## Why Having Cliff Lee Start Game 3 Isn’t A Big Deal

When the ALCS kicks off in Arlington on Friday night, Texas won’t have their best player on the field. Because he carried the Rangers past the Rays last night, Cliff Lee won’t start again until Game 3 of the ALCS. Because he’s not comfortable going on three days rest, he’ll only get two starts if the series goes the full seven games. The natural assumption is that this is a big problem for Texas, and gives the Yankees a huge advantage. In reality, it doesn’t matter all that much.

To advance to the World Series, the Rangers have to win four games. Lee will get no more than two starts, so the Rangers need to come out on top in at least two games started by somebody else to win the series. That would be true no matter whether Lee started Games 1+5, 2+6, or 3+7.

Psychologically, I get why 1+5 seems better. He starts two of the first five games, so you’re more likely to get that second start out of your best pitcher. But, it doesn’t change the calculus, really. They still need to win two of the five games he doesn’t start – pitching him earlier in the series just changes the dates of the games they need to win.

Let’s look at this in practical terms.

If the Rangers win three games in the series, then it’s obviously irrelevant – their best three pitchers each took the hill twice.

If the Rangers win two games in the series and lose out in Game 6 with Lee only taking the hill once, it will seem like they were significantly disadvantaged by only throwing him a single time, but the reality is that they wouldn’t have won the series no matter what. If we assume the Game 6 loss in this scenario, having Lee pitch an extra time would have only served to push back the date of the presumed loss to Game 7. Winning to earn Lee another start comes with the same requirement as winning Game 7 after Lee gets you there – a couple of wins from someone else in the rotation.

That is the requirement for the Rangers to win the ALCS, regardless of the order they pitch in. The task is the same, either way.

The other argument I’ve heard is that having Lee go in Game 1 and Game 5 would allow him to pitch in relief in Game 7. While that is true, he can still come out of the bullpen as the Game 3/Game 7 starter – he’d just do it in Game 1 on Friday night instead.

He pitched on Tuesday, and starts again on Monday. Friday is a natural throw day for him between starts, where he’d have had two days off to recover and another two days before his first ALCS start. In fact, it’s the exact same situation they would have faced when using him in relief in Game 7, where he’d be throwing with two days off after his prior start.

Whether its start in Games 1 and 5 and relieve in Game 7 or relieve in Game 1 and start Games 3 and 7, the overall effect is essentially the same. Having their rotation jumbled won’t present a significant disadvantage to winning the series. It just moves their must win games up by a few days.

Print This Post

What made Lee change his tune? He said he was ready to pitch on 3 days rest all throughout the 2009 postseason and now he can’t? Worried about his upcoming monster contract?

He did? Then why did the Phils throw Joe Blanton in game 4 down 2-1?

Because Manuel insisted that Lee get normal rest and use Blanton.

I agree with DC, but here’s a question: is the “leverage index” higher in games when the opposing SP is CC Sabathia?

This.

There would seem to be an appreciable drop-off between CC and the other Yankee starters.

Does home field advantage factor in here?

Let’s say the Yankees win the series in 5 games b/c Lee only pitched once. You would say, well, they didn’t win enough non-Lee games anyway.

But having Lee pitch in game 5 forces the Yankees to win the series on the road in Arlington.

I don’t know. I get the cold, hard logic here, but it seems a little too simplistic.

It ignores context, because context doesn’t matter in baseball…

And while the relief comparison of game 1 vs game 7 is interesting it also ignore another huge contextual issue…. If Lee come in in game 7 he’s pitching as long as he can as there is no next game (and the AL also has anextra offday prior to the WS). You bring him in relief in Game 1 and he has a clear limit if you are going to start him after 2 additional days of rest.

Of course Dave will wave this and all other arguments away as simplitic or insignificant….

Yeah dude, but, like, Game 1 isn’t Game 7, so…..

Totally agree. I feel like we’re discarding the value of Cliff Lee just giving Texas the CHANCE to play Game 7, which is obviously tremendously valuable. Yes, that game still would have to be won by somebody other than Cliff Lee, but the simple fact that they would have gotten the chance for a one-game-winner-take-all is a big deal.

Perhaps I am thinking of this wrong, but doesn’t the scenario above assume that all opposing pitchers (in this case, the Yankees) are created equal? Take for example the “lose in 6″ scenario, which seems to be the one most people are concerned about. Is it fair to assume that “having Lee pitch an extra time would have only served to push back the date of the presumed loss to game 7″?

I think the point can best be articulated by example. Suppose that Colby Lewis is projected to match up against Andy Pettitte in Game 6 and Cliff Lee is slated to match up against Phil Hughes in Game 7. Just for arguments sake (removing bullpens, defense, etc.) we can expect the Yankees to score 2 runs off of Lee and 4 off of Lewis. Let’s say that we can expect Texas to score 3 off of Pettitte and 5 off of Hughes. In this scenario, the Yanks win Game 6 (and presumably the series) 4-3 and the favorable Game 7 matchup for Texas never happens. Now, if Lee goes in 6 and Texas wins 3-2, isn’t it possible their probability of winning Game 7 (and thus the series) is altered, depending on the matchup of the pitchers?

Certainly we cannot make such accurate estimations, and my particular example may not be perfectly satisfactory. My only point is that is seems as if Lee not starting in until Game 7 COULD matter, at least in theory, unless we assume that all Yankee pitchers are created equal. I am not sure if this is a tenable assumption.

Right, you cannot say that the Rangers will score 3 off of Pettitte and 5 off of Hughes. You can say that is the median runs they will score off of them but must also realize that you will have nearly a normal (slightly skewed cuz you can’t score less than 0 runs) distribution of runs. The 3-2 and 5-4 scores each have a win probability that you can estimate with the pythagorean theorem of baseball. In one scenario (3-2, 5-4) you will have to closer to 50/50 games and in the other scenario (5-2, 3-4) you will have one large favorite game and one underdog game. In summation you have the same odds either way at chances for each team to win the series, just slightly different roads.

It probably does matter whether or not Lee pitches 3-7 rather than 1-5. The problem is that unless we do the fairly complicated matchup analysis, we simply have no idea which one will be better.

Thus, those arguing on a matchup basis aren’t even disagreeing with Dave. Dave perhaps is too hasty to suggest Lee starting game 3 “doesn’t matter,” but his analysis is very useful at showing there’s no reason ex ante to assume Lee starting game 1 is better than starting game 3. It makes sense to ignore the complicated matchup issue to get this point across.

True enough. There has been a lot of talk about this since it cuts against the typical wisdom of pitching your ace in game 1, so it makes enough sense to point out that it for the most part doesn’t matter what games you pitch based on the assumption that you pitch only 2 games.

However, it would be nice for a place like fangraphs to take the next step and at least run through an example match up calculation. I doubt it would be all that difficult really. Many of us here could probably do it in a hour, or two at most. But Dave or someone similar, who has all the tools at their finger tips, instead of you or me that would have gather than up de novo, could probably do it in 30 minutes tops.

Anyway, my gut reaction is that it probably does matter a little, and that we (fan and teams) might as well try to figure it out if we care to have this discussion.

Playing around with a crude win expectancy matrix, I’m finding there does seem to be a slight match-up advantage to starting to starting with your ace to the team with the stronger ace. The difference seems negligible under realistic circumstances – assuming an expected difference of about a half run between Lee and Sabathia (considering as well the Yankees’ superior hitting) there’s at most about a half run difference based on the order in which you use the pitchers. Within realistic ranges, I didn’t see the percentages vary by more than about 1%.

It’s entirely possible that other variations could produce different results, but from what I saw, the Rangers would probably be slightly better off with Lee in Game 1, but not by more than about a half percent, if that.

Thanks for looking at that some Wilson. That’s kind of what I might of guessed maybe .5%.

Though now that I think about it, might that be kind of a lot? If you assume both teams in an league championship series or world series are pretty good, it isn’t going to be too much more than a 50/50 that one team wins anyway, maybe at most 55/45. So an extra .5% is actually a far amount considering the kind of changes that are realistically possible just on talent alone (0-5% range above a random 50/50). Even on the upper end a team could add or take away at least 10% of the current advantage over the 50/50 mark.

Obviously, other factors come into this. You can’t just decide to start your best pitcher in game 1, because, like we see with Lee, you have to get to game 1 in the first place. And thus to actually start him in game 1 you might have to put him on short rest or reduce your chances of winning the previous series by holding him out. So, once you consider everything, its probably totally reasonable to do what the rangers are doing, but ideally, yes they should try to have him start game 1.

Anyway, I think this is an interesting topic that I would like to read more on, and unfortunately I don’t have the time to do anything myself.

this is brilliance. exactly what i was thinking. your numbers dont have to be correct for your point to shine through. here is how i see it

lee>cc>pettitte>wilson>hughes>lewis>burnett>hunter

therefore, lee wins vs anyone, and cc beats anyone not lee, and pettitte beats any other ranger starter, wilson beats hughes or burnett, while lewis only beats burnett and hunter beats no one.

ideally for texas it would be:

cc v lee, pettitte v hunter, wilson v hughes, and lewis v burnett

ideally for yankees its:

cc v wilson, pettitte v lewis, hughes v hunter and burnett v lee.

as you can all see, the second way is very close to how its going to go down.

