Avalanche Finally Tumble

The best part is… I DO watch the games!

I think I understand your question and to answer (hopefully), I don’t know of evidence to suggest a general pattern in either direction (better to play teams that end up with good records early v. late in the season). In general, the larger sample will be the most predictive though at some point “the past is the past” so to speak. I was more just ball-parking with the assumption that most bad teams would continue to be bad and most good teams would continue to be good over the last 30% of the season. In some instances this might not be the case (injuries, luck, legitimate improvement or regression…). It was surely better to play the Oilers, for example, after Khabibulin and Hemsky were injured than it was to play them before that. They were bad either way, but the degree was surely different.

The question becomes degree. From what I can tell your instincts are right here but that the degree to which it’s impacted their overall results isn’t overly large. At least, not if you consider “score tied” results a good way to factor out “playing to the score” as a variable.

Again, with the score tied, they get hammered. It has nothing to do with the lead. When the score is tied, they are worse by the shots data.

dude, you are smoking something. here’s what I wrote about the 2009-10 Avs:

http://www.behindthenethockey.com/2009/10/29/1106363/colorados-hot-start

That’s a pretty detached piece, talking about whether they were a 90-point team (like the bookies thought) or if perhaps they were a 98-point team. I didn’t touch on shot counts at all (mostly because they’re not very meaningful during a stretch where you win all your games.)

Here I talked about how Kyle Quincey was getting shelled:

http://www.behindthenethockey.com/2009/11/3/1108617/kyle-quincey-1-defenseman

And then four months later, I talked about how they had an historically high save percentage and shooting percentage:

http://www.behindthenethockey.com/2010/3/16/1370883/colorado-avalanche-worried-yet

I looked into why that was and concluded that they were a playoff bubble team and probably wouldn’t win a playoff series.

There’s a lot of luck in hockey. And sometimes all of that luck goes a team’s way. It’s my job (literally, it is my job) to determine how much is luck and how much is true talent. If you, as a fan, find it enjoyable to believe that outcomes match talent, that’s your business. But I don’t see how we gain any understanding of the game if we ignore luck.

Take yourself out of the fan’s shoes and put yourself in a general manager’s shoes. His team has been winning, but they’ve been getting shelled. Do you think he’s sitting there telling himself how great it is to be winning? Or do you think he’s worrying about how long it’s going to take for his team’s bad play to catch up with him?

I can tell you it’s the latter. Hockey guys look at turnovers, breakouts, forechecks, shots – all factors related to puck possession. And when their teams do poorly on those counts, they let them know – regardless of the score.

I think you’re unclear on what causality is. If a team controls the puck and controls the shot board, we should expect them to win. Winning primarily comes from having the puck. That’s a causal relationship.

There is also a causal relationship between high shooting percentages and winning, but high shooting percentages are luck-driven (as I pointed out to you, there’s no correlation between even- and odd-numbered games) while shot counts are driven by talent.

You’re really projecting onto me here with this notion of “worship.” I think Corsi explains about half of winning, and the rest is luck and data that we don’t capture. And I’ve taken great pains to show my work publicly. On the other hand, you are presenting no evidence for your alternative analysis of the game. But who needs evidence?

Good thing NHL teams use Corsi and count scoring chances. Maybe you should take your argument to coaches and general managers and see what they say.