Fancy stats say...

[Replacement*]

In terms of even strength puck possession, you can tell this team on ice is vastly better than last year... maybe only the Kings game and short handed against Canucks were the Oilers hemmed in their own zone for long periods. I really like what I saw in the preseason with the forwards winning board puck battles.

Bad goaltending and giving up too many grade A scoring chances are the big problems right now. Neither are easy to fix

But the assumption by analytics is that possession is a proxy for shots, which is a proxy for goals, which is a proxy for W's.

Analytics infers such things as "all other things being equal" for instance that all shots are equal and that teams don't exhibit the breakdowns the Oilers do. That we give up such dangerous scoring chances in every game confounds the stats.

If a team cheats for offense like the oilers do and pinch without discretion that will of course have other impacts. No statistical sampling can account for what is essentially irrational play.

Its like betting on a grand slam winner against an unknown and the Champion makes 50 unforced errors in the match. It just throws any other tabulation out the window. The Oilers have done this in 4/4 games.

 

[oobga]

But the assumption by analytics is that possession is a proxy for shots, which is a proxy for goals, which is a proxy for W's.

Analytics infers such things as "all other things being equal" for instance that all shots are equal and that teams don't exhibit the breakdowns the Oilers do. That we give up such dangerous scoring chances in every game confounds the stats.

If a team cheats for offense like the oilers do and pinch without discretion that will of course have other impacts. No statistical sampling can account for what is irrational play.

There is definitely a correlation between good corsi stats and making playoffs, but there are always exceptions. If we've learned anything about the Oilers over the years it's that we seem to do our best to always be the exception... in a bad way.

Like you say, there's no fancy stat that accounts for stupidity. A horrible bonehead play that leaves a guy wide open infront of your net for an easy goal only counts as 1 corsi against. Smyth (I think Hendricks is our new guy for this) could cancel that out with a muffin from the blueline that has 0% chance to go in. And if we now have Dellow convincing Eakins to pressure guys to just wildly take any shots they can for more corsi's then we're definitely screwed.

 

[Moose Coleman]

In terms of even strength puck possession, you can tell this team on ice is vastly better than last year... maybe only the Kings game and short handed against Canucks were the Oilers hemmed in their own zone for long periods. I really like what I saw in the preseason with the forwards winning board puck battles.

Bad goaltending and giving up too many grade A scoring chances are the big problems right now. Neither are easy to fix

I think going with a conservative, collapsing man-to-man system instead of whatever the hell they are playing now would go along way to fixing both these issues.

 

[Supermassive]

Isn't this just like the start of last year? All the analytics guys were praising Eakins for the good corsi numbers and all the blame was going to Dubs (and eventually the swarm) for why we weren't actually winning. Then the wheels eventually fell off, especially when we had to go on long runs of playing good teams.

Yup. It's funny how some people fall for the same bull, year after year. Blame the goaltending...wait, blame the defenders...wait, blame the coaching.

It's management, trading any veteran Oilers away for magic beans, and bringing in worse vets who get disillusioned (or is it reality?) early and give up and want out, while bringing up legit prospects too early and without proper mentoring. So terrible.

 

[Replacement*]

Yup. It's funny how some people fall for the same bull, year after year. Blame the goaltending...wait, blame the defenders...wait, blame the coaching.

It's management, trading any veteran Oilers away for magic beans, and bringing in worse vets who get disillusioned (or is it reality?) early and give up and want out, while bringing up legit prospects too early and without proper mentoring. So terrible.

As bad as people say Gilbert, Gagner and Hemsky were they could at least be interpreted as players that cared about the team and org, if not the fans.

How is it automatically inferred that a newcomer to a gongshow is going to be a better investment. If anything one would think the newcomer would book out quicker having little investment with the carnage and considering other options..

Only a fool of a team gives up on so many of its ongoing players and then wonders what happened to any notion of chemistry, identity, values, on the team.

Perron, my favorite player from last year is already looking like somebody that would like to be a deadline trade.

But this is the org that said one year that Stoll would be future captain material and who showed consummate leadership and then traded him the next.. With Stoll now being the quintessential LA leader and a guy who will perform any role gladly.