I posted this as a comment under the chat thread, but I think it fits more here:

Assuming that Cliff Lee starts games 3 and 7, I believe the Yankees give themselves the best chance to win the ALCS by throwing CC, Pettitte, Burnett, CC, Hughes, Pettitte, CC. Essentially, they would be conceding game 3 to the Rangers, but setting themselves up with the best chance to win the remaining games. The SP matchups would favor the Yankees in all the games except 3 and 7. Meanwhile should the series go 7 games, instead of being clearly overmatched by throwing Phil Hughes against Lee, they would be able to counter with CC and make it more even. I am under the assumption that CC’s performance does not drop off much if at all when pitching on short rest (this is based on observation and may not actually be the case).

I believe this follows logically from the you have to win 4 games line of thinking that Dave uses in this post. Essentially it doesn’t matter if you give yourself a lower probability of winning two games if it is outweighed by the increased probability of winning the remaining five.

I agree with the underlying assumption. CC is the one guy in baseball who, if anything, improves under the stress of three games. By shooting a blank with AJ, if surrenders a game, but if they’re going to do that, and I don’t think they should, then you certainly lose the least by lining him up against Lee. That said, AJ, and this would truly be an upset, but AJ does have the stuff to almost match Lee if he ever figures himself out. I don’t think Hughes or Andy can do that, though both impressed earlier.

I think the biggest problem here is the matchups. If lee pitches 1 and 5 he gets to pitch against CC. Assuming those are wins (the probability goes down that he will win against CC), the team has to win 2 non-Lee starts against Pettitte, Hughes and Burnett.

Since Lee pitches 3 and 7 he will be facing Hughes/Burnett, leaving his team to pick up 2 games against CC, Pettitte, and Burnett/Hughes. Obviously the Non-Lee Rangers have a much better chance of winning 2 against starters not named Sabathia.

And then the Rangers would have a worse chance of winning two games on the days that Lee does pitch. You can’t assume that the team’s chances of winning are the same with Lee vs Sabathia as they would be with Lee vs anyone else.

True. There is obviously a tradeoff to be made. Do you want your ace when he’s most likely to win, or when his team is most likely to lose– due to the tough opposing starter?

Another factor that must be considered though is how likely the Rangers are to win based on who they’d be matched up against if they wait to start Lee. If they’re confident they can get after the Game 3 starter, then you could argue it would makes sense for them to instead start Lee against the toughest opposing starter.

There is some more complex game theory at play here that none of us are comprehending. Your analysis is right on the surface, but there is a difference between the scenarios where Lee pitches vs CC and where he doesn’t.

Dave,

” You can’t assume that the team’s chances of winning are the same with Lee vs Sabathia as they would be with Lee vs anyone else.”

Of course not, but we need to take into account the marginal win probability changes in every case and run through the numbers. I suspect that Lee pitchign 1 and 5 is going to come out as the better choice, though not by much. If nothing else, it would allow for Lee to pitch 2-3 innings a do or die situation in game 7. And presumably they wouldn’t do that in game 1 if he starts games 3 and 7.

As Dave says, it is a big reach to assume anything other than that the Yankees and Rangers would split 2 CC vs. Lee games.

I don’t think several of the people posting understand the logic of what you said. But as someone who does… this is what I’ve been trying to explain to people haha.

If Cliff Lee were the game 1 starter, the Rangers would theoretically need to win 0 games not started by Cliff Lee in order to get him back in the hill for a game 5.

If Cliff Lee were the game 2 starter, the Rangers would need to win as few as 1 games not started by Cliff Lee to get him back on the hill for a game 6 start.

If Cliff Lee is your game 3 starter, you need to win at least 2 non Lee starts in order to get him on the hill for a game 7.

I understand the point that to get to the WS, Rangers pitchers not named Cliff Lee need to win 2 games.

The Rangers best chance at victory comes with him on the mound, and starting him in game 3 means that 5 games will have to be started by pitchers not named Cliff Lee before he would get(need?) a 2nd crack at the Yankees. Again, other pitchers still need to do their job to get to the series, but the Rangers win expectancy for games started by Lee is certainly higher than non-Lee starts. It must also follow that the Yankees best opportunity to accumulate the 4 victories they will need is against not Cliff Lee. Giving the Yankees 5 cracks at not Cliff Lee before they have to face him a 2nd time is certainly more likley to result in 4 Yankee wins over that stretch than facing him twice.

If it goes 7, its irrelevant and the Rangers have their best weapon on the mound. Any other outcome and it most certainly mattered.

The goal is not to get Cliff Lee on the mound twice. The goal is to win four games.

… Doesn’t mean they shouldn’t be motivated to get him to the mound twice, since that would obviously make their chances of winning four games that much better.

Dave … seriously. If this site really believes that Cliff Lee is one of the top 5 pitchers in the AL, then “winning four games” and “getting Lee on the mound twice” should not be mutually exclusive goals, but rather inter-dependent situations.

I don’t think it’s reasonable analysis to expect the non-Lee’s to win 3 out of 4 games against the Yankees, rather than expect Lee to win 2 or 3 games in the series. If you don’t need Lee to win 2 or 3 games in the series, why acquire him?

Meanwhile CC will start three games this series, if he needs to.

The Rangers need some breaks in this series. For them to compound that with only 2 starts from their ace, while NYY gets 3 … that’s just another disadvantage in a 7-game series.

I understand that, but the best way to win games is to not lose. Your best bet for not losing is starting Cliff Lee. Firing all of your not Cliff Lee bullets in 6 games reuslts in fewer wins than firing 2 Cliff Lee bullets over a 6 game stretch.

The series can be won in 4, 5, 6 or 7 games. Pitching Lee in 2 of the first 5(or 6 if he were the game 2 starter) games gives you a better chance at winning in 5 or 6 than only pitching him in 1 of those first 5 or 6 games. If it goes 7, it’s irrelevant(and likely advantageous).

There might be some more subtle impact on the overall probability depending on how optimal the ace-versus-ace approach is, or whether there’s a difference in how well the parks are suited to the different pitchers, but I agree with Dave’s analysis here.

If Lee doesn’t get to pitch game 7, it will mean one of two things:

1. That the Rangers have managed to win four games already, or

2. That the Rangers have managed to lose four already.

If Lee could go in games 1 and 5, then sure they’d have a better chance of losing the series in 6 rather than 5, and they’d have a better chance of winning in under 7, but on the flip side, if they fail to win all non-Lee games before Lee’s game five start, they’d be forced to win two in a row without their ace after he pitches for the second time. It might be different if Lee on short rest was an option, but they’ve already declared that it isn’t.

If the margin of loss or the number of games played mattered, then there would be some noticeable advantage in which games he starts. But their chance of winning four absolutely requires a minimum of two non-Lee wins out of five chances either before or after his second game, and if they get those two wins, they’ll guarantee him a second start unless they’ve already won the series.

If they lose in 5 or 6 it doesn’t matter because they already didn’t win the requisite 2 non-Lee starts.

Even if we assume that the Rangers will win every game started by Lee, which is obviously not a certainty and is especially not a given if he faces Sabathia – the Rangers will still have to win 2 games started by someone other than Lee. What difference does it make whether those games started by the other Rangers are #’s 1, 2, 4, 5 and 6 or #’s 2, 3, 4, 6 and 7?

If the Yanks are smart, they’d pitch Hughes game 2, and save Petitte for Game 3, in New York. You want the lefty on the hill at home.

they are doing this

That’s what they’re doing. CC/Hughes/Andy/AJ/CC/Hughes/Andy. Obvious matchup advantages to the Rangers in SP each time Lee and Pettitte face off, but Pettitte’s exactly the type of pitcher to neutralize NYS’s offensive advantages.

I think people are selling the rest of the Rangers’ rotation short. Their #2 and #3 both posted FIPs comparable to Sabathia’s, while Lee blew everyone else out of the water.

because FIP is the end-all and be-all of pitching, right?

Because that’s what I said, right?

If you want to go with ERA, the Rangers are putting out four ERAs below 3.80. Yankees are putting out two.

I you want to be blinded jerk, just keep those fingers in your ears, buddy.

I don’t actually believe he is your buddy, pal

Lee wants… more munnay

“he’s not comfortable going on three days’ rest” Nobody is. Anyone who says otherwise needs to check his machismo and bravado.

Jim Bunning says he could go on 3 days rest. Right now.

tell that christy mathewson

Old Hoss Radbourn wasn’t comfortable going on 3 days’ rest! How could anyone bear to wait that long between starts?

See Young, Cy

To the people disagreeing/not understanding dave:

Think of it this way. Pretend that the 2 cliff lee starts are automatic wins for Texas, no matter what happens. Whether those games are 1 and 5 or 3 and 7 doesn’t matter. They need to win two other games.

