THE BOOK--Playing The Percentages In Baseball

I think there may be some difference between “in-game momentum,” and “game-to-game momentum.” Within a single game, when a team storms back from a large defect to tie the game, doesn’t it feel like they have a better than 50% chance of going on to win?  The other team’s bullpen has likely been burned through trying to keep the lead as it is getting away, and their players are like to press more to prevent a would-be embarrassing loss.  As far as momentum over multiple games, as cliche as the phrase “momentum is only as good as the next day’s starting pitcher,” it really does make a lot of sense.  I mean, regardless of how much “momentum” the Red Sox have built up with their game 5 win, isn’t is all lost if Beckett comes out throwing 88 with a flat curve?

I think then, the big difference between in-game and game-to-game momentum is the opportunity for pitcher on the team being overtake to come in and stop the bleeding.  Within a single game, all of your best relievers could very well have been used.  Within a series, there is always a chance that you have a top starter pitching in the next game.

It would be fantastically easy to test the assertion that a team, coming from behind by at least 5 runs to tie the game, and then the next half-inning starts tied, that the team that came from behind had a better than 50/50 shot at winning.

Somebody want to do it?

I’ve recently thought that chemistry is most important (perhaps the only place where chemistry is more than marginally important) is on the defensive side of the ball, both with the pitcher-catcher relationship and the other 7 fielders. This is strictly intuitive, of course, but I’d like to know if anyone has seen any research/studies regarding this subject.

Those who really care about chemistry should study sports where players interact all the time, like hockey. 

Trying to find chemistry and momentum in a sport like baseball is a much harder endeavour than doing it in a sport like hockey.

Against my better judgement, I decided to run that study I noted earlier.

I looked at all the Retrosheet year (a 50 year time period) and I looked for all games where the batting team ended the inning tied, after coming from behind by at least 6 runs.

So, entering the next half-inning, the team that came from behind had super-duper non-stop momentum.

This happened 90 times.

How often do you think the team that had the momentum ended up winning the game?  46.  That’s a shade over 50 percent.

Let’s move along folks.  Nothing to see here…

That is really a terrific finding.  It then begs the question, if momentum doesn’t exist within a single game, how could it possibly exist between games?

I put myself through college betting against the Bengals....

I’ll go on record here as saying that I’ll gladly bet against any kind of momentum anyone wants to posit that can be checked like Tango did.  Of course, you’ll likely run into sample size issues.  Tango checked 90 games and found a 51.1% wp.  But that is plus or minus 10% at 2 sigma, so it doesn’t tell us anything other than if there is any effect of coming back from 6 or more runs it is likely not worth more than 10%, or something like that.

I’ve studied baseball inside and out for 20 years now.  I’m not saying that I couldn’t have missed something but I’ve never, ever found anything other than team A will win X percentage of time against team B, as predicted by an estimate of team A’s and team B’s talent for that particular game.  Never. 

Again, that is not to say that momentum or 100 other psychological and emotional issues don’t come into play in baseball as well as in other sports.  It is just that, like clutch play by individual players, it is next to impossible to identify and generally gets drowned out by all the noise inherent in events that have a large random component.

The way the media and every commentator in the world discusses these things, though, is complete and utter nonsense that any rational and intelligent person should ignore or chuckle at.

Example number 1,287,465:  2 days ago, in the 7th inning of game 5, Tampa was the team with heart, destiny, and great chemistry and Boston was the ho-hum team that had lost its edge and its desire for another championship.  Now, Tampa is one loss away from being the young, inexperienced team that couldn’t handle the pressure of a post-season series against the powerful, experienced, never-give up Boston team.

Oh, and the last game will be a coin flip.  I have not seen the Vegas line yet, but it will be abut even money.  So there are two scripts ready and waiting.  Each script has a 50% chance of being foisted on the public by the media and the fans.  No one knows yet what the script is going to be.  I will absolutely bet any amount of money, and I mean ANY, if anyone wants to lay me 2-1 or even 3-2 on either script.

One script of course, is that Boston broke the heart of Tampa in game 5 and that they are destined to win game 7 now that they have tremendous momentum and their ace pitcher going.

