2017-18 stats and underlying metrics thread [Mod: updated season]

[garret9]

I'm also considering the 2014-15 team.

Very good shot metrics. Quite good goaltending. But when it came to the playoff games, perhaps they lacked the skill to translate into wins. It always felt that they were maxing out on effort and scraping by with the results, whereas more skilled teams seemed to be able to go to a different level.

Obviously, you need the systems, skill, deployment and effort to drive both the shot metrics and the results. I think that's more possible for the Jets in the future than it has been until now.

Evidence needed.

OR they were injured severely and Pavelec went back to being Pavelec for that series (or really even worse than Pavelec).

I mean, you are basing your evidence on wins in a four game sample, where the goalie had a sub 900 save% and yet the team led going into the 3rd (IIRC) in every single game.

 

[Whileee]

What is the purpose of this question?

Would the Jets vs LA with 50% CF% always be like that in shot location? No.
That's the same (and very similar reasons to how the correlations work for in and out of sample) type of straw man that "rather be out scoring than outshooting crowd give".

Reminder, persistence in shot quality in absence of quantity is quite low:

It is, after all, a big reason why Corsi% out performs goals (and scoring chances, and all expected goal models aside from DTM's) in the long run.

I thought the LA game was a very interesting one to watch and analyze. I thought that the Jets completely dominated the Kings for at least two periods, and they were able to generate a lot more high quality plays than the Kings. The Kings, once they got the Jets on the ropes still hammered away with shots from far out, and crashing the net. It was such an interesting difference in style, generating the same number of shot attempts.

I guess the point is that I think that we might be seeing some of the reasons for the random error inherent in shot metrics, and some of it might have a bit of a pattern that is more subtle than our various analytical tools and data. That isn't a criticism of the robustness of shot metrics per se, but just an observation that there are things that we observe in the game that are more difficult to capture through data.

Of course, I could be totally wrong and this is an overly subjective observation. After all, I have cited shot metrics extensively and follow them closely. I'm just wondering where the 2/3 part of the relationship that isn't easily explained might come from.

 

[Whileee]

Evidence needed.

OR they were injured severely and Pavelec went back to being Pavelec for that series (or really even worse than Pavelec).

I mean, you are basing your evidence on wins in a four game sample, where the goalie had a sub 900 save% and yet the team led going into the 3rd (IIRC) in every single game.

I realize it was a small sample, and players were banged up, etc.

I also thought that their run to the playoffs was supported by a pretty unlikely run of goaltending (for any team), and a point total that was nice, but not upper echelon.

I have more confidence in the roster they have now than the one they had in 2014-15, for some reason. It's not because of their statistical performance (as yet). They just seem to have better players that are able to score and win more efficiently, even if they get mediocre goaltending.

 

[surixon]

I realize it was a small sample, and players were banged up, etc.

I also thought that their run to the playoffs was supported by a pretty unlikely run of goaltending (for any team), and a point total that was nice, but not upper echelon.

I have more confidence in the roster they have now than the one they had in 2014-15, for some reason. It's not because of their statistical performance (as yet). They just seem to have better players that are able to score and win more efficiently, even if they get mediocre goaltending.

Plus we have been without a couple of our biggest shot metric drivers for much of the year. This coupled with a terrible schedule imo explains most of the results. While I don't dismiss the models a lot of context is needed when interpreting the Jets results to date.

 

[garret9]

Plus we have been without a couple of our biggest shot metric drivers for much of the year. This coupled with a terrible schedule imo explains most of the results. While I don't dismiss the models a lot of context is needed when interpreting the Jets results to date.

That's not exactly contrary to any of my points though.
:p

 

[ps241]

So today on H&L the boys had Beyak on and the subject came up of "would the Jets protect Lowry in the expansion draft" (yes this will be about stats) and Dennis said yes they would. Dennis said they want to develop him into a face-off taking shut down centre. Then tight turns said the Jets track the "time of possession in the offensive zone" of each player and Lowry makes every player he plays with better at this stat (lets call it the keep away stat).

To those of you who are more astute have you heard of this actual measurement? I always thought of possession having a component of shot generation it not just hoarding?

 

[Aavco Cup]

So today on H&L the boys had Beyak on and the subject came up of "would the Jets protect Lowry in the expansion draft" (yes this will be about stats) and Dennis said yes they would. Dennis said they want to develop him into a face-off taking shut down centre. Then tight turns said the Jets track the "time of possession in the offensive zone" of each player and Lowry makes every player he plays with better at this stat (lets call it the keep away stat).

To those of you who are more astute have you heard of this actual measurement? I always thought of possession having a component of shot generation it not just hoarding?