 

[MoLarr]

There is definitely a correlation between good corsi stats and making playoffs.

Correlative analysis is hugely different than predictive value. Because hockey is such a fluid game with the continued input and influence of everyone else on the field of play (ice) at the same time, the creation of an individualistic measurement is impossible.

What's the goal? To be like baseball where the Pythagorean expectation can be calculated with shocking and surprising accuracy? Or retrospective analysis for in/off season improvement?

 

[doulos]

Correlative analysis is hugely different than predictive value. Because hockey is such a fluid game with the continued input and influence of everyone else on the field of play (ice) at the same time, the creation of an individualistic measurement is impossible.

What's the goal? To be like baseball where the Pythagorean expectation can be calculated with shocking and surprising accuracy? Or retrospective analysis for in/off season improvement?

Holy big words Batman!

 

[Jet Walters]

Our possession numbers are skewed greatly due to Justin Schultz standing behind the net with the puck for 5 minutes a game.

 

[Eirhead*]

Shots from the sideboards != shots from 5 feet deep in the slot

The 2005-06 Oilers were getting outshot like crazy in the playoffs. But the shot quality was largely very poor.

 

[MoLarr]

Holy big words Batman!

I've been listening to MacTavish pontificate.

 

[oobga]

Correlative analysis is hugely different than predictive value. Because hockey is such a fluid game with the continued input and influence of everyone else on the field of play (ice) at the same time, the creation of an individualistic measurement is impossible.

What's the goal? To be like baseball where the Pythagorean expectation can be calculated with shocking and surprising accuracy? Or retrospective analysis for in/off season improvement?

I would say retrospective analysis, mainly for player evaluation. I think you have to dig at least as deep as with/without zone start adjusted stats to really get value out of corsi to figure out which guys drive the bus on their teams and maybe you find the odd gem that no one else sees that you can try to acquire.

I'm not sure how you can really use corsi to change the way you play. If you're consciously keeping corsi in mind when you play, isn't the best strategy to just throw the puck at the net any chance you get no matter the situation and try to get off the ice before the puck goes back to your end? I guess a coach can make adjustments and try to look at corsi stat changes to see if they had a positive impact, but you would need big sample sizes and it's hard to take into account all the other factors like who you're playing against to really figure out if your adjustment had an impact.

 

[Horseradish]

Few points:

(1) What is their Corsi Close (or Fenwick Close)? I.e., how much of that is score effects. "Wow we played great against the Kings once it was 6-0."

(2) How much of Eakins' systems do we blame for their league-low 89 PDO (terrible shooting %age and goalie save %age)?

I can't remember where I saw it-- one of the major Oilers blogs over the past 2-3 days, I think. But their Fenwick close is suggestive of a very good team.

As for your second question, that's my concern. Our skaters may be able to play at a higher level than the opposition, but if the system and (arguably) relatedly, our goaltending, can't keep pucks out, it's a moot point.

 

[MoLarr]

I guess the long term analysis will be a lottery pick again.

 

[nexttothemoon]

I can refute the "fancy stats" and all the advanced statistics with one obvious one-liner that sums it all up...

Shots aren't goals.


Get back to simply analyzing goals for and against and what causes each to happen... that's the REAL advanced stats. Everything else is tangential and just obfuscates (my big word of the day) the real factors affecting winning and losing out there on the ice.

 

[McspOiler]

I believesuring staffers show yesterday we are 5 th in fenwick in the league a team with that level has never missed the playoffs. Problem is short sample size

 

[Panda Bear]

Corsi would be so much better if it looked at where the shots came from.

According to Corsi stats the great Russian terms were a bit **** because they kept passing the puck until a high-percentage chance arose.

Would the Oilers be dead last for shots in the slot Corsi?

 

[doulos]

I can refute the "fancy stats" and all the advanced statistics with one obvious one-liner that sums it all up...

Shots aren't goals.


Get back to simply analyzing goals for and against and what causes each to happen... that's the REAL advanced stats. Everything else is tangential and just obfuscates (my big word of the day) the real factors affecting winning and losing out there on the ice.

I disagree that that refutes any of it.

 

[Panda Bear]

"You miss 100% of the shots you don't take."