The other script is that Tampa recovered from game 5 and game 6, and their talent was too strong for Boston to overcome and their players were too strong of character to fold under the pressure.

Let’s hear which script it is BEFORE the results of game 7, and let’s see that person who posits that script put their money where their mouth is.  As I said, I am willing to take only 3-2 odds.  Surely if you believe in “stories” one team or the other has more than a 40% chance of winning game 7.  If your not willing to lay me 3-2, then you basically believe that the odds of one team or another winning the series at this point in time is equal to the relative talent of the two teams and their starting pitchers.

As Tango likes to say, talking about character, chemistry, heart, and desire after the fact is meaningless, worthless, and insults our intelligence.  If even a tremendous underdog in a series is going to win it 1/3 of the time WITHOUT any psychological or emotional factors whatsoever, what is the point of talking about heart, desire, and character one out of every 3 series?

What is the point of even discussing this kind of nonsense?  Seriously.

Boston -130 for Game 7.

Unloading on the Rays :^)

Ok, now that I can do some quality control, ignore what I just said.

The team with the momentum wins 64 and loses 44 (total of 108).  That’s a .593 win%.  Impressive, though not statistically significant.

This is the most recent game (in 2007):
http://www.baseball-reference.com/boxes/CHA/CHA200709070.shtml

Twins/Sox tied going into the 9th.  Twins score 6, then Sox score 6 in the bottom of the 9th!  Sox win in the 13th.

If I make it a 5-run deficit, the team with the momentum wins .558, with 496 games.

That is 2.6 SD from the mean.

So, consider me impressed.  That is not to say that .558 is the true mean for the momentum win.  It just happens to something north of .500.

Very cool.  I think I will write this up, and check to see if there is a potential home-bias.

I would certainly look for not only a home team bias but since you are basically looking at a team that scores at least 5 unanswered runs, they may be a much better team (I’m not sure about that since they were also down by the same number of runs), or at least the other team may have a bad bullpen.  I would guess that there is some kind of pretty large bias.  It may be that the teams are even overall, but that the team that was originally ahead has a much better starter and the team that caught up has a much better bullpen and much better offense.  I would definitely control or at least look at offense and home/road for starters.

I had said not statistically significant for the down by at least 6, but that was too quick on my part.  It is nearly 2 SD, so it is a bit impressive.

In any case, they are all significant, depending on how strong you want to make it.

Here’s the full data:
COME_FROM_BEHIND MOM_RATE N
-7 0.595 37
-6 0.589 90
-5 0.557 427
-4 0.524 1496
-3 0.522 4994
-2 0.533 15445
-1 0.533 41881

So, there seems to be some bias here.  The team that scores to catch up, even just by 1 run, seems to be better than their opponent.  There also seems to be a clear line at being able to catchup by at least 5 runs, giving rise to potential momentum talk.

Cool…

MGL, I’m willing to take that bet.  I said before game 6 that I was 99.9% sure Boston wins this series.  Now I’m as close to 100% as I can get. I don’t really believe in momentum, but there’s something terrible that follows around that team with a B on their caps, always ready to break the hearts of the teams I like.

Since gambling is illegal in most places, lets make it a charitable donation to retrosheet.  If Boston wins tonight you donate $100 to retrosheet.  If Tampa Bay wins I donate $150 to retrosheet.

I will consider the bet to be on if you confirm on this thread or by email before the game starts.

If I lose this bet I will consider it the best $150 ever spent.

Here’s the data broken down by whether the team that came back to tie it up was the road team or home team:
BAT_HOME_ID COME_FROM_BEHIND MOM_RATE N
0 -7 0.500 12
0 -6 0.536 28
0 -5 0.474 171
0 -4 0.433 624
0 -3 0.434 2179
0 -2 0.456 6802
0 -1 0.460 18474

The above is the road team.  We see that when the score is tied, the chance of the road team winning is .460 after coming from behind to score 1 run, and entering the bottom of the inning (home team now has 3 more outs).  But, coming from 2, 3, 4 runs puts them in a worse position.  Coming back from 5 or more gives them a tiny bit of momentum.