Teams obviously use more than Corsi. Whether what they are using is effective or not is a mystery though. It will probably remain that way IMO. So much secrecy is the analytics realm.

 

[Gm0ney]

Teams obviously use more than Corsi. Whether what they are using is effective or not is a mystery though. It will probably remain that way IMO. So much secrecy is the analytics realm.

I thought Corsi correlated very closely with offensive zone possession. So closely that it's a bit of a waste of time to track offensive zone time (TOA = Time On Attack in the chart below) when you can just look at the Corsi and get a pretty good idea.

Edit: That chart is from this Pension Plan Puppets article from a few years ago: http://www.pensionplanpuppets.com/2013/9/16/4727746/leafs-attack-time-at-the-halfway-mark

 

[Aavco Cup]

I thought Corsi correlated very closely with offensive zone possession. So closely that it's a bit of a waste of time to track offensive zone time (TOA = Time On Attack in the chart below) when you can just look at the Corsi and get a pretty good idea.

Edit: That chart is from this Pension Plan Puppets article from a few years ago: http://www.pensionplanpuppets.com/2013/9/16/4727746/leafs-attack-time-at-the-halfway-mark

Garret has been discounting the correlation lately on twitter. What he does say is corsi is better at predicting future wins than actual raw posession time. How that relates to individual player corsi vs possesion time is unclear though.

 

[Grind]

Garret has been discounting the correlation lately on twitter. What he does say is corsi is better at predicting future wins than actual raw posession time. How that relates to individual player corsi vs possesion time is unclear though.

Yah this is the important part.

The correlation may be there for TOA but the real purpose is identifying effectiveness at driving wins/goals.

As far as I know there isn't actually any solid proof that TOA is a strong predictor of future wins.

Part of that is inaccessibility of the numbers.

The "observational" argument is that corsi could be correlating/identifying "useful possession" where as straight time of possession doesn't necessarily mean your in a position to generate a scoring opportunity.

You can't get a shot attempt/scoring opportunity if your just mucking around the corners, below the goal line, or constantly under a check.

Anecdotal, but that's where my brain goes.

 

[Gm0ney]

Right - the idea is that Corsi's value is predicting future success. There's not much value in correlation with TOA - but they still correlate strongly. So Garret's been crusading to get people to stop going on about the link between Corsi and possession because WGAF...that's not the important part.

But my point is, you don't really need to manually track a player's time of possession in the offensive zone when Corsi does nearly the same thing.

A lot of the math behind this has either disappeared as stats guys were hired by teams, or been buried by more recent explanatory articles that just assume we all know the correlation - but IIRC guys were manually tracking TOA for long stretches 5+ years ago and they just stopped because there was no point when Corsi gave you the same results.

 

[Coach G]

Yah this is the important part.

The correlation may be there for TOA but the real purpose is identifying effectiveness at driving wins/goals.

As far as I know there isn't actually any solid proof that TOA is a strong predictor of future wins.

Part of that is inaccessibility of the numbers.

The "observational" argument is that corsi could be correlating/identifying "useful possession" where as straight time of possession doesn't necessarily mean your in a position to generate a scoring opportunity.

You can't get a shot attempt/scoring opportunity if your just mucking around the corners, below the goal line, or constantly under a check.

Anecdotal, but that's where my brain goes.

I agree with this, but there is another side to mucking below the opponent's goal line: they're not scoring on you from there.
I was interested in Beyak's comment about o-zone possession, too. I wonder what exactly they're tracking, and what they think it indicates or predicts.

 

[blues10]

I know that it has been discussed that faceoffs themselves have very little direct outcome of a game. I also know it has been discussed how hard it can be to get the puck back when losing possession off the draw.

All this got me thinking of a form of faceoff possesion stat. Although this model differs from the model that I had been thinking of what is the general thought on this stat?

Does faceoff possesion lead to overall puck possesion?

My eye test tells me the Jets do not do well against teams that gain puck possesion off the faceoff against us. I don't know if this is true or just a misconception on my part.

Thoughts.

http://puckbase.com/stats/post-faceoff-possession

 

[Daximus]

I know that it has been discussed that faceoffs themselves have very little direct outcome of a game. I also know it has been discussed how hard it can be to get the puck back when losing possession off the draw.

All this got me thinking of a form of faceoff possesion stat. Although this model differs from the model that I had been thinking of what is the general thought on this stat?

Does faceoff possesion lead to overall puck possesion?

My eye test tells me the Jets do not do well against teams that gain puck possesion off the faceoff against us. I don't know if this is true or just a misconception on my part.

