Ya, pretty interesting stuff. Corey is a workhorse. He watched all those games and tracked it all by himself. Do keep in mind this limits the sample size to make things 100% definitive, but the trends are still true.
I'm hoping to track all the exits for the Jets games this season. Corey was the only one who consistently tracked exits last season, but I think this season we have about 20 teams covered.
I've always liked Clitsome. Looking forward to see how the other Jets stack up this year.
Keep up the good work stats guys. Your dedication to the story beneath the story is admirable.
Corey gave me some of the data he has from the games he tracked with the Jets:
His sample is very limited as he only got so many Jets games... He also limited his defensemen research to those he had at least 100 touches with.
We'll get a really good look at this later this season with me doing the rest (with some help from a few other HFBoard members)
Some have gone on about Scheifele's FO% as a detraction on him. Now keep in mind I am NOT saying faceoffs do not affect the game... cos they do indeed. However, media tends to over exalt their power.
I decided to pull out some old Gabriel Desjardins stuff. Here is how we see the influence in shots after a offensive zone faceoff, whether that be win or loss and both PP and ES. As you can see there is a jump in shot rates, although small, even if you lose an offensive zone faceoff, just from the natural advantage of being in the offensive zone.
You can subtract the increase from a win by the increase from a loss to remove the natural zone start advantage, which gives you this:
The area under the curve then is representative of how many shots you gain on average from winning a faceoff. We can then use league average shooting percentages to estimate the amount of goals.
* it takes 41 ES OZ faceoff wins to equal a goal on average
* likewise then it takes 41 ES DZ faceoff wins to equal a prevention of a goal on average
* looking at how avg goal +/- equals a win: 245 zone face off win +/- on average equals one win in the standings
PP is different than ES as the sh% for and against isn't equal (due to advantage of having an extra player and getting more prime scoring chances):
* it takes 27 PP faceoff wins to equal a goal on average
* again looking at avg goal +/- to equal a win: 164 special teams faceoff win +/- equals a win in the standings on average
But then how many faceoffs does each team take?
* teams on average take 790 special teams zone faceoffs, so the chances of being 164 faceoff +/- is very slim
* teams on the ES take about 2200 ES zone faceoffs, so it's realistic to get a single win from the value of faceoffs alone over the span of your whole team
So when you look at a guy who takes 400 faceoffs (about 1/4 of the teams) and absolutely sucks at about 40%, he's costing the team about -50 in the FO +/-
For neutral zone it's about 657 FO +/- to equal a win
Moneyball wasn't really about stats but more about making the best out of your resources while giving yourself a market advantage. Problem with moneyball is your advantage is no longer an advantage when everyone knows about it.
Tyler in the article above shows how TOR could have saved themselves from the Komisarek buyout with adv stats.
It's been shown before how GMs tend to make choices that align with PDO (just on-ice sv% and sh% combined), which isn't controlled much at all by player talent...
Schiefele and Setoguchi have made Kane a good PP player IMO. No stats to back it up, but it looks to me like they've got more PP time than LLW. Also appears that Noel is using both PP forward lines with both PP D pairs, not set 5 man units. Working so far.
A lot of people have difficulty believing that shot differentials can be better determiner of future goal differentials or win percentage than GD or win% do for themselves.
We don't do much in research on shot metrics vs GD or win% anymore as it's mostly been proven long ago. Now we mostly work on improving shot metrics vs themselves.
Problem is a lot of that proving happened long ago around 2004-2007 era, prior to hockey blogs or advance stats being popular, so most people don't realize the work that went on before.
Most of the great work was done by Desjardins (arcticicehockey.com before it was Jets site), JLikens (objectivenhl.blogspot.ca/) and Ferrarai (vhockey.blogspot.ca/).
I've been doing some re-reading of this old stuff for fun because my workload in the lab has gone down a bit.
Going to resurrect a bit of the stuff here:
Here is a really good data from JLikens when he was trying to determine which is more predictive of future success:
First we'll show how predictive a particular stat is of future occurrences from that same stat:
This is basically looking at 1 random set of 40 games for a team, comparing it to another random set (allowing for overlap) in the same season, and looking at the correlation... and repeating it 1000 times.
It makes sense, the more common variable (Corsi) is the one more reliable at predicting itself, while the least common variable (goals) is the one least reliable at predicting itself.
Then we can determine how the two correlate to goals using the same method above:
I just accidentally deleted the rest of the post... crap. Will fix on later date.
Interesting to see theta Scheifele is already middle of the road in terms of matchup against both D and F. I would have thought he'd be more sheltered than that, but I guess that's what happens when you're on the 2nd/1st lines. The Frolik usage surprises me a bit.
So Mirtle tweeted something about the Leafs being among the bottom feeders in shots by defencemen and that was contirbuting to their poor Corsi.
That leads to the question as to whether Corsi and Fenwick close by forwards rather than by the team overall would be a better predictor of success? Is this tracked anywhere. (I am making an assumption that sots by forwards have a greater probability of enterring the net than shots by defencemen.)
I'd imagine the Jets are one of the better teams of generating shots by defencemen, though with the injuries, maybe not. There's also the question of whether a shot by a defenceman leads to a shot by a forward.