Dan Lewis at The ArmchairGM has been crunching some numbers on momentum. Here's the bottom line:
World Series teams down 3-0 have gone a putrid 2-15 (note: in game 4) over the period of my survey. That’s a .117 winning percentage, which over a 162 game season extrapolates to nineteen wins.
There is simply no way that a skill differential or even dumb luck can account for a large portion of this fall-off. The conclusion must be that the adage "'Momentum' is the next day’s starting pitcher" is entirely untrue when down 3-0 in the Series, and quite possibly in the LCS as well. The psychological barrier a team faces is exceptional, and perhaps more than most people realize.
I think that self-belief is a big part of streakiness, and it works both ways. Winning breeds confidence, and losing erodes it.
Post-season baseball is an interesting experiment due to its terminal nature. Suppose you randomly sampled 100 regular season games after a team was swept in a 3 game series. Even though the chances are that such teams are differentially poor, what are the odds they would win as few as 12 out of their next 100? Post-season series are tailor-made for producing streakiness, because the psychological weight is much bigger.
Again, as I've said before, television announcers tend to over-hype such phenomena. Dan's next post wonders if a late 1st half score truly shifts momentum in a football game. It was prompted by the commentator in yesterday's Giants-Vikings game, and is filed intriguingly under "Media Madness." He's offering an opportunity to be an Armchair GM for the best attack on the puzzle.