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.
7.7 In Game Basketball Decision-Making
Loading...
来自 University of Houston System 的课程
Math behind Moneyball
40 个评分
从本节课中
Module 7
You will learn about advanced basketball concepts such as Adjusted plus minus, ESPN’s RPM, SportVu data, and NBA in game decision-making.
与讲师见面
Professor Wayne WinstonVisiting Professor
Bauer College of Business
Okay, let's talk a little bit about in game basketball strategy.
There are lots of interesting issues,
we'll just scratch the surface here, and I think we'll leave the Hackett Jordan,
which has become sort of a big debate, till the next video.
Okay, so let's go through some situations.
You've got the ball, you're down two points, and you are going to
basically take the last shot of the game, should you go for three, or not?
And, I have the stats on pro,
college I don't have the stats on how well people shoot.
If you look at this naively, and this was a Microsoft Interview question.
Let's look at the whole game.
Suppose you're going to take the last shot and you're going to let it go in the air,
the buzzer is going to go off to simplify things.
So in the whole game we know you shoot about 35% on threes,
you shoot 45% on twos.
Okay, so if you're going to replicate that shooting percentage to the end of
the game, what should you do?
Well your chance to win if you shoot the three and you make it,
you know I've seen Reggie Miller do this several times.
You have a 35% chance to win the game.
Okay, now if you shoot the two and it goes in, you tie the game,
and you think you win, the crowd is cheering, if you're at home.
But actually, you only have half a chance probably to win in overtime.
So your chance of winning would be half the chance of making that too,
which is 22 and a half percent.
Okay.
So it would seem like you should go for the three,
unless things differ at the end of the game a lot from the whole game.
And in the pros, at least, they do differ a lot.
Because I went on reference.com and checked this out.
So you're down two points at the end of the game,
what happens when teams go for three?
They make 21%, what happens when teams go for two?
They make 40%.
So if you just do some crude calculation that that's the last shot of the game,
your chance of winning is 21% if you go for the three.
If the buzzer goes off, you're thrilled.
And your chance of winning if you go for two.
40% chance of overtime.
And then let's suppose you went half the overtimes.
so your chance would be 20% and so it doesn't really make much difference.
Okay, and so that's pretty interesting.
In other words, Edwards Deming,
the great American quality control statistician said, In God we trust,.
All others need data." So I mean, in my book I did the analysis.
Sort of assuming that the stats at the end of the game would
be like the game as a whole, and that's just not going to happen.
Okay, so now, a really controversial issue.
Your team's up three.
The other team has the ball.
The clock is running out.
Should you foul them?
And I guess I had a student showing Vince I'll do some good work on this.
But again, it's hard to know what assumptions to
make about the various [INAUDIBLE] athletes.
You could foul the other team, and then they get the rebound.
I saw the Mavericks lose a game when they kicked it out.
It was Dwayne Wade, who never hits a three, hit a three.
Okay, one of the few Maverick games I ever went to was game
6 of the playoffs against the Phoenix Suns, the Mavs are up 2.
Steve Nash, a great three point shooter, is bringing the ball up.
Dirk is saying, foul him, foul him, foul him, Jason Terry doesn't foul him.
The Mavericks were up three.
Steve Nash hits the three and the Mavs lose in double-overtime and
they're eliminated.
And I mean Terry and Dirk had words after the game, I'm sure they made up, but
basically the only definitive study I've seen on this ending,
we'll see it's not really definitive, was on the [INAUDIBLE] blog,
which is great on college basketball stats, February 12th, 2013.
So if you're up three at the end of the game, should you foul or not?
So again, you can make all fancy assumptions about the chance of making
the shot, but what you should probably look at is when they foul,
how often do they win when they let him shoot the three, how often do they win.
And did the detailed analysis, so
usually teams will not foul at the end of the game, in college.
And again, the college three pointer is much easier than the pro three pointer.
Then you've fouled.
When you foul, you went 122 loss 5 and 11 went overtime.
Now going overtime counts as half a win in his analysis, and
you would win 92% of the time when you foul.
Okay.
And when you didn't foul and let them shoot,
there were a lot more cases of that.
You won 598 lost only two, and 76 went to overtime,
which counting the overtime is half the win.
Means you win 94% of the games.
Not really much of a difference, okay.
But the problem with this is, the assumption is you're assuming
the teams you found, the teams you let shoot, have equal three point ability.
Now probably what happens,
is coaches are fairly intelligent, are smart, know more than you or
I do about the team I'm playing against, because they're coaches after all.
And even if you believe in moneyball, you should never discount the knowledge
of the coaches and scouts because they know a lot more other basketball people,
or baseball people, or football people because they've been around the game and
they know a lot they should look at the data, but the data
is not going to tell you everything and here's a really good example.
Okay, so you win more when you let them shoot the three but guess what,
when you let them shot the three,
sort of the moral hazard problem if you know anything about incentives.
When you let them shoot the three it's probably because they're bad at shooting
the three so you should win.
So I mean we can't really block out the effect of the ability
of the other team shooting the three, so
I think this is really an unanswerable question at this point, okay.