Thoughts.

http://puckbase.com/stats/post-faceoff-possession

We definitely suffer against teams that own us in the dot. Spending the entire game chasing the puck has an effect. How much is hard to say.

 

[csk]

Patrik Laine and Kyle Connor have had some pretty horrid corsi numbers thus far, despite often playing with good players. Is this typical of young high-end rookies? Are they usually better at the end of a full season?

 

[surixon]

Patrik Laine and Kyle Connor have had some pretty horrid corsi numbers thus far, despite often playing with good players. Is this typical of young high-end rookies? Are they usually better at the end of a full season?

Yup, Scheifele was in the low 40's for the first part of his rookie year.

 

[MardyBum]

Gaaaaarrrrret.

What's the best site (iyo) to use for advanced stats that is available to us?

Was debating with a Leafs fan on PP oiSH% and apparently there was a 3.0% gap between hockey reference and hockey analysis for Ehlers.

Any personal preference to which is the best/most accurate available to the public?

 

[garret9]

One is probably 5v4 and the other all PP mins.

 

[Gil Fisher]

Is someone (the League?) subjectively tracking scoring chances or are they just the result of binning and weighting shot location? I think teams are subjectively tracking them, but what does hockeystats and corsica do for their SC numbers?

 

[Grind]

Is someone (the League?) subjectively tracking scoring chances or are they just the result of binning and weighting shot location? I think teams are subjectively tracking them, but what does hockeystats and corsica do for their SC numbers?

That's a good question, I had always thought they were just shots binned from certain positions but had never looked into it...

 

[YWGinYYZ]

That's a good question, I had always thought they were just shots binned from certain positions but had never looked into it...

War on Ice used to define it like this:

(...) based on these measures, the average probability of a goal given the type and locations, and the consideration of team defense, we have these conditions for a “scoring chance”:

• In the low danger zone, unblocked reboundsÂ¥ and rush shotsƚ only
• In the medium danger zone, all unblocked shots.
• In the high danger zone, all shot attempts (since blocked shots taken here may be more representative of more “wide-open nets”, though we don’t know this for sure.)
• Â¥Rebounds are defined as attempts within 3 seconds of a missed, blocked, or saved shot

ƚRush shots are defined as attempts within 4 seconds of any event in the shooting team’s defensive or neutral zones.

In essence the measure is attempting to identify a higher percentage attempt, that has an increased likelihood of resulting in a goal.

More here: http://www.hockeyprospectus.com/burtch-examining-scoring-chance-data/

 

[Joe Hallenback]

I noticed on TSN that they are tracking shots from what they call the homeplate box. Does anyone know if the NHL or some site does this for NHL games?

 

[YWGinYYZ]

I noticed on TSN that they are tracking shots from what they call the homeplate box. Does anyone know if the NHL or some site does this for NHL games?

The NHL tracks the raw shot location data (I can get it through my webscraper(s)), but I don't think that they display the shot data anywhere that's easily accessible in graphic format.

hocketstats.ca plots the shot location data for each game, and will also show 5v5 vs other shot data, and "high danger" shots versus all shots as well.

 

[garret9]

I noticed on TSN that they are tracking shots from what they call the homeplate box. Does anyone know if the NHL or some site does this for NHL games?

NHL tracks shot location for their GC apps. You can scrape for that.

Issue is that looking only at certain shots ends up being worse than including all shots. Adjusting is a far more effective tool, which we use with expected goals.

In terms of predicting future performance at either player or team level, Corsi which views all shots as equal still outperforms any scoring chances or high-danger scoring chances model thus far created.

This is a common phenomenon in stats when binning a continuous variable (like a shot being relatively closer than another is relatively better) into a false categorical variable (shots in home plate vs shots outside).

Some more information on that can be found here:
http://biostat.mc.vanderbilt.edu/wiki/Main/CatContinuous


Aside:
I should point out that adjusting for just location of shots, which Corsica's xGoal model does, isn't sufficient enough to outperform Corsi.
DTM's xGoal model does outperform Corsica's xGoal model AND Corsi. It doesn't just use shot location but other variables as well (such as regressed shooter history).
I also want to point out that Brian MacDonald's xGoal model that was from way back in 2008 IIRC did also outperform Corsi, mostly because it is Corsi but weighting goals more heavily than other shots, and then introduces penalty, hits, and faceoff differentials.

 

[garret9]

Is someone (the League?) subjectively tracking scoring chances or are they just the result of binning and weighting shot location? I think teams are subjectively tracking them, but what does hockeystats and corsica do for their SC numbers?

Everyone is subjectively choosing their own definition for SC.
HS and Corsica are using their own definitions, just like coaches, TSN, etc. are all subjectively defining their own scoring chance.