1 -7 0.640 25
1 -6 0.613 62
1 -5 0.613 256
1 -4 0.589 872
1 -3 0.590 2815
1 -2 0.593 8643
1 -1 0.591 23407

The home team, coming back from 1, 2, 3, or 4 runs to tie it up gives them the same win rate.  At 5 or more runs of comeback, they get a tiny momentum bump.

"But, coming from 2, 3, 4 runs puts them in a worse position.  “

Clearly, this has to happen in the later innings.  And coming back from 1 run might happen disproportionately in the earlier innings in comparison.

Here’s a good chart to be familiar with:
http://tangotiger.net/innwin2.html

The team that scores to catch up, even just by 1 run, seems to be better than their opponent.  There also seems to be a clear line at being able to catchup by at least 5 runs, giving rise to potential momentum

Is it possible though that the underlying conditions of the game changed (eg a depleted bullpen, a strong starter removed from the game), and those conditions, and not momentum, are responsible for the superior winning percentage of the comeback team? For example, here’s a table of runs per game by the road team based on how many runs they scored in the first inning (based on Retrosheet’s linescores through 2005):

1st  r/g   Diff
 0   3.77  3.77
 1   4.90  3.90
 2   6.05  4.05
 3   7.10  4.10
 4   8.32  4.32
 5   9.36  4.36
 6  10.84  4.84

Diff is just the runs per game minus runs scored in the first inning. You’d think that “diff” would be the same regardless of how many runs were scored in the first inning. In other words, that runs per game would equal runs in first inning + (8/9 * team average runs per game). But ... scoring first-inning runs implies that the conditions of the game are conducive to run scoring, and it appears that those conditions carry over into the other innings. For example, scoring three runs in the first means it’s more likely the team is facing a weaker starting pitcher, or that the game’s being played in a hitter’s park etc etc.

So, with regard to observed “momentum” after five-run innings, I wonder if it would still be observable if you controlled for the opposing pitcher, park etc, and all of the other conditions that scoring six runs in an inning implies.

Read #11 above Tango about controlling for other factors.

Rally, you got it.  How about charity of our choice, to be verified?  We have to send the money to a third party, maybe by Pay Pal or something like that to make it easier. Or make the donation through Pay Pal and email a verification to the other.  (I already donate all of my profits from The Book to retrosheet.) We can leave it at Retrosheet if you want. Of course I trust you 100%.  First bet I offered ever taken!  I give you mad props for that.

FWIW, the Vegas line is now around Bos -126, TB +116 which means they think the true odds are BOS 54.75%.

It’s on.  Why don’t we leave it at retrosheet then?  If I lose, I can send the money to a third party (perhaps Tango?) or I can send you a copy of the check.  I trust that you will uphold your end of the bet, so I’m not too worried about verification

Thanks, yeah, #11.

Here’s another way you could slice and dice the numbers. Take all of the half innings where a team scored five runs, and then show the average run scored over the remaining innings based on the score differential at the end of the five-run inning (ie, five-run innings that brought the score to -2, to -1, that tied the game etc). If momentum exists, then r/g should be highest when the big inning ties the game (or when it pulls the score to within a run).

Also, for the improvement in w% noted by Tango, it would be interesting to break that down by runs scored and allowed. If the improvement was equally split between offence and defence, then momentum could exist (although I don’t think it does). But if it was purely offence, then it’s likely that the improvement is caused by changes in the underlying conditions.

FWI also W, “friendly” wagers between two people is rarely illegal (in the various jurisdictions), as far as I am aware.

Retro is fine.  Good luck!  I went to the game last night, and I am going tonight.

PayPal provides a “receipt” of all transactions.  I think it would be cool to take an image copy of that, and post it on a website (with personal information redacted), as “proof” for bets, for tracking purposes.

(Just do “printscreen” and then paste to MS Paint.)

Btw, thanks to Adam for the original idea.  Sometimes, going against my “better judgement” is good.

dackle, I like your idea.  I’ll see if I can do something with that later tonight.

