, if you need a rule of thumb (so as to prevent the runner from having a brain cramp), let him take off all the time.

And, I really don’t care how many years people have watched or played the game. 

- I’m pretty sure Psycho Lyons has played far more ballgames than I have, and I know for sure he’s played more pro ballgames than I have, but the very last baseball person I would listen to about baserunning is Pyscho Lyons.

- I think a 22yr old Tim Raines knows more about baserunning, than the combined mindpower, wisdom and experience of every single person reading these words.

Quality, not quantity.

I think that if you include the chance of an error at home plate making everybody safe and moving the runners up one base to be fair you also should include a chance of an error at first making everybody safe with the bases loaded.

IF I am reading your chart correctly (and that’s a big if) and assume the same 5% error chance that would increase the “start state” from .382 to .392 which increases the break-even point by 3-4%, small but perhaps not totally insignificant.

As for a good rule of thumb for base runners how is this: “If you’re trading outs be aggressive, if you are risking outs be conservative.”

One other thing to remember, MLB managers have another objective besides maximizing their RE’s: minimizing decisions that make them and their team look bad.  You can argue that they shouldn’t, but they do.  Running into an out at home where you’re nailed by a wide margin (which would happen frequently if you *always* run) is one of those look-bad plays that stay on a manager’s, coach’s or runner’s “resume” a long time. 

Then there’s the human tendency to postpone potential discomfort (e.g. watch “casual” blackjack players stand on 16 when the dealer shows a face card) even when the numbers don’t back it up.  Holding the runner and hoping the next hitter comes through seems “safer"… So I think Brenly probably summed up the consensus opinion in the league when he said what he did.

Doesn’t change the logic here, of course.  We’re arguing that runners should go much more often than they do, and the numbers are clear on that.  I’m just not surprised that in an industry with very limited job security, the perceived “safe play” is still dominant…

Greg, yes, that is definitely how some of these “wrong ideas” evolve - because of the pressure of wanting one’s job to be safe, avoiding immediate discomfort, etc.  It is not a conscious thing for a manager (Brenley does not whisper to his partner, “I know that is wrong, but that is how I managed because I didn’t want to lose my job") of course.

As I said, no matter how you look at it, (9%, 26%, or 40% BE rate), Brenley’s comment and his thinking, assuming that he wasn’t just talking out of his a**, and would not follow his own advice in a game, was clearly, 100%, wrong.  If the BE point were 85 or 90%, I would give him the benefit of the doubt and assume that when he says “never” he means “almost never,” but that is not the case - the BE point is most certainly less than 85 or 90%.

As far as how you instruct your players, you simply tell them the truth, although you can put it in more understandable terms.

“If you think you have a decent chance of making it, like 3 or 4 times out of 10, then go for it.” Simple as that.  Then each time a player does or does not go for it, you make a mental note and speak to him if you have to.  You also want to make sure that when he gets thrown out, if you think that he still made the right decision, that you let him know that.  Some players will tends towards running the bases to minimize their times thrown out no matter what you tell them the requisite BE point is.  If in your estimation as a manager, the player had a 30 or 40% (or whatever the BE point is) chance of making it, and he gets thrown out and says, “Sorry, Skip, I made a mistake,” you need to be sure and tell him, “Hey, you did the right thing.  Remember I told you on that play as long as you think you have a 30% chance of making it, you should go.  That means you will be thrown out more often than you will be safe. That’s OK.  That will win us more games in the long run, which is the only thing I care about.  No need to apologize.  You did the right thing.  Good job!”

How hard is that?

For players brought up on the idea of a .300 batting average is great, you’d think they’d understand that being safe 30% of the time on a 13% breakeven play should be celebrated.

Tango/#16:

You’re forgetting how far out on the tail of the intelligence distribution you are…

When one of these percentage plays goes wrong, it does the managers little good if the Book Blog regulars applaud the choice and shrug off the outcome, but the GM, Owner, Media and other 99% of the fans all have a cow…

It will take time for the mass of people to see things this way, but I think if we keep talking about this, and keep patiently explaining, we can eventually mainstream the concepts and the implementation.

Sort of reminds me of nuclear power.  The public perception is long on emotion and hyperbole, and short on facts, but it is beginning to be rehabilitated after a long time in the doghouse…

I think there is no question that the level of sophistication of thinking by GM’s and managers has and will continue to progress.  As with player evaluation (which has gotten infinitely better), you either have to keep up with the best or you fall so far behind that you can’t compete.  Plus, people love to follow what other people are doing and generally don’t like being at the forefront of new (and scary, to them) things.

I’m not a defender of Brenly in general or, for that matter, of Soriano’s baserunning, but I will say this, on the play in question:

The ball was hit right down the third base line.  Soriano would have been just past the pitcher when he fielded the ball and it would have been a much easier play going home than to first.  Running gives him a 40 ft throw instead of a 110 ft throw, and the longer throw is also directly opposite the direction his body is moving.

Second, your argument from averages is fine, but I think the disconnect is more likely to be that they’re working from the median instead of the mean.

dbt, as I said, I was NOT making any comment whatsoever about the Soriano play.  Only on Brenley’s comment, which was about that situation in general.

Later on I mentioned that I thought that Soriano should have gone, but that I would have to watch the replay and even then, I might not be sure.  But whether Soariano should have gone or not had nothing to do with my initial post.