Learn how probability, math, and statistics can be used to help baseball, football and basketball teams improve, player and lineup selection as well as in game strategy.

Loading...

From the course by University of Houston System

Math behind Moneyball

43 ratings

At Coursera, you will find the best lectures in the world. Here are some of our personalized recommendations for you

Learn how probability, math, and statistics can be used to help baseball, football and basketball teams improve, player and lineup selection as well as in game strategy.

From the lesson

Module 4

You will learn how to evaluate baseball fielding, baseball pitchers, and evaluate in game baseball decision-making. The math behind WAR (Wins above Replacement) and Park Factors will also be discussed. Modern developments such as infield shifts and pitch framing will also be discussed.

- Professor Wayne WinstonVisiting Professor

Bauer College of Business

In this video we'll begin our discussion of baseball decision making.

And a lot of this, I think, will relate to what you know if you've

read the book Moneyball or seen the movie Moneyball.

So the first thing we'll talk about is bunting.

So if you've read anything about Moneyball, you would know

teams that believe in analytics or Moneyball don't put very much.

Now why is that?

Well, the key to understanding this in a lot of baseball decision-making

will get what's called the run expectancy table.

So remember from our Monte Carlo simulation,

there are eight on-base states.

Okay, we've had none on, man on third, man on second, second and third.

Man on 1st, 1st and 3rd, 1st and 2nd, bases loaded and zero, one or two outs.

And the key to really understanding decision making is to say

how many runs can I expect to score from a given situation?

So when an any starts with none on but

nobody outs you can average 2014 teams averaged about.46 runs.

How would you figure it out?

You look at every inning at the start.

How many runs were scored?

And how would you figure out this number, for instance,

if we start with the bases loaded and one out?

How many runs on average are scored?

You look at every inning where that situation occurred and

see how many runs were scored from that point on.

The answer is 1.51.

Okay so then it's really simple to explain in theory why bunting should be.

All right, so let's suppose we have man on first, nobody out.

How many runs on average can we score?

So man on first, nobody out.

It's going to be better than 0.45.

We expect to score- 0.82 runs.

That's right here.

Okay, let's suppose we bunt.

What we're doing is giving up an out for a base.

We'll have one out, a man on second if the bunt works.

And the bunt may not work.

We can hit into a double play.

I mean we can basically strike out.

We could beat it out.

But, basically, what happens if the bunt succeeds?

What you're trying to do when you bunt is have a man on second and one out.

Okay.

If bunt succeeds

We expect to score 0.62 runs which is way less.

So giving up an out, and the outs are really scarce resources.

You don't want to give them up.

Giving up an out for a base is really pretty dumb.

Now this is assuming sort of an average hitter is up.

So if you've got a really bad hitter up like Lester pitching for

the Cubs so I think it's over 50 at this point, maybe you should bunt.

But I guess the proof of the pudding would be if you look at actual bunts,

what happens to these the runs created.

In other words would you expect to score less

runs given this matrix after the player tried to bunt than before?

And we'll look at that in a second.

Now you might say, sometimes you don't want to maximize expected runs scored.

So in bottom of the ninth, let's say.

High score.

You have a man on first, nobody out.

So you might say well maybe I should give up that out to get him to second because I

won't get the big inning.

But I don't need the big inning, I just need one in the front.

But it works out.

We can see later the change of winning the game does not increase with

a successful hook.

Okay, and now there is a file we can look at called Buntz where I downloaded

from the great baseball reference.com site.

All bunts from the 215 season through I guess May 20th, okay.

Through May 19th.

Okay.

And so here is the runs created by bunts.

And it minus fifth each bunt and here's the results of the bunts,

on the average created negative runs.

So it can't be a good idea.

I know there's managers who are bunting too much, maybe with a guy who can't.

If I was at bat, I should probably bunt.

because I'm never going to get a hit of a big league pitcher, okay.

But in general, bunts have created -0.15 runs per bunt attempt.

And this win probability added we'll talk about later.

The average bunt knocks down your chance of winning the game by %1.

So there's no way you can say managers or teams are making

bunting decisions correctly, because they're hurting themselves, on average.

And we'll see this, we'll see teams in football, don't go for enough on fourth

and down in short yardage, which I think you've been hearing a lot about lately.

That we can prove that's true, and we'll get to that when we talk about football.

So in the next video we'll talk about, again, baseball decision making, and

we'll try and figure out if you have a man on first, in a given situation,

what probability of success do you need before you should go to steal a base?

Go to try and steal second.

Where it's if you have, you need an %80 chance, a %55 chance,

what is the answer to that interesting question?

Coursera provides universal access to the world’s best education,
partnering with top universities and organizations to offer courses online.