***

Here’s the data broken down by inning.  I also took care of the “double-counting” issue, whereby a come from behind 1 run would also exist in the come from behind 2 runs and so on.  That is, if a team comes back from 6 runs, they also came back from 5 and 4 and 3 and 2 and 1 run.  Not a big deal, but I made sure to count it as the lead at the start of the inning.

Anyway, here it is, with no quality check yet:
BAT_HOME_ID INN_CT COME_FROM_BEHIND MOM_RATE GAMES
0 1 0 0.438 91924

0 2 -7 1.000 1
0 2 -5 0.500 4
0 2 -4 0.462 13
0 2 -3 0.409 93
0 2 -2 0.477 516
0 2 -1 0.457 2215
0 2 0 0.449 49647

0 3 -5 0.400 10
0 3 -4 0.412 51
0 3 -3 0.425 174
0 3 -2 0.435 689
0 3 -1 0.452 2672
0 3 0 0.448 32597

0 4 -6 0.500 2
0 4 -5 0.684 19
0 4 -4 0.475 59
0 4 -3 0.432 236
0 4 -2 0.453 935
0 4 -1 0.456 2809
0 4 0 0.451 23425

0 5 -7 0.500 2
0 5 -6 0.333 3
0 5 -5 0.500 20
0 5 -4 0.425 87
0 5 -3 0.442 292
0 5 -2 0.468 827
0 5 -1 0.460 2435
0 5 0 0.452 18125

0 6 -7 - 2
0 6 -6 0.750 4
0 6 -5 0.391 23
0 6 -4 0.427 89
0 6 -3 0.424 302
0 6 -2 0.462 1004
0 6 -1 0.460 2467
0 6 0 0.451 14898

0 7 -7 0.667 3
0 7 -6 0.800 5
0 7 -5 0.500 30
0 7 -4 0.373 83
0 7 -3 0.418 318
0 7 -2 0.463 832
0 7 -1 0.465 2109
0 7 0 0.457 12989

0 8 -7 - 1
0 8 -6 0.400 5
0 8 -5 0.419 31
0 8 -4 0.381 105
0 8 -3 0.475 297
0 8 -2 0.450 753
0 8 -1 0.473 1862
0 8 0 0.450 11582

0 9 -7 1.000 1
0 9 -6 0.400 5
0 9 -5 0.350 20
0 9 -4 0.582 67
0 9 -3 0.434 244
0 9 -2 0.453 664
0 9 -1 0.466 1576
0 9 0 0.446 10182

0 10 0 0.436 8718
0 11 0 0.439 4862
0 12 0 0.440 2714
0 13 0 0.441 1515
0 14 0 0.438 864
0 15 0 0.442 462
0 16 0 0.428 264
0 17 0 0.438 144
0 18 0 0.457 81
0 19 0 0.475 40
0 20 0 0.500 24
0 21 0 0.571 14
0 22 0 0.500 10
0 23 0 0.333 3
0 24 0 0.333 3
0 25 0 0.500 2

1 1 -6 - 1
1 1 -5 0.667 3
1 1 -4 0.432 37
1 1 -3 0.582 182
1 1 -2 0.586 836
1 1 -1 0.592 3285
1 1 0 0.564 65107

1 2 -7 0.667 3
1 2 -6 0.500 4
1 2 -5 0.400 10
1 2 -4 0.659 41
1 2 -3 0.589 158
1 2 -2 0.569 744
1 2 -1 0.576 3101
1 2 0 0.560 39768

1 3 -6 0.667 3
1 3 -5 0.548 31
1 3 -4 0.566 83
1 3 -3 0.550 282
1 3 -2 0.562 918
1 3 -1 0.574 2922
1 3 0 0.558 27290

1 4 -7 - 1
1 4 -6 0.667 3
1 4 -5 0.643 28
1 4 -4 0.600 95
1 4 -3 0.604 303
1 4 -2 0.557 1030
1 4 -1 0.581 3130
1 4 0 0.563 20251

1 5 -7 0.500 2
1 5 -6 0.800 5
1 5 -5 0.545 22
1 5 -4 0.576 92
1 5 -3 0.565 322
1 5 -2 0.552 938
1 5 -1 0.568 2422
1 5 0 0.566 16318