Let’s say they lose in 6 so they didn’t get to a game 7. Well, in your head you can pretend the rangers won game 7 with lee throwing a shutout, and the rangers still lose the series 4-3 instead of 4-2. He’s pitching two games, and that’s all that matters.

…Thanks Dirk, think you really cleared that one up for us…

?

That’s awfully pretentious. There are plenty of people on here who disagree with Dave and understand his arguments.

They are just saying there’s levels that Dave doesn’t address (i.e. does the team benefit more having CL face CC versus Hughes/Burnett?)

The arguments referencing matchups can be valid. I’m just talking about the arguments of getting Lee a second start being valuable in itself. Didn’t intend to say that the pitching matchups are necessarily irrelevant.

If Cliff Lee’s starts are automatic wins, you want him taking those automatic wins as soon as possible, because 4 losses sends you packing. If you’d rather get 1 automatic win in the first 6 games, and take your chances with 5 other games, instead of 2 automatic wins in the first 6 games, feel free, but the rational thinkers will stack our best opportunities for victory as close to the beginning as possible, so as to limit our opportunities for loss. Even in a world where Cliff Lee isn’t an automatic W(like ours for example), if he gives you the best opportunity for victory, you want him starting first. 4 wins moves you forward, 4 losses sends you home. If pitchers could only make 2 starts in a series, and rest was irrelevant, would you choose Lee to pitch games 6 and 7? Why or why not?

In this scenario, I wouldn’t care which two games he started (again, I’m not denying that things like matchups could matter and in real life there could be an emotional edge of being up 2-0 instead of down 0-2).

If rest is irrelevant, and Lee starts are auto-wins, then the Rangers are basically just playing 5 games, and need to win 2 of them, since they already have two in the bag. Whether those two in the bag come before or after the other 5 games irrelevant, you know you have those two games won.

Dirk you are incorrect. You would take the first two games. Think of it this way. As you said you only need to win 2 of five games. But if you reserve your bullets for the 6 and 7th games, then you cannot afford to lose 4 games, and your opponent will have all 5 chances to win them. If you take the first two games, the scenario of having to win 2 of five games remains, but you can simply win games 3 and, precluding any chance your opponent has. You can knock your opponent’s probability of winning to zero in just four games, instead of having to wait out your foe and giving them every opportunity to win they can have if two games are already yours. You want to crush your opponent, give them as little possibility for victory as possible. In this case, you you take your automatic wins as soon as they’ll come.

Think of it this way, you may never get to game 6 or 7, but you have to play games 1&2 no matter what. Given these two option, I’ll take the sure wins in games 1&2 every time.

No I’m not incorrect. Either way my opponent needs to win 4 out of 5 games to beat me.

Not considering matchups, bullpen use, psychological factors etc., Dirk’s right, it doesn’t matter.

If you save your big guns for the end, you would need to win two of five games without the luxury of your ace, and then trust your ace to finish it off.

If you use your big guns early, you would need to win two of five games without the luxury of your ace, and then trust your ace to finish it off.

No,

Your opponents task is the same, but his chances of defeating you are not as constant. Would you chose to instantly win games 6&7 or games 1&2? Because how you answer that question will, statistically speaking, change the dynamics of the outcome.

Winning games are mutually exclusive events. Winning game 1 and 5 is not going to increase the chances of you winning the other games relative to winning game 3 and 7. That’s all there is to it.

“Your opponents task is the same, but his chances of defeating you are not as constant. Would you chose to instantly win games 6&7 or games 1&2? Because how you answer that question will, statistically speaking, change the dynamics of the outcome.”

Would you care to walk us through those statistics?

“Winning games are mutually exclusive events. ”

no, they are absolutely not.

What about Lee’s impact on the bullpen? If he pitches 8 innings in game six, no matter what’s going on, the Rangers have all of their bullets rested and ready to go for game 7.

A stud pitcher has impacts on more games than the one he pitches in. If he’s pushed back to game 7, than his innings eating skill provides no help to the team..

If Lee pitches deep into game 3, then the bullpen will have had Sunday and Monday off, so they should be rested for games 4 and 5. They then get another day off. Not to mention CJ Wilson would be available for game 7 out of the bullpen. I don’t think bullpen readiness is a big deal in this case.

The innings eating skill helps the team by allowing it to treat the pen in games 5 and 6 the way you suggest they would for game 7. There’s a certain amount of risk there but it’s important to remember that the Rangers can always bring back Wilson, their #2 starter (and game 1 starter), in relief behind Lee in game 7 if needed. Therefore, Washington can fire all his pen’s bullets in game 6.

I think if you’re the Rangers you might actually like your chances of winning a little bit more against Pettitte than Hughes. Pettitte had the roundly superior season, however, especially since Josh Hamilton appears to be playing at less than 100% all of the Rangers’ best hitters right now are righties who have hit lefties at a better clip than same-handed pitching for their careers.

If the plan is for Pettitte to start Game 2, then especially since it would be in Texas, I think in that instance it would make sense to save Lee for Game 3. The Rangers should like their chances at home against a LH starter, and there is certainly something to be said for having your best pitcher increase your chances of winning on the road.

i think the analysis is solid in terms of probabilities, etc. but it assumes that each game is an entirely independent probability from the preceding games, like coin flips or throws of the dice.

we’re dealing, however, with human beings and not coin flips. i’m not sure, as a matter of psychology, that each game can be considered a statistically independent event.

if the rangers find themselves in a hole early (down 2-0, for example), isn’t there concern that the hitters might start pressing, lee might put too much pressure on himself and overthrow, etc.?

Wait cliff lee is pitching against the Yankees???? I was certain I read somewhere the Twins were awesome and their depth would knock the Yankees out in the first round!!!!

Yeah Dave was wrong about something he predicted. It’s pretty easy to pile on him for that, in addition to being really trite and tired. I’d love to see your predictions for all the series to pan out.

I’m not an apologist, but I get tired of seeing all this DC hate just because he predicted something wrong.

Good grief.

Assuming Lee wins his starts …

[1] If he wins game one, he’s guaranteed to pitch game 5. The NYY cannot win a 7 game series by winning games 2, 3, and 4.

[2] If Lee wins game 3, he is not guaranteed a game 7 start. The NYY could win the series by winning games 2, 3, 4, and 6.

So, if the series goes 7 games, then yes, it does not matter if Lee wins games 1 and 5, or 3 and 7.

But, by him pitching games 1 and 5, you have a much greater probability of him pitching two games, versus just one by starting games 3 and 7.

IMO, the problem with this sites playoff analysis is [1] the trend to just go contrary with the opinion, just to be unique (it’s annoying, seriously), and [2] the dwelling of seaslong stats against all teams instead of focusing on more relevant stats against strong teams.

It’s like FG is surprised when the more talented team wins. Like, duh.

Talent thrives in the playoffs. Lee is the best pitching talent for Texas. Thus, it is a big deal for him to pitch two games instead of three …. it is also a big deal for him to pitch 1 and 5 instead of 3 and 7.

If you were managing the Rangers, and the league said “Lee cannot start games 1 and 5, he can only start games 3 and 7″, you would be absolutely pissed … and for good reason, it’s a disadvantage.

Even if Lee started and won, games 1, 4, and 7 … the Rangers would still need to win ONE non-Lee game. Okay, great. The odds of winning one non-Lee game are better than trying to win two non-Lee games.

Folks, there are ONLY seven games in a series. Any single game where you do not have your best pitcher on the mound, when you could, is a big deal.

+1

+1

err, NYY could win the series by winning ANY FOUR of games 1, 2, 4, 5, 6, and not face Lee in game seven.

There is also psychological aspects in being up/down in a series … ‘specially with a team that isn’t in the playoffs a lot.

The psychological impact of anything seems to be discounted, but I think you’re right to bring it up. The series feels differently to each side depending on if they’re leading or behind in the series, particularly when one team gets to within one game of winning. More importantly, the series *plays* differently as well. Each game in the series is not played in a contextless vaccuum. Strategies change depending on the situation. Teams do things like press when they’re pressured, or take/avoid risks if they’re pressured, and none of this seems to be factoring into the thinking re: Lee.

The counterintuitive argument that the Rangers aren’t hurt by not having Lee available in game 1 is unconvincing. Teams will not play the same way if they’re up 3-1 vs. down 3-1 in the series, and players won’t react the same way, either. So saying that it wouldn’t matter if Lee never gets to pitch a game 7, because if the Rangers don’t get that far they would’ve lost anyway, is spurious logic. Do not buy.

I don’t doubt that psychological effects can play a role, but we shouldn’t be so quick to presume what those effects could be.

Does being down in a series demoralize a team, or does it add a sense of urgency that makes the players sharper? How does the knowledge of having an ace at the end compare with the knowledge of having to close out the series without your ace if he’s used up early?

Ballplayers are people and psychological effects can certainly affect their performance, but different people respond to different situations differently. Some people might be motivated by an early lead and play with more confidence while others might become lax by the comfort of an advantage. Some people might press when they’re down, while others might perform best when they’re under pressure.