Corsi comes from Wayne Gretzky, the greatest player to ever play the game.

#obvs

 

[oobga]

Corsi would be so much better if it looked at where the shots came from.

According to Corsi stats the great Russian terms were a bit **** because they kept passing the puck until a high-percentage chance arose.

Would the Oilers be dead last for shots in the slot Corsi?

Speak of the devil. I thought Michael Parkatti was sick enough of the Oilers to not do analysis of games this year, but he just posted stats for the Canucks game.

http://www.boysonthebus.com/2014/10/13/game-stats-oilerscanucks-oct-11/

He does an analysis that takes shot distance from the net into account. I hope he does more of these. In this canucks game, accounting for shot distances we did pretty badly with goals expected 36.9%, worse than the already poor corsi of 42.5%.

 

[Horseradish]

I disagree that that refutes any of it.

No kidding. There is a HUGE correlation between Corsi and/or Fenwick and playoff teams-- especially when looking at consistent playoff teams (there are always anomalies like Toronto two years ago, or Colorado last year that go against the model, but they almost never sustain).

Are Corsi and Fenwick perfect? No. But their logic is very simple as well as being tried, tested, and most importantly, TRUE. Possession is very important in winning hockey games. Can anybody debate that logic? I doubt it. Well, Corsi and Fenwick are just imperfect statistical representations of that simple truism.

 

[Panda Bear]

Speak of the devil. I thought Michael Parkatti was sick enough of the Oilers to not do analysis of games this year, but he just posted stats for the Canucks game.

http://www.boysonthebus.com/2014/10/13/game-stats-oilerscanucks-oct-11/

He does an analysis that takes shot distance from the net into account. I hope he does more of these. In this canucks game, accounting for shot distances we did pretty badly with goals expected 36.9%, worse than the already poor corsi of 42.5%.

Holy crap.

THIS. THIS THIS THIS. Give up a hundred shots at 25+ if you give up no shots closer than that.

 

[Horseradish]

Speak of the devil. I thought Michael Parkatti was sick enough of the Oilers to not do analysis of games this year, but he just posted stats for the Canucks game.

http://www.boysonthebus.com/2014/10/13/game-stats-oilerscanucks-oct-11/

He does an analysis that takes shot distance from the net into account. I hope he does more of these. In this canucks game, accounting for shot distances we did pretty badly with goals expected 36.9%, worse than the already poor corsi of 42.5%.

I wish he would have looked at ES vs. Special Teams. I think it would be a lot more even.

 

[CornKicker]

i checked the advanced stats and we are still 0-4

 

[oobga]

I wish he would have looked at ES vs. Special Teams. I think it would be a lot more even.

I think he only works with the ES stats. Didn't we get hammered pretty badly on the PK that night? I would expect the numbers would be worse including special teams.

 

[nexttothemoon]

No kidding. There is a HUGE correlation between Corsi and/or Fenwick and playoff teams-- especially when looking at consistent playoff teams (there are always anomalies like Toronto two years ago, or Colorado last year that go against the model, but they almost never sustain).

Are Corsi and Fenwick perfect? No. But their logic is very simple as well as being tried, tested, and most importantly, TRUE. Possession is very important in winning hockey games. Can anybody debate that logic? I doubt it. Well, Corsi and Fenwick are just imperfect statistical representations of that simple truism.

How about doing a correlation of goals for/against and especially 5on5 goals for/against and seeing if that has a higher correlation to a team being a playoff team or not.

Corsi and Fenwick... (which are just a fancy way of adding in missed/blocked shots)... are yet another level of abstraction away from what actually counts... goals for and against.

Even goals for/against can be a slight abstraction... if a team wins 9 games by a goal each and then loses the 10th by 9 goals... they'll be even in goal differential yet have only 1 loss in 10 games. Obviously in that case battling hard and winning 9 games means more than the 1 blowout loss.

The point remains that goals are a better indicator of "goodness" than some fancy stat that takes into account blocked shots into someones skate and misses that end up in someones beer in the stands.

Better yet... actually watch the games and see what actually CAUSES every goal for & against... and then you'll know the strengths and weaknesses of the team much better than looking at how many shots were fired 20 feet high and wide of the opponents net.