THE BOOK--Playing The Percentages In Baseball

"The game probabilities were then used to simulate the games in question a total of 100,000 times, more than enough to get an accurate distribution of how many games the team with ‘momentum’ would be expected to win. The median of the simulation showed that the team with ‘momentum’ should have won 1157 of the games while in reality they won 1167. However, the simulation showed this difference was not nearly enough to be statistically significant at the .05 or even .1 level. Therefore, we cannot reject the null hypothesis that momentum does not exist in baseball showing yet another example where momentum in sports is most likely non-existent.”

I agree with everything but the italicized portion (my emphasis).  Isn’t it most likely that momentum is like clutch, which exists in very small denominations at the pro level?

It seems reasonable to me that the true impact of momentum is 10 games in 1167.  And in fact, assuming p < .50, this study shows it’s most likely closer to 10 games in 1167 than it is to 0 games in 1167.

Not if your prior is 0 to begin with.

Otherwise, you can say the same thing if you flip a coin 100 times and you get 56 heads.

He goes too far to suggest momentum doesn’t exist in other sports.  Lots of other sports are more fluid, and have fewer breaks/interruptions than baseball, such that fatigue in the near term can set in (as on a long drive in football, where the offense generally is more able to dictate play and exploit the defense’s exhaustion), or fatigue in the long term can set in (as in extra time in the other football, where a team with superior fitness and/or judicious use of subs can generate a big advantage sometimes).

If he defines momentum strictly as something like “we tied it up, now we just can’t lose!”, then I would probably be inclined to agree, but fatigue is real, and it does not always affect both teams equally…

I am not clear on their methodology, but the teams that came back in the 9th likely had slightly better than average offense and the teams that blew the lead in the 9th likely had a pitcher who was slightly below average (compared to the average 9th inning pitcher I guess). Both of these things might be enough to account for the 10 win difference.

(The pitcher thing would only apply if the 9th inning pitcher sometimes pitched the 10th, which I assume is the case.)

Also, and perhaps most importantly, when you blow a lead in the 9th you likely have used up your closer whereas the other team likely still has his available.  I don’t know if the authors took this into consideration. This is especially true if you are only looking at home teams who blew a 9th inning lead. In extra innings the road team usually brings in their closer if they take the lead.  The home team, in this case, has no closer to use at all, assuming they used him in the 9th.

"momentum” is the granddaddy of stupid sports cliches announcers fall back on when they have nothing interesting to say.  I’d love it if it didn’t exist.  And I suspect there’s very little of it at the highest levels.  The assumption is that the game situation affects the players’ level of concentration and effort.  I say baloney.

That’s why I have the comic on the right side of every page, perhaps the greatest illustration ever.

You know, I don’t think that cartoon depiction is completely accurate.  I think the word it was really meaning to use was:

pseu·do·ran·dom adj \-ˈran-dəm\

: being or involving entities (as numbers) that are selected by a definite computational process but that satisfy one or more standard tests for statistical randomness

Perhaps substitute the word “baseball” where it reads “computational”.

...teams that came back in the 9th likely had slightly better than average offense and the teams that blew the lead in the 9th likely had a pitcher who was slightly below average…

I suppose there’s also an opposite effect working against the team “with momentum” that scored in the 9th:

That team was more likely to have the strongest part of its lineup batting in the 9th, which means they are likely to have relatively weaker hitters batting in the 10th.  (Not sure if the simulations account for these effects, not that it would make a huge difference...)

"That team was more likely to have the strongest part of its lineup batting in the 9th, which means they are likely to have relatively weaker hitters batting in the 10th.”

Yeah, could be.

I think the strongest effect is going to be the closer thing.  The team that came back likely still has its closer available and the team that blew the lead likely does not.  That is a big thing…

I agree with #1 that there’s no way to conclude that “momentum is likely non-existent”. All we can conclude is that the effect is at most very small.

Honestly, you don’t really “conclude” anything (unless you are forced to) when dealing with sample stats.  You simply result in a certain estimated likelihood that one or more hypotheses are true and even then it is usually not an either/or.  It is a matter of degree…

Greg/7:  No, the comic is right.  Sports (especially baseball) is a “weighted random number generator” in the sense that what happens in a given moment is random but weighted by the individual skills that the players have.  Some things are more likely to happen when Albert Pujols is batting than when, say, Juan Uribe is batting.  Of course, on any given at-bat, Albert Pujols could strike out and on another given at-bat, Juan Uribe could hit the winning home run.  Hence, the weighted random number generator.

A pseudorandom number generator is one where if you knew the algorithm and its initial seed, then you could predict with 100% accuracy what will occur despite the fact that to an outside observer who does not know the algorithm that the events appear random.  This is not the case with baseball.