If there are psychological factors in play here – and there could be – can we really say with any confidence how those factors are likely to play out?

Wilson, we could probably attempt to answer this question by looking at past 7 game series. Do teams facing a 0-2 hole go on to win more, less or the same number of games that we would expect?

@ Wally: Even in that kind of test, you (realistically) wouldn’t be able to isolate out psychological effects from the ‘the winning team was simply better’ effect. Teams that get a lead in a series tend to win them – is it because their confidence and momentum carried them to victory or because they were the better team anyway?

Good vibes is something that is intuitive, but prohibitively difficult to prove. Micro behavior isn’t very predictable.

blaff,

I’m sure from regular season data, current line up construction, and even future data we could get some pretty good guesses at who’s actually better. Especially if all we look at is league championship series and then don’t have to deal with league effects. Even if we just went back to 1995 or something, we’d still have at least a sample size of 30. Could be 90 if you want to use the division series and 105 with the WS. With that kind of sample, I think true talent estimates are going to come out just fine.

I’d even go on to bet that there would be little point to doing all this and if you basically tested against a null of a 50/50 chance of each additional win, that you’d arrive at the same conclussion.

Psychology aside, it makes zero difference to the probability of winning which two games he pitches.

You hate to lose the series having him only pitched one game, but if you do that it means you’ve lost four of the five games that he didn’t start. If you lose four of the five games he doesn’t start, it doesn’t really matter whether that’s 1,2,4, and 5 or 3,4,6, and 7, unless you consider it better to lose 4-3 than 4-1. If he were willing to go on short rest and he were willing to go 1/4 and have the possibility of coming back for even 75 pitches in a possible game 7, that would be different.

Er, how can you say ‘psychology aside’? It can’t be removed from the equation.

Because there is simply no evidence out there that shows what the effect of psychology will be. Maybe they play better, maybe they play worse, maybe it won’t make a difference. So you can’t simply use “psychology” to argue the Rangers are better off with Lee pitching on Game 1, because we simply don’t know if it’ll have that effect.

@ q: Even if you can’t definitively say whether or not being ahead or behind in a series hurts a team, is it really safe to assume that it’s a wash? Psychology clearly has some impact, so I’d err on the side of it mattering in a good way. Better than whistling past the graveyard and assuming it can’t hurt you.

Yes, but how do you propose we factor psychology into the equation without making wild guesses about the mental makeup of the players and delving into discussing “intangibles”? This is a serious question. If you have an answer for it I would like to hear it. If you don’t, that’s fine as well because neither do I and I don’t think anyone does.

@ Terminator X: Any number you ascribe to good vibes in an equation is going to be subjective. Simple as. But that’s not a reason to ignore reality. Teams that get off to a good start in a series tend to win them, don’t they? (I may be making the wrong assumption, but I’m almost sure that’s true.) Are we so confident that none of that is due to psychology that we simply ignore psychology? I’d rather err on the side that getting a lead matters in a good way.

I don’t assume it’s a wash. I’m saying our analysis won’t change if we factor in psychology, because we simply lack the ability to do that. Also, why would you err on the side of it mattering in a good way when we simply do not know how it will matter? Why not err on the side of it mattering in a bad way (i.e. teams play relatively better when they’re down)? Better than whistling past the graveyard and assuming it can’t help you.

@ bflaff – teams that get off to a good start in a series do win series more often, but I would generally ascribe that to having more wins as a result of that good start. We would expect a team up 2-0 to win the series, because the odds of the other team winning 4 out of the next 5 are low. Also, if a team wins the first two games the odds are probably slightly in favor of that team just being a better team than the other one.

I blame you for the box of toothpicks having instructions on it.

It’s also a simple matter of probability and resource management. Work it out for yourself, assume all starts happen as promised and assume both teams are evenly matched. Ignore pitching matchups. If BP has already done their playoff Monte Carlo simulator then use those win percentage projections but for the purposes of this exercise I’m happy with a 50% win chance for either team for each game.

We’re measuring probability each team gets to start their ace. There’s a 100% chance that games 1-4 happen, so one unit of (potential) awesome happens for each team. What is the probability either teams sweeps? 6.25%. So allowing for both teams’ chances, there is a 87.5% chance Game 5 happens. CC’s “start value” is 1.875 then. Extrapolating to Game 7, there is a 31.25% that we even reach that game. This Cliff Lee’s “start value” is 1.6875.

Sabathia’s 1.875 vs/ Lee’s 1.6875, not as much divergence as some might guess, but some marginal advantage. Somewhere between an advantage and “not a big deal”.

Dave it may get lost in the flurry, but can you address Lee’s impact on the bullpen? If he pitches 8 innings or complete games in his two starts, that’s a huge help to the bullpen for the other games. You lose that advantage if his second start is game 7.

Why are people trying to apply dependency and exclusive events into a basic logical point Dave brings up. I don’t understand =(

…because those people would argue that they have everything to do with whether Dave’s argument is actually valid. I think everyone understands his logic. Some people just don’t completely agree with the conclusion he reaches.

Oh ok. I think the Phils should start Halladay in Game 3 too, and then the Giants will counter with Lincecum in Game 3. Etc. We will all end up sitting through a game 1 of Bumgarner vs Blanton. That does sound exciting as I say, rolls right off the tongue…

Max,

No one is saying that Texas gets an advantage from starting Lee in Game 3. If Lee is not going to start on short rest, Game 3 is the earliest that he can pitch and Dave is stating his opinion that this is not a disadvantage to Texas.

I generally agree with Dave with the exception that if Lee pitches in Games 1 and 5, he might be available for an inning or 2 in an “all hands on deck” relief role in Game 7.

That’s the first problem, but we can’t do anything about that.

True again. But as I said, by starting games 3 & 7, instead of 1 & 5, Lee and the Rangers give the NYY the opp to win the ALCS while only facing Lee once. I would not be surprised to see the NYY do just that.

While it is true, that in a 7-game series, with Lee pitching only twice, you force the non-Lees to win 2 games. Fair enough. With Lee pitching game 3 and 7, you are forcing the non-Lees to win two games out of four, instead of 2 games out of 6 … against (perhaps) the best team in baseball.

I’m sure you can figure the probabilities of …

[1] Non-Lees winning 2 of 4 vs. NYY

[2] Non-Lees winning 2 of 6 vs. NYY

It’s like saying 2004 was no big deal, since BOS had to win 4 games anyway … big deal if it was the last 4. What was the probability of BOS winning 4 out of 7 vs. winning 4 out of 4?

I honestly don’t know what to say. While the math may be the same in terms of innings pitched, games appeared, etc … having Lee pitch (1) in relief in game 1 and start 3 and 7 is completely different than (2) starting game 1 and 5 and pitch in relief in game 7.

Also, it’s irrelevant. Pitching in relief in a short series is for guys that are willing to pitch on short rest. That’s not Lee.

“While it is true, that in a 7-game series, with Lee pitching only twice, you force the non-Lees to win 2 games. Fair enough. With Lee pitching game 3 and 7, you are forcing the non-Lees to win two games out of four, instead of 2 games out of 6 … against (perhaps) the best team in baseball.”

In either case, non-Lees have to win 2 out of 5 games. Not sure how you got 4 and 6.

Lee throws 3 and 7: Non-Lees need to win 2 of games 1,2,4,5,6

Lee throws 1 and 5: Non-Lees need to win 2 of games 2,3,4,6,7

I think the unstated idea is that it’s easier (not easy – just easier, in a meaningful way) for the Rangers to win 6 and 7 if Lee pitches 1 and 5. Of course, it depends on how the Rangers do in those first 5 games, but in terms of strategy life is probably easier for the Rangers if Lee can go 1 and 5. The strategy for later in the series will be dependent on what happens earlier in the series, so the better you do earlier in the series, the more strategic options you have later. Teams already down 3-1 have to take risks that a team tied at 2-2, or up 3-1 don’t have to consider.

I would think the difference is minuscule at best. In a short-series, teams already have a very short time horizon. Yes, one’s time horizon is shorter if you’re down 0-1 versus 1-0, but I doubt the difference is significant enough that you’ll see in a game or two a situation that calls for a different decision.

Dirk your scenario is very good… considering the Yankee pitcher in those games which would you prefer if you were the Rangers?

Would you rather have CJ Wilson go up gainst Sabathia twice vs Pettitte (or Hughes) twice?

If Lee = autowin, don;t you want him taking down the other team’s best pitcher to give your other pitchers a better chance at more run support?

To put it in simpler terms, if pitching match ups are a game of war, then the Rangers have the ace and won’t be able to use it at ideally against CC (a king by relative terms). Yes Dave, it still presents the problem of lowering their chances of winning that particular match up vs him starting against someone like AJ, but you need to make more than the superficial statement you just presented. You know, maybe try calculating the series outcomes based on the match ups if he were starting 1&5 vs 2&6 and 3&7. This is a statistical site, pretend like you actually use real statistics.

Just to use the war example further. If you have an Ace and two Kings matching against two Aces and a King and you flip a coin in the case of a tie. Ideally you lose 1 game out right and flip coins for the other games vs losing out right to unfavorable match ups. Obviously baseball games involve a lot more elements, but to brush aside the importance of match ups is a serious oversight.