1 6 -7 0.625 8
1 6 -6 0.700 10
1 6 -5 0.553 38
1 6 -4 0.545 112
1 6 -3 0.560 352
1 6 -2 0.615 951
1 6 -1 0.587 2460
1 6 0 0.570 13731

1 7 -7 0.667 3
1 7 -6 0.417 12
1 7 -5 0.519 27
1 7 -4 0.644 87
1 7 -3 0.551 316
1 7 -2 0.631 873
1 7 -1 0.594 2121
1 7 0 0.579 12260

1 8 -7 1.000 1
1 8 -6 0.625 8
1 8 -5 0.621 29
1 8 -4 0.621 103
1 8 -3 0.624 274
1 8 -2 0.604 747
1 8 -1 0.607 1885
1 8 0 0.603 10817

1 9 -7 0.800 5
1 9 -6 0.800 5
1 9 -5 0.791 43
1 9 -4 0.570 93
1 9 -3 0.697 271
1 9 -2 0.657 749
1 9 -1 0.686 1794
1 9 0 0.656 10415

1 10 -4 0.667 3
1 10 -3 0.769 13
1 10 -2 0.551 49
1 10 -1 0.762 239
1 10 0 0.650 7079

1 11 -4 1.000 1
1 11 -3 0.545 11
1 11 -2 0.581 31
1 11 -1 0.653 118
1 11 0 0.653 3950

1 12 -4 1.000 1
1 12 -3 0.333 3
1 12 -2 0.571 21
1 12 -1 0.726 73
1 12 0 0.647 2187

1 13 -3 0.500 2
1 13 -2 0.727 11
1 13 -1 0.706 51
1 13 0 0.642 1214

1 14 -3 0.500 2
1 14 -2 1.000 3
1 14 -1 0.545 22
1 14 0 0.646 698

1 15 -3 0.500 2
1 15 -2 0.750 4
1 15 -1 0.571 7
1 15 0 0.646 379

1 16 -1 0.750 8
1 16 0 0.634 216

1 17 -3 - 1
1 17 -1 1.000 2
1 17 0 0.617 120

1 18 -1 - 2
1 18 0 0.625 64

1 19 -1 - 1
1 19 0 0.571 35

1 20 0 0.550 20

1 21 -3 1.000 1
1 21 0 0.500 10

1 22 0 0.714 7

1 23 0 0.667 3

1 24 0 0.667 3

1 25 0 1.000 1

Anyway, this presentation leaves alot to be desired.  What I need to do is to take tie-game line for each inning, and use that as the baseline, for a weighted average.  I’ll do that later.  This way, we can compare apples to apples.

Interesting actually that the w% seems to tail off when the game is tied. If I’m reading the very first set of #s correctly, if the visitor came to within a run in the second, the w% is .457, but if they tied it the w% is .449. Seems to be a similar pattern in the other groups as well. There could be a “counter-momentum” argument, in that tying the game forces the team that gave up the lead to bear down. This is my issue with “momentum” arguments—why are the Rays viewed as cardboard scenery that must bend to the character and will of the Red Sox? They’re all professional athletes, and the Rays have as much say in what happens on the field as the Red Sox do. You see that a lot in the gambling forums—team A is “not going to let” team B win. Well, I think team B is going to have a say in the matter. If momentum exists, it can exist for the Rays as well (ie it forces them to bear down), and the two cancel eachother out.

Now that I think about it, it’s going to be very difficult to isolate the momentum aspect using historical linescores. Let’s say you look at all of the five-run innings and the score at the end of those innings, and you see a spike in w% after those innings that tied the game, vs those that brought the team to within a run. Well, the fact that the big inning still leaves the comeback team down by run carries some information about the relative strengths of the teams—the team with the one-run lead is still probably a slightly stronger team. You could tease it out and only look at runs scored following the big inning, but again, I think someone might argue that pitching is part of momentum, and that a big inning should lead to not only more runs scored, but fewer runs allowed.

