Player Discussion Simon Benoit discussion.

[MCR74]

Me personally I’d take Benoit over Holl

that’s just me though, everyone is entitled to a opinion.

So would Detroit.

 

[banks]

Other teams are already overplaying their top guys in the playoffs. FLA played Seth Jones like 27 minutes in one game without OT.

Here, meanwhile, because we can trust 3rd pair guys like Benoit with 20 minutes, our top defencemen won't wear out on a deep playoff run.

 

[Zack47]

lol. Please, he was bad last playoffs and through the first 60-odd games of the season. He's had a passable two-week stretch of playing adequate hockey (buoyed, as usual, by stellar goaltending from Stolarz and some puck luck). Great, I hope he keeps it up. It's funny that you're bringing up stats teacher chops only now when it's convenient though, and dismissing larger samples of largely non-subjective stats.

To that point, the panel barely had anything to say about Benoit when he was playing poorly (again, most of the season) and now gurus like Bieksa are going on about how he's his favourite player ever. And for all that they're about ~just watching the game~, their reads are nonsense. They were calling the pass interception that Benoit made prior to setting up Domi's OT winner in game 2 a "risky read" when it was a 3-on-3 and Benoit was in good position. They didn't even mention that it was the drop pass to Domi that was the risky play - Cozens barely missed poking it away from Domi to set up a rush the other way. It's a game of inches and if you choose to only remember the good bounces in forming your opinions of a player, so be it, just don't pretend there's any principle behind it.

I have to respond. I have a major problem with bivariate statistics in sports, for three reasons: 1) we don't know their standard error, so we can't judge differences with other players, and small differences are cited over and over; 2) we don't know their predictive value -- these are descriptive stats only and they are not used to predIct differences in outcomes; and 3) unless someone takes a multivariate approach, to find out which individual stats matter and which don't, I am not impressed.

In the meantime, cut your "lol, please" snark.

 

[TMLBlueandWhite]

I have to respond. I have a major problem with bivariate statistics in sports, for three reasons: 1) we don't know their standard error, so we can't judge differences with other players, and small differences are cited over and over; 2) we don't know their predictive value -- these are descriptive stats only and they are not used to predIct differences in outcomes; and 3) unless someone takes a multivariate approach, to find out which individual stats matter and which don't, I am not impressed.

In the meantime, cut your "lol, please" snark.

Yeah but the VAST majority of stats in hockey are simple counting stats.

If you're looking for multi-variate statistics with predictive value then you're talking about xG models and WAR. Most xG models are fairly advanced running regressions and correlations across multi-variable inputs and then aggregating the results. WAR models do something similar for evaluating players.

You can find a bunch of stuff on hockey-graphs and scattered throughout the internet.