Jeff Mahi you probably know if you saw the movie 21 he was the modeled for
the main character.
I probably the MIT Blackjack team that won wrote a book called the house
advantage and he's now a columnist for ESPNs gambling site.
ESPN chalk invites a lot of good posts.
So he has a great analysis of the Belichick Conundrum.
In his book I think he went through that pretty well.
But he has something I hadn't seen before.
Jeff says basically,
hey how can you tell if a team really is playing money-ball in the NBA.
You should look at the two for 1 at the end of the quarter.
So when a team has the ball, let's say 30 to 40 seconds left at the end of
the quarter, do they go for the two for 1 assuming it's not the end of the game?
So how would you tell when you should shoot the two-for-one,
although it could disrupt your offense?
Okay, and so what Jeff says in his book, he had somebody tabulate for
him, okay, if the other your team has the ball, like if you shoot the two for
one, you may be able to get the other team the ball.
You're probably going to lose on average, if the other team has the ball.
So, he looked at the amount of points you're going to lose by if the other team
just takes the ball out give the time remaining in the quarter.
And so, basically, this is what it looks like.
And so, you can see It's always negative, but
with 33 points left in the quarter,
you're going to lose by the least about 0.4 points per quarter.
So if you shoot with 33 seconds left, and you can see there's a really pronounced
sort of maximum to this curve shoot with 33 seconds left.
You lose .4 points per quarter.
And that's bad but suppose you shoot with 20 seconds left.
You lose by .7 points per quarter because you probably won't get the ball back.
So we've got through 20 seconds of that.
We'll use that .7 points a quarter, which is much worse.
I mean that can add up to [INAUDIBLE].
If you shoot with 20 seconds left instead of 33 seconds left,
.4 points per quarter over three quarters that's about,
if you had the ball at the end of each quarter, that's like a point a game.
We know a point a game is worth three wins, so this could be fairly important.
Again, you don't want to disrupt your offense in that situation.
Okay, so then a really interesting question and I'm not sure,
here's two interesting situations.
Okay, and then we'll stop and
we'll talk about hacking Jordan, based on a great post on 538.com,
which I think skewers the conventional wisdom about hacking being good.
Thank God, because none of us want to watch this.
Okay, other team has the ball, 20 seconds to go in the quarter.
Should you foul them?
Okay, and in Europe, supposedly this is a common strategy.
So if you look at the data for a whole game, this actually makes sense,
assuming they won't value back, which would be a horrible thing for
us to subject those fans to.
Okay, but if the other team has the ball with 20 seconds to go in the quarter,
okay should you foul them?
Okay, if you foul them there are about 70% foul
shooters they would get two free throws and
make 70% of them lets say, they would get 1.4 points.
Okay.
And then you get the ball back and a possession is worth about one point.
You would get one point.
Okay, so you would lose point four points.
Okay, if you fouled them, now if you let them keep the ball.
You would think if they take the last shot, okay, a point per
possession, they beat you by one point.
If they shoot like they do the rest of the game, they probably do.
So that would indicate you should foul them at the end of the quarter.
Okay, but if you look here with 20 seconds to go you can use Jeff Mazis chart
if you give them the ball with 20 seconds to go you're not going to lose by one
point a quarter, you're going to lost by about 0.6 or 0.7 points per quarter okay,
but it still looks like it might okay, and
if you get the ball back, you probably won't score that one point.
So it looks like it's going to be a wash.
In other words from this chart, okay if you would, okay with 20 points, it's
20 seconds to go and they've got the ball, you get a lose by .7 points per quarter is
what it looks like in that, since almost nobody fouls in that situation.
That's probably what's going to happen, okay.
But if you foul them, now you're the team with the ball, you
should win by let's say .7 points, and you've given them 1.4 points,
you'll lose by .7 points, it looks like it doesn't make any difference.
Thank God.
Now you [INAUDIBLE].
Okay, in college basketball, because of the one in one,
in the pros you would never do this.
Suppose it's a time game at the end of the game.
Should you foul the other team?
If you can foul you can [INAUDIBLE], it's probably a good idea.
Okay, if you foul the other team, if they have the ball, okay.
And there, Ken Pom had a post on this,
which I've sort of been unable to find with the search engine today, but
basically I think this is a pretty open topic here.
And you don't really have the data because teams almost never do this.
But you can see the argument for fowling the other team,
because they put you in the driver's seat I mean, even if they make two free throws,
you have a chance to tie the game and win the game.
And if they miss the first of the one and one, you're in the driver seat, but
of course, they could foul you back.
So I'm not sure about the of this one.
Well that's some quick analysis of some interesting in-game basketball decisions.
And they're harder to analyze than you think.
Especially when you look at the hack-a-Jordan in the next video
we're going to see.
That really requires some sophisticated analysis because a naive analysis
says [INAUDIBLE] and basically if you do a sophisticated analysis which was
done brilliantly by 538.com and Mason you should in fact unless
you believe it disrupts the momentum of the Clippers which is.