So the study might require a different approach altogether, maybe turn to our old friend multiple regression. Regress the number of runs scored in each of the first six innings against the number of runs scored in the seventh (or the average in the seventh through ninth). So the first six columns of the spreadsheet would be the runs scored in each of the first six innings, and then the seventh column would be the average runs in innings 7-9. It would be interesting to see if the relative weight was heavier on the most recent innings. For game results for teams, I really don’t think much emphasis is needed on recent games. But within games, I could see that recent innings might carry more weight, simply because the personnel available in the early innings isn’t always the same as in the later innings (ie after the starter gets yanked the expected runs allowed changes).

Finally, a person who believes in momentum could argue that it doesn’t matter what the underlying causes are—the only thing that matters is that five-run innings to tie games means that you’ve got a better chance of winning.

I found a bug in my code.  Here are the correct numbers:

Down by at least 5 runs, to tie the inning: 383 games.  Team with the momentum wins .504.

Anyway, now that I’ve got my program working fine I think (never program and baby sit at the same time), I’ll write up something lengthier later in the week.

You mean you can program and NOT babysit? That sounds amazing. How can I try that?

MGL, what are your total losses from this game?

Total losses?  I think MGL had a series of bets on the Rays.  Including mine listed above.  I just sent $150 to retrosheet via paypal, and sent an email copy to MGL.  I have never been happier to lose a bet.  Also, I can take some pride in being the first to pick up one of the many bets MGL has offered on baseball sites.

One thing we humans do is pattern recognition.  I’ve seen this before, with Red Sox teams pulling off impossible wins against teams I was rooting for, and it seemed like a bad dream, it was all happening again.  Even worse this time, as the Angels and Yankees never blew 7 run leads.  But just because we see a pattern doesn’t mean it always has to turn out that way.

Maybe there’s a tiny bit to momentum, but for the most part I think momentum is just an obscure college running back.

Mentum, Maurice: 5’11 235 lbs.  Powerful runner, not especially quick on the first step.  But once he gets running in the open field, impossible to bring him down.

Momentum..Shmomentum..!

Mick, congrads. on the wins!! Good luck with the WS.

yup, i had tb to win the pennant and series for a few bucks (legally bet in a nevada casino).

looks like so far tb is around a -140 fave over philly.  i am a little surprised it is that low as most people know that the al is far superior in general (which does not necessarily mean that the AL pennant winner is necessarily a lot better than the NL pennant winner, but i think that most people would agree that tb and philly are both two of the better teams in their respective leagues - in fact, they are both probably the second best teams in their leagues).

Anyway, good game Rally.  As I said, I figured it to be a coin flip and TB came out on top.  It was a very good game.  I just got back from it.  Price was amazing.  I am not sure I would have brought him in in the 9th to finish the game, but who knows.  I don’t.  Maddon finally made all the right moves in the 8th, bringing in lefties and righties as needed, as if the whole series were on the line (as opposed to game 5, which he appeared to manager as if the game didn’t matter, or at least as if they were going to win no matter what he did).

You could just tell that Maddon was managing this game so that he would not get criticized if they lost.  He was not trying to “be smarter than everyone else,” which I think is one of his weaknesses as a manager.  (Being smarter than everyone else is OK of course, if you really are.  But he is not.)

The only mistake he continues to make, which was obvious if you were at the game (you get to see everything that goes on on the field as opposed to only what the TV station lets you see).  He almost never sacrifices bunts, which is an example of him trying to be smarter than the “old school” managers.

As I have explained many times, you need to sac bunt sometimes (with a good bunter and fast runner - those are the only criteria - obviously it has to be a close game) in order to keep the defense from being able to play back, thus taking away some hits when swinging away.

With Upton up in the (I forgot what inning it was) and a runner on first and no outs, and I think the Rays were down by a run, Youkilis was playing behind the bag at third, confident that Upton was not going to bunt.  I don’t know if Upton is a good bunter or not, but he is awfully fast.  If he drops a bunt down the third base line, he is going to be safe around 40% of the time.  That means that on any sac attempt by him with Youk playing back, he is going to be safe maybe 20% of the time or so.  That is enough to make the sac attempt better than swinging away.

