THE BOOK--Playing The Percentages In Baseball

I assume that even pitching as long as he did in game 5, Lee, on 2 days of rest, will be available if needed out of the bullpen anyway.  There is a lot of precedent for moves like that when there’s no tomorrow.

What would have really been interesting is if the Phillies pulled him after 50 pitches, and then brought him back to start game 7.

They have less than a 50% chance to win games 6 and 7, so I’d like to see the calculation redone with correct numbers.

I second post #2.  Should use numbers like,
Game #6: 0.36
Game #7: Anywhere from 0.32 to 0.35

I think they had to let Cliff Lee go fairly deep (90 to 100 pitches) into this game as the empirical WE chart is probably off a little on this one due to the potent Yankee offense and the sizeable advantage the Yankees have in the bullpen.
vr, Xei

Where do you get the .36 and .32-.35 WE for the Phillies for games 6 and 7?  They have all the advantages (though Pettitte pitching on 3 days rest is a bit of ?) and home field, but ~.35 seems a bit low.

If you give them .4, .4, then it becomes:

.93*.4*.4 = .15

.99*.4*.4 = .16

.93*.4*.5 = .19

The typical star pitcher is worth +.02 wins per inning (makes him a .680 pitcher).  So, it would be better to say:

.93*.4*.44 = .16

So, it’s a wash.

I recall Josh Beckett pitching a complete game shutout of the Cubs in ‘03 for Game 5 of NLCS, then coming back Game 7 and recording four innings of relief on similar rest. Beckett went the distance throwing 115 pitches. Of course, that’s anecdotal evidence, and I don’t know the relative endurance differences between the two, but I agree with Rally that it would not surprise me to see Lee out there anyway.

Vegas has the Game #6 odds out, that’s where I got them from.  The Yankees have a 65% win probability, the Phillies 0.35 on Game #6.  On Game #7 I am taking a swag based on previous Vegas lines and my experience in knowing what the odds should be.

Game #6: 0.35
Game #7: 0.32 to 0.36 is very reasonable. 

With a somewhat rested Cliff Lee vs a somewhat rested Sabathia in Game #7, you are probably looking at around 0.39 to 0.41 at the most imo.

Game #1 was 0.3922

vr, Xei

I’m not sure I agree it’s a wash.  Whatever you think Lee’s win value is, five innings of Lee with a 6-1 lead top of the 4th has to be less than his average projected value in game 7 (assuming you also get 5 IP from him then). 

At a minimum, I certainly thought they should have pulled Lee before the top of the 8th (and yes, I thought that before he gave up 3 straight hits).  You have to win 3 straight, and preserving the chance for several innings from Lee in game 7 should have been a priority.  Especially when Lee was about to face hitters for the 4th time.

For that matter, I thought it was a mistake to allow Lee to pitch the 9th inning of game 1 with a 6-0 lead.  That was 16 pitches Lee didn’t need to throw.

Tango/5: If you use average star pitcher rather than perfect pitcher, then the Phillies’ chance of winning doesn’t go up to 99% anymore.  Probably more like 96%.

If you use that you get:
.93*.4*.4=.149
.96*.4*.4=.154
.93*.4*.44=.164

You gain 3 times as much by taking Lee out (.015 by taking out Lee, you gain .005 by leaving him).  Even if you use the 99% value, taking him out is still 50% better.  I wouldn’t call that a wash.

Sky: Fair enough.

***

“assuming you also get 5 IP from him then”

The assumption was 2 innings.

***

The main point stands that I’d like to know who would have the b-lls to make that move?  That you would justify that, on average, you gain .01 wins.  And at the same time you go from a 4% chance of blowing Game 5 to a 7% chance of blowing Game 5 if you take Lee out early.

No one here is going to do it, are they?

I think the kinds of managers that would do it would be :
A) a manager with nothing to lose.  Somebody who has nothing to lose.  If Lou Piniella thought it was a good move, I think he would make it, media be damned.
B) A LaRussa type who enjoys defying conventional wisdom and has a proven track record and job security.
C) An unproven stat-minded manager who could become a “saber-class hero” by making such a move, thereby endearing himself to stat-minded front offices and fans.