Just to nitpick the analogy a bit, each ’round’ of War is independent of the rounds that come before and after. The playoffs aren’t like that. Each game in a best of 7 series is not fully independent of the games that come before and after. Your strategy is influenced by how you’re doing in the series, and so are your players.

Not in the sense that a rotation is typically set before a series starts. It is more or less forced unless you decide mid series that you are going to start pitchers on short rest. Neither team has said that so at this point we can assume we know the pitching match ups for the whole series.

Sure, but a team’s strategy in a series goes far beyond naming starters, doesn’t it?

I think everyone would agree the starters comprise 90% of the variable from game to game barring someone getting hurt. Sure, maybe they bench a guy or shuffle a couple spots in the line-up but really the only major changing factor from game to game is the starting pitcher.

Each round of war is not independent, once you’ve used that ace or king, it can’t come back. Further choices and outcomes early cause differences in later choices. It is similarly faulty to just assume each game in a seven game series is an independent event. They are most certainly not. For simplicity you might want to assume that, but you then have to rationalize or even quantify that assumption. I haven’t really seen anyone try to do that yet.

I think the horse has taken enough beating, but here are some scenarios to really simplify Dave’s logic:

Lee wins 1/5:

Rangers lose all non-Lee: lose series in 6

Rangers win 1 non-Lee: lose series in 7

Rangers win exactly 2 non-Lee: win series

Lee wins 3/7:

Rangers lose all non-Lee: lose series in 5

Rangers win 1 non-Lee: lose series in 6

Rangers win exactly 2 non-Lee: win series in 7

I guess if Lee started 1/5 and the Rangers can win 2 non-Lee games, you can prevent the series from going to 7. Also, I personally believe that winning game 1 can give your team a psychological advantage. But based on pure logic, the 2 situations are exactly the same.

Everyone is overly simplifying the importance of match ups. Obviously you have to win at least 2 of the games Lee doesn’t pitch, but what you’re missing is the chances of winning those 2 games or even the 2 games Lee does pitch is greatly affected by the pitcher he faces. You would really need to simulate the series from different perspectives to see which has the greatest chance of the Rangers progressing to assert that it does or doesn’t matter which games he starts.

Unfortunately, that’s simply too difficult and until we do that, it’s impossible to say if the matchups would favor Lee pitching Game 1 rather than Game 3. Thus, it’s certainly valid to ignore match ups when countering the argument that the Rangers are at a disadvantage because Lee is pitching Game 3 rather than Game 1. Ceteris paribus, it doesn’t matter.

q: Really? That’s as ridiculous a statement as I’ve heard. You are saying with all of the statistical and simulation power we have today that you couldn’t attempt to gain any advantage by setting a rotation a certain way? Nonsense.

@q. Discounting something because you can’t account for it is a terrible mistake. Scientifically speaking, you can make incorrect assumptions as you have but of course that just makes your results irrelevant.

@Nick:

Wait, what about the KISS principle? (head explodes) :)

I’m almost 100% sure you are incorrect Dave. I haven’t run the full binomial, I hope someone has the spreadsheet handy, but here’s my logic.

The issue is path dependence. What happens in games 1-5 affects the number of opportunities the Rangers get to win the 4 games they need. They may get as few as 4 chances or as many as 7.

To simplify, let’s assume the Rangers have a 60% chance to win Lee starts and a 40% chance to win non-Lee starts.

Your analysis would be correct if you assume that the series must go 7 games. In that case the Rangers play 2 games with a 60% chance of winning and 5 with a 40% chance, needing to win 4 of 7 So when Lee pitches doesn’t matter.

Likewise, if the series goes 4 games, the Rangers face 60% odds once, and 40% odds three times, needing to win all 4, regardless if Lee pitches Game 1 or 3, he can only pitch once.

However, the Series can also end in 5 or 6 games. In those cases, the Rangers face radically different odds depending on when Lee pitches.

In a 5 or 6 game series, they only have 60% odds once with Lee in games 3 and 7, as opposed to twice with Lee in games 1 and 5.

“Likewise, if the series goes 4 games, the Rangers face 60% odds once, and 40% odds three times, needing to win all 4, regardless if Lee pitches Game 1 or 3, he can only pitch once.

However, the Series can also end in 5 or 6 games. In those cases, the Rangers face radically different odds depending on when Lee pitches.”

This doesn’t make sense. It’s not just randomly determined how many games a series will go.

Yes, not having Lee pitch again until game 7 makes it more likely that the Rangers will lose in 5 or 6 games, but if it goes 7, the rangers have a much higher chance of winning the series.

Somebody really needs to just do the math, because it will even out exactly.

It should even out if I’m not mistaken.

As a simple example, the chances of the Rangers being down 2-3 after five games should be the same if Lee pitches in game two or game three – either way you’d have one Lee game and four non-Lee games played.

From that point on, if Lee starts the sixth game, the Rangers chances would be as follows:

0% – Win in 6

40% – Lose in 6

24% – Win in 7 (.6*.4)

36% – Lose in 7 (.6*.6)

If Lee starts the seventh game, the odds would shift as follows:

0% – Win in 6

60% – Lose in 6

24% – Win in 7 (*4*.6)

16% – Lose in 7 (.4*.4)

In either scenario, the Rangers would have a 24% chance to come back and winning, and a 76% chance of losing. The order greatly affects the distribution of the number of games, but the ultimate outcome of winning or losing shouldn’t shift.

I haven’t run all the numbers or anything, but I’d expect the logic to follow as you expand. If Lee pitches earlier, he increases the chances that the series goes to 6 or 7 games, but that’s offset by the decreased likelihood of winning that game 6 or 7.

I did the numbers for a 5 game series where it’s between Lee pitching 1-4 and 2-5.

Probability of the Rangers winning in 5 games when Lee goes 1-4: 0.1024

Probability of winning in 5 games when he goes 2-5: 0.2016.

Probability of winning in 4 games when Lee goes 1-4: 0.2112

Probability of winning in 4 games when Lee goes 2-5: 0.1408

Probability of winning in 3 games: 0.096

Now of course, you can’t simply say Lee should start 2-5 since those probabilities add up to a higher number, because (P(3) or P(4) or P(5)) isn’t a simple sum as these are not mutually exclusive (yes they sound mutually exclusive, but the actual events that occur to get to those situations can overlap). I imagine once you factor that in, it evens out.

Of course, one simple way to do that is to simply map all permutations of W-L in 5 games, then add up the probability of those that have 3 or more wins. But since winning games are mutually exclusive, it’s pretty easy to intuit that it makes no difference whether Lee pitches 1-4 or 2-5 if you do it this way.

I ran the numbers for a 7 game series.

Lee 1, 5:

W4: 0.0384

W5: 0.11904

W6: 0.1216

W7: 0.12595

Probability of winning the series: 0.40499

Lee 2, 6:

W4: 0.0384

W5: 0.07936

W6: 0.16128

W7: 0.12595

Probability of winning the series: 0.40499

Lee 3, 7:

W4: 0.0384

W5: 0.07936

W6: 0.10752

W7: 0.17971

Probability of winning the series: 0.40499

No change in the probabilities based on when he starts. No doubt matchups play a part, but it’s not clear how the matchups would change things overall.

q: I think there might be something wrong with your calculations when the probability of winning a best-of-7-game series in 3 games is greater than 0.

Don’t know the math, so no help there, but I do think that ‘path dependence’ logic is at the heart of the con position.

Yep, and unfortunately its pretty damn complicated issue that would will probably leave us with a lot of assumptions. So, I’m sure its real, but its probably small and we probably don’t have much chance at really figuring it out.

Interesting. I will think about this.

I think even more interesting then this debate, is discussing the merits of CC pitching three games or two. I think this is itself worthy of a post. I personally think this is the best strategy for beating the Rangers. AJ Burnett doesn’t belong in important games, and CC is the guy known for getting it done on three starts. Doc might be the only pitcher I would take over CC on three days, I’m not sure that’s clear cut. I think this means the yankees need to do what it takes to take an AJ start and make it a CC start.

The only way the Yankees can avoid starting AJ is by pitching CC, Pettitte and Hughes all on 3 days rest in games 4-7.

Or pack it and crack it. Start AJ against Lee in Game 3, and play to win in 1,2, and 4.

AJ is the kind of pitcher that you just might catch lightning in a bottle with too. Plus, you’ll be at least be at home in game 3 and coming off an off day, so your bullpen will be rested…

I see how I can vote down commenters.. how do I vote down an article?

Another point, we’re conceding that Lee wins game 3 because he’ll be the better pitcher. Then we should also concede that CC wins games 1 and 4. Game 7 would need to be considered equal, as the differences in lineps probably balances out and edge Lee has over CC.

So, to get to that game, TEX needs to win every non CC game … And then eventually beat CC in g7.

I think TEX is starting off in a bad position, and could very well be down 0-2 before Lee takes the hill. I think the probability is high that Lee only gets one start in the series. To get to g7 TEX is going to need quite a bit of (literally) luck.

