I’ve read that before.

The problem is that while overtime taken by itself itself is a crapshoot, it doesn’t follow that you should throw out overtime data from the overall data set on that basis. Overtime alone isn’t predictive and is very random, but it’s because it is so short. The same could be said of any short stretch of regulation play, too — for example, the result of the first five minutes of regulation are liable to be pretty random, or the 12th to 17th minutes of the second period, for that matter. Are the last five minutes of regulation in a tie game very different from overtime? Should we discount goals scored in these circumstances, too?

The difference is that overtime stands out more because it happens more occasionally, and whoever wins overtime always wins the game, which is not the case for any given stretch of regulation time. That lends to overtime an intuitive importance that it doesn’t deserve. It’s very nature as the conditional last stretch of a hockey game makes it stand out in a way it really shouldn’t. I just don’t see a basis for not treating it as just more sample size, except for the fact that its randomness as a short stretch of hockey stands out more to us human observers.

Naturally, throwing out any specific portion of a data set will tend to skew the results against any team that is favored by the selected portion. So throwing out overtime data on the basis that “OT is a crapshoot” will make any team with an unusually good overtime record (ie. the Habs) look much worse, where it’s not necessarily deserved.

To call that the Habs are having a ‘shockingly lucky’ season, one must look on the season as a whole. Gabe mentions their early luck with overtime and their recent stretch of favorable percentages, but doesn’t that simply cancel out the opposite tendancies from those stretches? For example, the Habs have, over the course of a season and discounting empty-netters, a 8.87% shooting percentage and a .911 save percentage. That doesn’t seem “shockingly lucky” to me over the course of a season. In fact it seems perfectly average.

Of course, this was brought up to this level by a recent stretch of very good percentages while being badly outshot… which implies that they were having poor percentages despite shooting more earlier, back when they accumulated all those overtime wins (which was indeed the case). Their bad luck in regulation was masked by their overtime luck, as they lost games in regulation that should have gone to overtime and won overtime games that should have been regulation wins. Bad luck one way, good luck the other, but the overtime record was more unusual and so stood out more.

The Habs are not a good team right now, I think the standings speak for themselves, but to call them lucky on top of that, apparently solely on the basis of overtime record, is unwarranted. Their luck and their bad luck on the ice seem to have balanced out mostly, so they’re more or less where they should be (and in the age of OTL points, being “.500” is nothing to write home about). Given that, the only element of luck that remains is all those injuries, and I don’t think one can claim the Habs have been lucky there.

And why does this matter? They were outshooting, and their shooting/save percentages were below league average. As a result they ended up losing games they “should” have tied, and bringing games to OT that they “should” have won in regulation. Frankly, that they outshot all these teams and came out of this stretch roughly .500 seems like bad luck to me, and if they had lost half their OT games, we might be talking about how “unlucky” they had been instead. As you said, it’s not an impressive spread, and Montreal lost several games despite outplaying/outshooting during that stretch.

You need to present compelling evidence that the Habs have been lucky across the board to call them “shockingly lucky”. Your entire argument is based on their ‘lucky’ overtime record.

Why don’t we just forego the games and let Tom tell us who will win.