There’s actually a higher chance that you’re a hero and the move “works” and you win the World Series (16%) vs. being a goat and the move “backfiring” and you lose Game 5 (7%).  The problem though is that you get much more blame than credit.  Truly though, I’ll bet 29 out of 30 MLB managers just don’t think outside the box enough to even consider the move at all.

I would like to see the “Win Probabilities” ran through mgl’s simulator as opposed to using the “one size fits all” win exptectancy tables for the Game #5 numbers.  I think to make such a bold move, you’d need more precision.
vr, Xei

The relevant comparison points is that it’s a 16% to a 15% chance of winning the World Series against a 4% to a 7% chance of blowing Game 5.

And blowing Game 5, mind you, ends the Series right there.  While, logically, they should manage based on winning each of the next 3 games (as a set), they actually manage on not losing the game in hand (one game at a time).

There’s simply no way anyone is going to increase the chance of losing today (AND end the season), against increasing the chance of winning Game 7.

The relevant point is that if they lose, the Series ends.

When we say that managers are risk averse, it is not necessarily the criticism of the media and fans they are afraid of. They get that all the time anyway and are used to it. It is the second guessing of themselves and the self-flagellation (if the move doesn’t work) that they fear.  They are results oriented as most people are.

Especially since even if they make the right move, they don’t really know it is the right move.  They don’t have any of the numbers to defend themselves against themselves if the move backfires. They will beat themselves up.  So it is not that, “I am going to LOOK like an idiot if this backfires,” it is, “I am going to FEEL like an idiot.”

Yeah, it’s the old “if you lose Game 5, there won’t BE a Game 7” fallacy.  Very few people see past that. 

The rebuttal to that of course, is that if you’re going to lose Game 7, then it doesn’t matter what happens in Game 5.  At which point, Experienced Baseball Man will repeat, ad nauseum, “Yeah, but if you don’t win Game 5, there won’t BE a Game 7”

I agree that no one trades 4-5 innings of Lee today for 2 innings of Lee in a hypothetical game 7.  That really isn’t an interesting question.  But what about Rally’s point about starting game 7 (assume 6 IP)?  Then you’re talking about WS win going from 15% to 19%, against the increased chance of losing game 5.  That’s much more interesting. 

One assumption being made here is that wins have no value, unless you win the WS.  But is that right?  Should a manager place zero value on winning game 5?  Extending the series has financial rewards, certainly.  I imagine Phillies fans would prefer losing in 6 or 7 games over losing at home in 5.  So maybe some weight should be given to winning game 5?

I’ll continue to maintain that no one here has the b-lls to make the move.  Which is really what it comes down to it.

You decrease your chance of losing today by 3% points (and the Series ending right there, and you KNOW you will get the hit for the WHOLE 7% chance of losing, even though the chance of losing was already at 4% to begin with) against increasing your chance of winning a possible game 7 by 4% points.

Really.  There’s not a single manager in any sport at any level that’s going to make that trade.

Those are the two salient points:
- it’s “match point”
- you get the whole hit for the loss, even though there was already a non-zero chance of losing to begin with

I’m all for agreeing that the percentages agree to take him out.  But, who would do that?  No one.  And that’ll never change.

There is furthermore the additional unknown of what happens to Lee on such short rest.  The uncertainty around that estimate can easily turn the move into a bad move.

Really?  You have to assume that someday there will be a manager who knows the percentages (if there already isn’t one).  And if a manager knows and is confident that it’s the right move, do you think he will knowingly decrease his team’s chance of winning the World Series because he’s scared? 

1% of a WS title sounds small but is actually huge as far as managerial decisions go.  Especially when you consider that a Manuel snapping his fingers and winning Game 5 automatically (from the start) would have been only worth about 10% of a WS title.

I agree that he’ll need the support of his front office, but someday the GM will conference with the manager before games like these and go over these types of situations.  If they do, and the manager feels confident about the move, the manager will pull the trigger.  I’d love to see it done.

Funny though, that the conventional wisdom inside a typical non-playoff, middle-of-May type game is the opposite.  Crucial situation in a close game in the 7th inning, and who comes out of the pen?  Not the best reliever, who is saved for later, but the third best reliever, if he’s available, and if not, the 4th best.