Equally importantly though is that even if Maddon sacs occasionally (say 25% of the time), he forces Youk to play up, thus increasing the number of hits when he swings away the other 75% of the time (and of course, he now only reaches base safely on a bunt attempt 15% of the time).

The same thing happened when the game was tied or maybe TB had taken a lead and there were runners on 1 and 2 with no outs and I think Bartlett at the plate.  Bartlett is also fast and a good (or decent at least) bunter, I think.  Youk was playing about even with the bag, maybe a little in.  Boston was pretty sure that TB was not going to bunt, but not 100%.  Again, no attempt by Maddon to sac because he rarely does.  He does not understand how to use game theory in a potential bunt situation, but he thinks that he is smarter than everyone else and that bunting is a “bad thing.”

I was a little surprised that Francona left Lester in for so long.  At some point in time around the 5th inning or so, all of his fastballs were only registering 88 or 89 on the stadium gun. I don’t know if he was only throwing cutters for some reason, or that he lost his fastball.

I think that Maddon left Garza in a little too long as well.

But it was a good game.  It was a great, loud crowd, but anyone who thinks that certain teams in baseball have large HFA (other than because of something quirky in the stadium) should think again.  I doubt that the players on the field pay any attention to the crowd whatsoever, and even if they do, it is hard to tell when they are cheering or booing, and 20% at least of the fans at the Trop were Boston fans anyway.

The idea that fans at a stadium or a stadium itself (again, other than some quirk) has any significant affect on the players in a baseball game is ludicrous.  In football, a team sometimes cannot hear the QB’s signals on the road, and in basketball, one, the crowd is right on top of the players, two, they are waving things right in the foul shooters faces, and three, refs in basketball are known to favor the home team.  In baseball, the players play the game and that’s about it.  Thus the small HFA in baseball and thus the fact that there is no evidence that any team has any more of a HFA than any other regardless of attendance, etc. (other than Boston’s wall, the Metrodome’s hard to see roof, and the altitude in Denver.  BTW, in the Trop it is REAL easy to see a fly ball against the roof of the dome, unlike in Minnesota.  I have NEVER seen a player, home or road, lose a ball in the Trop dome, although I am sure it has probably happened before.  So when you hear the commentators talk about what an extra HFA the Rays have because they play in a dome, just chalk it up to the 10,837 things they say that are wrong.

BJ Upton bunted more often than Jason Bartlett this year.  He also has ALOT more foul bunts, which is probably a sign of someone who is not a good bunter, or someone who is very precise in trying to bunt.

I was really surprised how deep Youk was playing n that situation.  The thing about that is the defense has to “show their hand” first so that it is critical that they understand game theory and play at a position such that it doesn’t matter whether the team bunts or not (same WE).  The offense of course can always take advantage of the defense if they don’t play optimally.  In this case, I think the defense was not playing optimally (because they know how stubborn Maddon is about not bunting) and that Maddon is not smart enough to figure out that he has to make them pay for that mistake (actually it is not a mistake if Francona knows that Maddon will still not bunt and if Maddon does not make him pay).

Surely a player with the speed and athleticims of Upton could teach himself to bunt well in a relatively short period of time, if he doesn’t already.  Of course the fact that Upton is a pretty good hitter (not nearly as good as he did in the post-season of course - which brings up another point - I am sure that Maddon was thinking, “Why would I want to bunt a guy that has 7 HR in the post-season?” - which is another mistake of ignorance) means that the bunt attempt has to have more success than if he were not a good hitter, and another reason why Boston was not playing for the bunt.

I submitted an article to studes on this topic.  It’s a nice little piece…

If you want to read a disgustingly stupid article, which I consider irresponsible, yellow-rag journalism, check this out by a well-known, mainstream sports writer for NBC sports, Mike Celizic:

http://nbcsports.msnbc.com/id/27282889/

Can an otherwise smart person write a truly ignorant column like that?  I think so actually.  Sports (and gambling and politics) really brings out the stupidity in people.

Some people are writing that TB has a pretty good chance of sweeping the series. Do they realize that the odds of them sweeping is greater than 10-1 against?  IOW, little chance of sweeping (7.5% actually).

