Puck Daddy just posted a pretty sizable piece on the Wild and "regressing to the mean." We get a few indirect mentions.
What's really interesting is that, among some entirely valid statistics, there are some pretty glaring flaws. Specifically, they repeat the notion of "regressing to the mean," which suggests that we're dealing with dependent events––that the outcome of one game directly affects the next. If they're independent events, which are usually used in probability, then the odds of winning any game is always about the same, regardless of how many games we may have won in unlikely fashion before it. That means we should have gone about .450 (or whatever our Fenwick suggested) in the span after our winning streak--not that the team would lose enough games to fall to .450 in the span of the next 30 games.
It's frustrating that they use the losing streak to validate their prediction that the Wild's success wouldn't last; in fact, probabilistically this losing streak is just as unlikely as a winning streak.
Also, there's no mention of the inconvenient fact that our losing streak coincided with an improvement in our Fenwick and Corsi.
Last edited by llamapalooza: 03-01-2012 at 04:57 PM.
Puck Daddy just posted a pretty sizable piece on the Wild and "regressing to the mean." We get a few indirect mentions.
What's really interesting is that, among some entirely valid statistics, there are some pretty glaring flaws. Specifically, they repeat the notion of "regressing to the mean," which suggests that we're dealing with dependent events––that the outcome of one game directly affects the next. If they're independent events, which are usually used in probability, then the odds of winning any game is always about the same, regardless of how many games we may have won in unlikely fashion before it. That means we should have gone about .450 (or whatever our Fenwick suggested) in the span after our winning streak--not that the team would lose enough games to fall to .450 in the span of the next 30 games.
It's frustrating that they use the losing streak to validate their prediction that the Wild's success wouldn't last; in fact, probabilistically this losing streak is just as unlikely as a winning streak.
Also, there's no mention of the inconvenient fact that our losing streak coincided with an improvement in our Fenwick and Corsi.
Not only that, but 30 games is NOT a relatively small sample size over an 82 game season. I believe the Wild will be the first team in NHL history to not make the playoffs after leading the league so late in the season.
Not only that, but 30 games is NOT a relatively small sample size over an 82 game season. I believe the Wild will be the first team in NHL history to not make the playoffs after leading the league so late in the season.
So, Hackett or Harding tomorrow? Harding had over a full game tonight (60:30 TOI) so I'd be inclined to go with Hackett. Of course, I would have been going with Hackett regardless so that might not be a fair assessment.
Gotta be Harding tomorrow only because of his record vs. Detroit. After that, it's Hackett time! I really want to see how this kid can handle some pressure.