I still don’t get why Lee won’t pitch on short rest. That seems so counter to what aces do in a 7- game series, where you’re only guaranteed 4 games.

If it does go 7, however, it should be one of the best pitching matchups in a big game that we’ve seen in recent history. Neither team will be trying to piece together a game 7 plan. It’ll be horse against horse.

I’m not sure what you are saying here. Sabathia is not going to start three games in this series. The order as of now appears to be Sabathia, Hughes, Pettite, Burnett, Sabathia, Hughes, Pettite. I think you have made some other good points in this thread, but I don’t think this one is based on accurate information about the Yankees’ plans.

I was under the impression that CC was going to start games 1, 4, and 7. If that is incorrect then i will need to correct my statements.

The opportunity for CJ to watch how Lee handles the hitters is lost if Lee doesn’t pitch first. #1.

#2, although statheads may disagree, I think momentum is not meaningless. And the earlier (and therefore more frequently) you can run Lee out there, the more likely momentum is to work in your favor.

Yeah, I guess the videos of Lee pitching against the Yankees mean nothing, nor does the fact that CJ Wilson himself pitched against them a little over a month ago.

I suppose that’s true if the Yankee approach is exactly the same as it was a month ago, and that the Yankees, watching the same tape, decide to change nothing at all in how they approach the Rangers pitchers.

Maybe there’s nothing at all to be gained from sitting back and watching Lee bury another team’s hitters two days before you face them. Maybe the benefit you’d get is the same as watching tape of your opponents playing a meaningless non-playoff game 30 days ago. But if there *is* something to be gained, then it’s better to have it than not.

“I suppose that’s true if the Yankee approach is exactly the same as it was a month ago, and that the Yankees, watching the same tape, decide to change nothing at all in how they approach the Rangers pitchers.

Maybe there’s nothing at all to be gained from sitting back and watching Lee bury another team’s hitters two days before you face them. Maybe the benefit you’d get is the same as watching tape of your opponents playing a meaningless non-playoff game 30 days ago. But if there *is* something to be gained, then it’s better to have it than not.”

Sure, but there could also be something lost. I mean, I can think of all sorts of plausible scenarios where it would be better for Lee to see how the Yankees approach CJ Wilson than vice versa. Especially when you factor in the matchup. In any case, we simply don’t know if CJ needs to watch Lee, or if Lee needs to watch CJ, or if tape will be enough, or if the Yankees will even change their approach from merely a month ago, or if changing their approach will even help the Yankees, etc., etc. Willy’s position that so much is lost from Lee not pitching Game 1 is simply not rooted in something realistically perceptible or measurable. It’s far more likely his position is based on an unquestioned assumption he’s built up over the years that one should pitch one’s best starter first. There isn’t much reason to give unthinking assumptions much credence.

Yeah, or maybe my position is based on CJ’s statements that it has been extremely helpful for him to watch how Lee handles a team immediately before CJ faces them. But I guess if he can’t prove to us that it’s helpful by showing us statistical proof, it isn’t real and more likely an unquestioned assumption that CJ is making…

In addition to what q wrote, there’s no way that Lee is pitching Friday night on 2 days’ rest.

Why is everyone assuming that a series loss in 5 games is exactly the same as a series loss in 7 games?

It seems to me that whether or not the series is perceived as competitive will have a large effect on the Texas organization. If they get swept, then they won’t make as much money from the post season (east post-season game is worth a huge amount of cash), which might affect their budget for the offseason (ie re-signing cliff lee). And it seems to me that if Lee hasn’t already made up his mind to sign with the yankees then his decision might well be affected by how well texas performs against the yankees in this series (if they get blown out of the water, then he’s more likely to sign somewhere else if the money is just as good and the team is demonstrably better, right?)

So if one is only concerned with whether or not the rangers win the series, then Dave’s point is more or less correct (there might be some slight differences in matchups and bullpen usage, but it seems likely that those things will more or less even out). But that analysis assumes that all losing outcomes are equal (and that all winning outcomes are equal as well), which is not necessarily true.

The Rangers possibly lost the ALCS by not clinching in either games 3 or 4 at Texas against the Rays. Many pundits would likely be favoring the Rangers against the Yankees if Cliff Lee was starting game one. If I was manager, I would put him in game two and pull him after 80-90 pitches.

It would be funny if Lee blew this all up by losing his start.

Real aces are willing to pitch on short rest.

Have a cookie… hell have 40 boxes of cookies tubs

IMO, a “sense of urgency” in elimination games plays more like desperation than it does adrenaline or focus.

At this point, if u need the pressure of an elimination game to deliver a stronger performance, you’re likely doomed.

But, overall in this thread, whatever games Lee pitches not mattering is only valid if the series goes 7. If it ends in anything but 7, your best pitcher only started 1 game.

This would be like an amatuer team saving it’s ace for the state/national championship game, it’s only no big deal if you make it that far.

Given that random things can happen in baseball, it is important to keep the series close. A big break in a game where you’re down 2-0 is less than one that works in your favor when it’s 1-1.

To me, the big deal is that Lee only gets one start in this series, and TEX misses out on the finances from hosting an extra playoff game.

It’s still weird to me that a low PC pitcher requires 4 days rest in order to start. It’s entirely possible that three 6 IP starts by Lee is more valuable than two 8 IP starts. With all the off days in a series the BP gets more rest than usual. They can also rest until Spring Training if they lose. TEX should treat this as if it’s their last LCS for a decade, because it could be.

“IMO, a “sense of urgency” in elimination games plays more like desperation than it does adrenaline or focus.

At this point, if u need the pressure of an elimination game to deliver a stronger performance, you’re likely doomed.”

Sure, and I could say IMO, if you need the cushion of being ahead to play better, you’re not a competitor and also you’re likely doomed. Both opinions are worthless as they are not rooted in anything factual.

“But, overall in this thread, whatever games Lee pitches not mattering is only valid if the series goes 7. If it ends in anything but 7, your best pitcher only started 1 game.”

If it ends before game 7 with the Rangers winning, it doesn’t matter. If it ends before game 7 with the Rangers losing, it means your non-Lee pitchers failed to win the prerequisite two (three if Lee loses his first start) games they need to win the series. It simply doesn’t matter that Lee pitched 1 or 2 games, all else being equal.

“Given that random things can happen in baseball, it is important to keep the series close. A big break in a game where you’re down 2-0 is less than one that works in your favor when it’s 1-1.”

This makes absolutely no sense. Whether you’re 2-0 or 1-1, that win is no less important. In the playoffs, every win you get is important.

Well, purely anecdotal, but look at the Phillies/Reds in game 2. The Reds were up 4-0 in that game before turning it into a comedy of errors and losing 7-4. Did they play too conservatively thinking the game was in the bag, breaking from their natural rhythm? Were they thinking ahead to the next game rather than focusing? Did they feel added pressure trying to hold onto a lead? We can only speculate what kinds of psychological factors, if any, led to their downfall in that game, but clearly an early lead didn’t help them.

One team could be up 2-0 and go for the throat, while another could feel added pressure of trying to close out the series. It’s just guesswork to assume how teams will respond to particular situations.

Calculus the same, really?

Dude, game 7 is a crap shoot unless of course you don’t reach game 7. All evidence points to game 2 as the most important game of any series. Teams that go up 2-0 have historically won 81.4% of all 7 game series.

Statistics fail.

meaningless rank littered with quotation marks

That was a nonsensical reply. But I’m glad we agree your use of statistics is amateurish.

To me, the more interesting debate here is the game theory element of it. In other words, should the Yankees adjust their rotation now that they know what the Rangers’ is? Dan earlier brought up using Burnett in game 3 instead of 4, thus essentially punting that start in order to gain more starting pitching advantages in the other games. Would it make sense to use their best starter, CC, against Lee? Most here would say no, b/c they disagree w/ Dave on the 1&5 = 3&7 argument. Now that the Rangers have played their hand, what should the Yankees do? Stay the course? Change it? How?

Here’s a way to end the argument:

Run two different simulations. Every player is average, except for Cliff Lee, who is well above-average.

Have Cliff Lee start Games 1 and a presumable 5 in one of them. Run it 1000 times, or whatever.

Then, have Cliff Lee start Games 3 and a presumable 7. Run that one 1000 times or whatever.

This will end the argument.

Well that scenario wouldn’t lead to a difference. You can do a simple probability calculation to see that. For example, if Lee wins 70% and everyone else wins 50%, Rangers probability of winning is .625 either way.

The truth that makes the article hard to believe is that the starters not named Lee _aren’t_ identical. However, running the simulations with some actual measurements of performance would certainly get us somewhere.

What if Game 3 is rained out in NY? Now you have Lee pitching Game 7 on short rest. What if you have weather delays after Game 5? Lee’s still going to be pitching on full/extra rest, but now the game 1/5 starter might be able to come back for game 7

Dave says well the bullpen thing is a non-starrter as an enlightened manager can use Lee in the pen in Game 1 just as easily as if it were game 7?