I guess it just proves that the “win or go home” factor is the one that drives the decision.  A 7th inning in a game may or may not determine the winner, and in any event you get some more hacks to try to even the score.  A game in May may or may not determine the division title, and in any event, you get some more games to try to make up the deficit.  But down 3-1 in the WS, you win or go home, so you play for today and worry about Game 6 tomorrow…

1% of a World Series title
* * * * * * * * * * * * * * *
My guess is that the margin of error is greater than the 1%.
vr, Xei

The fact of the matter is that not having Cliff Lee available for Game 7 does not mean the Phillies lose this mythical Game 7.

As for the 93% chance of winning with a 6-1 lead in the 3rd inning. A lot of the data accumulated to determine that 93% win% is:

a) against bad teams
b) where managers continue to manage the way Charlie Manuel managed the following innings.

We really don’t know what the odds of taking the road almost never taken would be. Theoretically, it makes a lot of sense but it is very untested. And plenty of sound, untested theories have turned out to be untrue.

What if this was a practice that started to be employed and the win% went from 93 to 90% in those situations? How often does someone on this site blast a manager for giving away 1% WE? It’s very possible that you could be giving away more than 1% WE.

Then, we have Lee who had already thrown 50 pitches. Is limiting him to 50 pitches any guarantee he can return on 2 days rest and be effective? He’s probably going to be available for an inning or two if there is a Game 7, anyway since it is his bullpen day… does throwing only 50 pitches give you a 3rd inning from Lee.

You can’t win Game 7 if you don’t win Game 5… you also don’t necessarily lose Game 7 because your best pitcher isn’t available to start that game… ask the ‘01 Diamondbacks. And having your best pitcher available to start Game 7 doesn’t guarantee a win, either… ask the ‘01 Yankees.

Once you start playing for tomorrow, you can be asking yourself for a world of trouble. Manuel’s decision to yank Shane Victorino to preserve him for the rest of the series cost the Phillies their defensive sub. In any other situation, Ben Francisco would have been in LF instead of Raul Ibanez and the muffed Ibanez play is probably an out. Next thing you know, that game is 8-5 because of managing for Game 6 and beyond. Not saying it was the wrong move… just saying the results surely weren’t optimal. Especially since the concern was Shane Victorino’s difficulty with throwing the ball and Francisco later proved that Victorino could probably run the ball to home plate faster than Francisco could throw it there… OK, maybe that’s getting a little carried away!

I guess I take it the other way around.  You not only have to have a game 7, you have to win it.  You have to get (Lee innings)+(non-Lee innings) equal to 20 (assuming no extras in any game), with a lead at the end of each game. So which are the optimal Lee innings, given that you have an 8-1 lead with two innings to go against these exact Yankees? 

You have the cojones only if you have another pitcher you can go to in the eighth whom you trust.  2008 Lidge, yes.  Today, that sort of begs the question, doesn’t it?

Has anyone pitched Games 2, 5 and 7 since Mickey Lolich?

This seems like the type of call that should have been decided on before the game (If we are up X1 runs and Lee is under Y1 [X2 and Y2, etc.] pitches we pull him). . . so the numbers should have been there. 

Also, if there was any chance of a weather delay for game 7 those numbers should be figured in as well.

Tango:  I’m not clear on your exact position.  Is it that managers will never trade some of today’s WE for greater overall WE, or only that the gain in this example isn’t big enough to justify even small increase in risk of losing today?  And if the latter, what ratio would be enough?  Also, do you think this applies only to a potential deciding game, or any game?

I think we know the Phillies will make such a trade, at least in a non-deciding game.  They probably believe that Lee on 3 days rest is better than a rested Blanton, yet they started Blanton in game 4.  Totally the right move, since Blanton had to pitch either G4 or G5.  But they did reduce their chance of winning today (G4) in order to maximize their overall chance of winning Series.

I agree with those who say that taking Lee out after 3 IP probably isn’t a reasonable propostion, since we don’t know if it allows him to start G7 and bringing in middle relievers against the Yankees in the 4th has non-trivial chance of blowing up in your face.  But I would still argue managers are too cautious about removing great pitchers with a good lead.  Lee should have been pulled after the 8th in G1, and after the 7th in G5.  While it’s hard to quantify, I think that would increase the likelihood Lee could give them 2, 3 or 4 quality innings in G7.