How about someone predicts TB (or PHI) in 5 games? Greater than 7-1 against for each team separately!

Also, unless one team is a very large fave to win the series (greater than around 2.5 to 1), you HAVE to pick one team or the other in 6 or 7 games.

I wouldn’t go into predicting how many games the series lasts.  I’ll just leave it at Tampa Bay is a 60/40 favorite over the Phillies.

MGL, are you going to stay in Tampa for the series?  I might try to catch one of the games in Philly.

Nah, we all had to get back (dogs, kids, etc.).  We might go back down for games 6 or 7 though (FLO nice this time of year!). Or maybe we’ll go to Philly.

You live near Philly? I have never been to CB park, but my son went this year and said it was real nice.

The Trop is not a bad place to watch a game, for an indoor stadium.  The Kingdome was horrible, and Skydome was pretty bad too.  I have never been to the Metrodome.  Of course, the new stadiums with the retractable roofs are usually nice.

I got the Pay Pal receipt, BTW.  It wasn’t really necessary and I should not have even suggested that we verify the donations, but thanks.

The momentum article:
http://www.hardballtimes.com/main/article/momentum-shmomentum/

I looked for every single game in the last 50 that a home team

Now that’s a small sample!

Nicely written and very clear.  I wonder why you didn’t bring up the notion that without controlling for other things that we discussed in this thread, you were not even sure of the results you got?

IOW, wouldn’t you want to tell the readers that it is possible that once you control for the overall quality of the team, the bullpen, offense, etc., that it is possible that the “momentum” edge could disappear?

Or did you control for all that?

Also, while I am glad that you didn’t dismiss the results, as so many researchers do, because it is not “statistically significant” (you guys know how much I hate that), I think you should have at least mentioned that the fact that the significance was low, even in the combined home and away sample, means that there could fairly easily be little or no momentum effect and the results you got were purely by chance.

(IOW, I think that all researchers, rather than accepting or dismissing a conclusion based on some arbitrary sigma level, should merely tell their audiences how “confident”, statistically speaking of course, they are that the “true results,” if they had an infinite amount of data, are within a certain interval of the results they got.)

Along those lines, isn’t this a misleading statement:

Statistical significance tells us it is much less than that. But, I know you don’t want to hear that.)

Statistical significance does not tell us that the “true” effect is much less than that.  It only tells us how confident we are that the “true results” are within a certain interval centered on our results.  Now, if a priori, the results we get (if they are a deviation from the null hypothesis) were less likely than the null hypothesis, then yes, the significance also tells us how much less the likely “true results” are than the sample ones.  But not in all cases is a deviation from the null hypothesis less likely than the null hypothesis.  In this case, I am not sure which one is more or less likely a priori.

In this case, and often, it depends on the size of that deviation.  For example, let’s say that we think that a momentum effect is MORE likely than no effect, going into the experiment.  If we find a sample effect of .03 or so, like you found, which seems reasonable, the statistical significance does NOT tell us how much smaller the true effect likely is, as you imply with your above statement.  However, if we find a gigantic effect in the sample data, say, .20 wins, then it is true that if that is 2 SD from the null hypothesis then it suggests that the true effect is closer to the .20 than if the significance were only at the .5 SD level.

But, at the risk of repeating myself, if you found a .03 win effect, and you think that is reasonable, and it is 3 SD from the mean, that does NOT mean that the true effect larger than if the significance were at the .5 sigma level.

To this:

However, if we find a gigantic effect in the sample data, say, .20 wins, then it is true that if that is 2 SD from the null hypothesis then it suggests that the true effect is closer to the .20 than if the significance were only at the .5 SD level

I should add this:

and that at the .5 level, the true effect is likely LESS than .20 (if we apply a Bayesian analysis or regress the sample results).

That should have said “years”.  That was my fault.  Funny how you can read something five times and still not pick it up.  But, you and I both know how incredibly long and arduous a process editing is.

I don’t disagree with anything MGL said.  I think the article served its purpose.  You lose your audience if you don’t win them first.  And I think I would have lost them.