I wonder if the limit on the pitcher would be the same in both scenarios? If you use Lee in the bullpen in Game 1, it’s likely 1 or 2 innings unless you want to potentially jeopardize his Game3 start. You bring Lee into a game 7 out of the bullpen? He pitches as long as he’s feeling OK as you don’t have to worry starting him after 2 days of rest.

The probabilities of these thing will likely be dismissed as minor, but having your best pitcher throwing in game 1 gives you the most options and flexibility.

the calculus? this is algebra.

This thread makes me want to beat my head against a wall. Or beat some commenters’ heads against a wall.

This scenario does allow the lesser pitchers to pitch at home and let’s Lee kick off the games at Yankee Stadium, which he’s done before. This might be some help to the the young guys. That’s about the only positive I see.

I haven’t been reading around, but is Lee taking any heat for not pitching on short rest? Seems like countless other elite players would be getting bashed for the same decision.

This is a moot point, because the Yankees already announced Pettitte in Game 5. Wouldn’t the Yankees be better of if they gerrymandered their rotation? Run Burnett (who will probably be terrible) against Lee in game 3 (effectively ceding the game) and then play hard for games 1,2, and 4. Why aren’t they doing this?

Why isn’t a professional baseball team conceding a game in a best of 7 series? I have no idea!

You are assuming that the Yankees won’t score enough runs in AJ’s start, why? There are 2 ways to win in baseball. Scoring and preventing runs.

Also, AJ has pitched decently against the Rangers this year. They just need him to go 5 innings and give up 3 runs. That’s a terrible start in a vacuum, but certainly doesn’t preclude them from winning the game.

Because the Game 3 starter would be the only one who is on regular rest in a possible Game 7.

Why do you want to pitch your worst starter earlier? It gives you the chance of a loss earlier rather than later, which invariably increases the chance that you lose the series.

This is what’s lost in this argument. The series ends once one team has four losses. Losing games earlier increases the chance that your team loses the series. Thus, you want your best chances to win earlier in the series. It’s very, very simple, made complex because Dave wants to appear schmart.

It’s hard to say.

The fact remains that, since Lee is only pitching twice, Texas must win at least 2 non-Lee starts to win the series. If those were 1 and 5, sure, you’d have the highest chance of seeing Lee twice, but Texas would still have to win 6 and/or 7.

If I were New York I’d pitch Sabathia 1, 4 and 7 but perhaps still use a 4 pitcher rotation. I would think CC 3x > CC 2x even factoring rest. Maybe the other guys would truly operate better on normal rest. With Lee pitching games 3 and 7, they could throw their non-CC starter with the highest day-to-day variability in performance at Lee in game 3. Lee is very consistent. Let’s say he’s a standard deviation on average above the non-CC NY starters who are themselves roughly equal on average. The one with the greatest variability would have the greatest chance at having an out-of-his-mind performance to trump Lee’s.

Either way the series will come down whichever non-ace starters are stepping up and winning games. Lee can only win 2 games max, Sabathia 2, maybe 3.

If Texas looses the series before getting to Lee twice, it’s the rawest of deals.

Under no circumstance do you want one of the best pitchers in baseball not throwing his first start until the 3rd game in a 4 loss elimination series. The match-ups may work out for Texas, but either way it’s a big deal.

I’ve done a full matchup listing of the Win Expectancy using the following formula based Cliff Lee either starting (1,5), (2,6), or (3,7) using the following pythagorean model:

Runs Allowed, Texas (RAt): (FIPt*IPt + FIPbt*(9-IPt)) / 9

Runs Scored, Texas (RSt): (FIPy*IPy + FIPby*(9-IPy)) / 9

Win Expectancy, Texas (WEt): RSt^2 / (RSt^2 + RAt^2)

where IP(t/y) is the expected innings pitched by the given t/y starter (equal to starting IP divided by games started)

I’ve also included data relative to xFIP in the spreadsheet, denoted with an x in front of the signifier (e.g. xWE).

I’m not entirely sure this model is correct, but it is the best idea I could come up with to factor matchups fairly quickly. Problems include lumping the entire bullpen into one FIP, whereas in the playoffs bullpen usage might differ (e.g. Rivera pitching two innings), altering the FIP slightly. I feel using the average IP per starter and the average FIP for the bullpen (assuming 9 inning games) is the dirtiest way to get an estimation of runs allowed in a given game.

Also, notice that there is no offense/defense being introduced here. I’m toying with some ideas currently (such as integrating team wOBA vs (L/R)HP with FIP to create a “better” RA model). I am by no means a statistics wiz, and I’m sure someone can do better than me.

For the Yankee lineups, I did my best to predict Joe Girardi, and used a lineups consisting CC, Pettitte, Hughes, AJ, CC, Pettitte, Hughes and CC, Pettite, Hughes, CC, AJ, Pettitte, CC and give the matchups for both circumstances, assuming Girardi might not have an issue pitching CC three times if need-be. When CC pitches on short rest, no dropoff is assumed.

Here are my findings: https://spreadsheets.google.com/ccc?key=0AjOG9z6t1CetdExNM3Z2VFMxY3lTWC1icVJRdi0zYXc&hl=en. Please note that I am not a statistical wiz and I might have (probably?) had errors, especially regarding cumulative win probability. WE/4 is simple, and for WE/5/6/7 I summed every permutation I could think of that resulted in the give scenario (please let me know if I missed one, if anyone cares to double-check).

Here are quick results (FIP/xFIP with CC going 3 times second):

Lee 1,5: .6193/.6045 or .5975/.5772

Lee 2,6: .6068/.5951 or .5905/.5701

Lee 3,7: .5996/.5910 or .5837/.5068

Again, remember that this does not include any offensive data into the expect runs scored. I’m working on remedying that for tomorrow.

Hopefully someone can make use of this data and I didn’t make too much of a mockery out of this idea. Just trying my first hand at baseball number crunching.

**************IMO, the problem with this sites playoff analysis is [1] the trend to just go contrary with the opinion, just to be unique (it’s annoying, seriously), and [2] the dwelling of seaslong stats against all teams instead of focusing on more relevant stats against strong teams********************

It can be very frustrating when you’re trying to learn and just can’t seem to get it. Keep trying and eventually it’ll click for ya. You can do it!

I am referring to …

[1] Using total team WAR as an important piece of deciding who should win a series).

[2] Using FIP instead of runs allowed when predicting who will win a game/series. Different tools, different purpose.

[3] The slant of finding evidence that favors teams whose actions align more closely with FG’s preferences (namely how good the Twins team and Reds offense is).

[4] Using season-long data against ALL teams, rather than using performance data against strong or similar/playoff teams.

Eventually, if enough non-traditional things are said, an author will get it right … then the parade can begin.

There are, recently, been some rather drastic non-traditional things said, that ended up being decisively wrong, and many reader comments seemed to “get it” … whereas the comments in the articles were what ended up being way off.

Please let me know where the non-traditional thinking of the recent FG articles was more accurate that traditional thinking. I’m afraid “you’ll eventually get it”, it not enough.

I am always willing to learn. But, I’m afraid what I am reading are conclusions based upon erroneous ideas or misapplied data (or the selection of the wrong data).

Keep trying. You’ll eventually start to understand a little bit at a time. You can do it.

you are all overthinking this. the rangers strategy for the series should simply be to wish that games 1 and/or 2 get rained out to push the series back a day.

dutchbrowncoat makes the best point in hoping for rain. Not sure about all the statistical calculations, but two things make sense:

1 – If Yankees start AJ Burnett to concede a game, then the Rangers basically have a one game advantage.

2 – Rangers should try to start CLee as soon as possible in case there is rain and he gets rest because of weather.

In any case, should be a good series!

“you are all overthinking this. the rangers strategy for the series should simply be to wish that games 1 and/or 2 get rained out to push the series back a day.”

Well, they can wish for rain, but hoping for the weather to intervene can hardly be called a “strategy.” Maybe I’m missing some sarcasm.

terrible article. go back to math.

The point that’s being missed in the article is this:

“The natural assumption is that this is a big problem for Texas, and gives the Yankees a huge advantage. In reality, it doesn’t matter all that much.”

Nowhere in the article does Dave suggest there’s NO benefit from the order in which the rotation is set, just that it’s not some enormous benefit. There’s some variance for matchups, and a slight difference in the expected playoff revenue due to different likelihoods of playing more games, or how the possibility of a rain delay changes things, and there could be some subtle differences in performance depending on how the teams respond to different situations psychologically.

The thing that’s being missed is that it’s never suggested that the order is completely irrelevant, just that it’s not the substantial disadvantage that it might seem intuitively to some. We’re not looking at a detailed evaluation of an unorthodox strategic decision here, we’re looking at whether the Rangers put themselves at a big disadvantage by having to play a game 5 and pushing back Lee’s first start. Granted some people have a different view on how you’d define “big deal,” but I didn’t take from the article that Dave’s denying there could be some small impact based on optimization of matchups or other factors. What I took from it, which is correct logic, is that there’s no fundamental advantage in probability from the order in which the ace pitches.

We can debate about how much of an impact the psychology of being up compares to the possibility of rust after a longer break or how much the extra rest helps players with nagging injuries, and we can research exactly how much of an advantage the different matchup optimizations create, and then argue about whether those differences are enough to be considered a “big deal.” That’s not what the article’s about though. What Dave’s saying is that there’s no fundamental difference in ensuring your ace’s second start and relying on lesser pitchers to close out the series versus relying on those lesser pitchers to get the ball to your ace so that he can close the series. In that regard, his logic is correct.

It wasn’t int he article, but in the chat yesterday Mr Infallible said that the effect is 0. There were several questions about it.

And in that regard I’m in agreement with you. Dave has a tendency to weaken his points by using absolute language when there’s an intuitive gray area. Like he did in his chat.

That’s not present in this article, whatever his beliefs on the matter are.

he said there is 0% chance and was dismissive toward some guy with a pretty good idea of why it isn’t 0%.

Serious question:

Isn’t this the perfect analogy to using your closer in a tie game on the road??

In both cases, you still need a non-ace (relief ace) to succeed to win.

Yet in one case, the “SABR” argument is clearly that you should use your ace as soon as you can to avoid the situation of the game (series) ending without getting that chance.

Am I missing something? Seems inconsistent.

It’s a little different. I don’t know the math behind the closer issue, but the way I look at it, the difference is this:

In a 7 game series, you always have exactly 7 games to try to get 4 wins. It’s a finite number we’re dealing with. We don’t play the extra games if a team reaches 4 wins early because the outcome doesn’t matter.

In a tie game, there’s no finite number of innings or runs. Your team could score three runs after you hold the other team to a zero, making the job easier for the next guy. Comparing the two:

7 game series: down 2-3: You can either go with your ace in game 6 and another pitcher in game 7 or vice versa, but either way you need to win both games.

Tie game, road team pitching, bottom of the ninth:

If you use the closer first, then score two runs, then your second best reliever gives up one run, you still win. If instead your second best reliever gives up a run first, you never have the chance to score those two runs, so the order in which they’re used becomes more important. The outcome of an inning isn’t a binary event like a win or a loss, which is a subtle but important difference between the two. If both relievers absolutely had to pitch scoreless innings, the order wouldn’t matter, but that’s not always the case.

Thanks Wilson. That does make sense.

I guess I was thinking about it like this: If you use Lee in game 6 to force a game 7, maybe your offense comes out and scores 10 runs in game 7 rendering your starter irrelevant. Similar to the extra-innings argument. Your inferior game 7 starter can “close out” the series, even by pitching poorly.

“If your 2nd best reliever gives up a run first, you never have the chance to score those two runs”

Sounds an awful lot like, “If your 2nd(or 3rd and 4th) best starters lose 4 games first, you never have the chance to use your best starter(again)”

Certainly, Lee could win 1 and 5 and the Rangers lose the series, but the series ends before a game 7 more often if Lee only throws once in the first 6 games than if he throws twice.

Put differently, the Rangers chances of losing the series before a game 7 went up when Lee had to throw game 5 against the Rays.

Steve:

Your offense can also score 10 runs in game 6 to make your started irrelevant. That’s where it differs from a closer: the need for perfection is always there when the game is tied on the road, whereas it’s not so predictable when the need for perfection is greater in two separate games.

Rob,

The Ranger’s chance of losing before 7 is offset by their increased chance of winning IN 7.

Using some unrealistically extreme examples, suppose you have a relief pitcher who you know will give up zero runs and one who will give up exactly one run. If you use the second in the tie game, you will always lose, whereas if you use the first, you retain a chance to win.

Now take the series after five games. Suppose you’re down 2-3. You have two pitchers, one who you know will give up one run, and the other, seven run. Your chance of getting to game 7 is greater with the better pitcher in game 7, however your best chance to win the series is for your weaker pitcher to pitch against whoever’s more likely to give up more than seven runs, regardless of the order.

On the other hand, if you’re UP 3-2, you’re better off conceding the likely loss with the weaker pitcher and using the better pitcher against whoever’s least likely to throw a shutout. So the matchups DO make a difference in the preferred order, but the optimal order varies depending on the outcomes. The article’s not looking at matchups, though – it could, but it’s a more general point than that. We need to look at both possibilities – the possibility of being up and needing to hold onto the lead versus the possibility of being down and needing to come back.

Thanks again, good stuff.

WilsonC, i appreciate that you ran these numbers:

Lee 1, 5:

W4: 0.0384

W5: 0.11904

W6: 0.1216

W7: 0.12595

Probability of winning the series: 0.40499

Lee 2, 6:

W4: 0.0384

W5: 0.07936

W6: 0.16128

W7: 0.12595

Probability of winning the series: 0.40499

Lee 3, 7:

W4: 0.0384

W5: 0.07936

W6: 0.10752

W7: 0.17971

Probability of winning the series: 0.40499

however, your numbers also say that:

probability of TEX winning in <= 5 games

Lee 1, 5: 15.74%

Lee 2, 6: 11.78%

Lee 3, 7: 11.78%

probability of TEX winning in <= 6 games

Lee 1, 5: 27.90%

Lee 2, 6: 27.90%

Lee 3, 7: 22.53%

as you can see, there is a 4-5% greater probability of winning the series earlier when lee starts in games 1 or 2. that's the difference b/w a .500 team and a 89 win team. but this is looking at it from the viewpoint considering all possible outcomes before the series starts, as your numbers do.

in reality, path dependence matters becase we have to re-run the numbers after each game is finished. if lee starts game 1 and actually wins, you can immediately eliminate all the paths involving TEX getting swept. we've moved up the tree of possible outcomes. if TEX gets a clutch start from wilson in game 2 (or whatever) and actually gets the win, then all paths involving a TEX loss in 5 games are eliminated. we've moved further up the tree with more outcomes adverse to TEX being eliminated.

obviously, losses for TEX in games 1 or 2 would have the opposite effect per your initial outcome distribution. so you want to maximise your chances of favourable outcomes early on so that you're going the right way up the distribution tree, to maximise the odds that the new data we're inputting to recalculate the probabilities is favourable. there are also the real-life ramifications of frontloading and stretching a series as long as possible, such as letting lee pitch in game 7 knowing you don't have to save him for anything.

of course, i'm pretty bad at probability theory, so i could be totally wrong.

I like Lee pitching 3 and 7. The difference is very marginal, but you have Lee pitching in NY where he destroyed in the WS last year, and the other guys who maybe are more comfortable in Texas starting games 1 and 2. NYY are been-there-done-that and probably just as good at home or on the road, but Texas is a young upstart team with just 1 pitcher who has proved he can win a playoff game in NY.

But game 7 is in Texas. Either way, he’d be pitching 1 home and 1 road game.

This is crazy talk.

If we were going to play all 7 games no matter what, then yes technically the expected probability of winning would be the same.

But different pitchers (and different pitching matchups) influence the probability of winning a game, and the probability of winning the next game in a best-of-7 IS dependent on what happened in past games (for ex., if the Yankees sweep Games 1-4, p(Rangers win) for Games 5-7 = 0, no matter who is set to pitch).

Here is a thought experiment: The Rangers staff has two pitchers, Tommy Hunter (p[Rangers win] = .4) and Cliff Lee (p[Rangers win] = .6). Rest doesn’t matter.

Which rotation leaves them more likely to win the series? HHHLLLL, or LLLLHHH?

This is the sort of logic that leads someone to heavily favor the Twins in a short series.

But you are applying variables to a basic logical argument. There really is not much room for statistics or variables in the fact that the Yankees play NOT Cliff Lee 5 times.

Of course different pitchers and match-ups influence probability of winning. Also, past results probably do impact the next game’s results. That being said, this analysis is based on the assumption that Texas is going to win both of Lee’s starts, which is also dependent on match-ups and clearly becomes less likely if he faces Sabathia in one or both of them. Another factor that may need to be considered is the comfort of young pitchers pitching at home. Colby Lewis gets to pitch Game 2 at home instead of pitching Game 3 in NYY. If we run a sim of Games 1 and 2, it would seem that a split is the most likely scenario (albeit with the Yankees winning 2 as the 2nd most likely). If Texas wins the series, when we look back at it in hindsight, it may well be because they split the first 2 at home without Lee, took a 2-1 lead with Lee starting Game 3, won 1 of the next 3 and won Game 7 with Lee.

The point of the article was not that starting Lee in Game 3 is the most optimal scenario. The point was that Lee is starting Game 3 because he pitched on Tuesday and having him start 3 and 7 instead of 1 and 5 is not as much of a negative as some (many?) are making it out to be.

So far its looking like C.J. vs CC wasn’t a bad idea for the Rangers….5-0 beg of 7th

OOOppppps Yanks 6-5

It matters now!

The chances of reaching Game 7 just dropped a ton. Lee is only going to get 1 shot in this series. If he was going Game 1 and 5, he’d at least get his two starts. not anymore.

Pyschological advantage and actual advantage to the Yankees