The Advanced Stats Thread

[Mr. Canucklehead]

By popular demand. This thread can serve a few purposes - an introduction to advanced stats, discussion and debate on their value and place in today's NHL, and so on. I'll update the OP with relevant info as the thread goes on.

Click to read more...

 

[WTG]

Good resources for advanced stats
http://stats.hockeyanalysis.com/
- for general stats

http://www.behindthenet.ca/nhl_stat...+3+5+4+6+7+8+13+14+29+30+32+33+34+45+46+63+67
- for QoC stuff

http://chlstats.pythonanywhere.com/
-CHL advanced stats

http://hockeystats.ca/
-game by game stats


What is Corsi/Fenwick TL;DR

(CF)Corsi for: Shot attempts for (Shots missed + Shots Blocked + Shots on net)
(CA)Corsi against: Shot attempts against (Shots missed + Shots blocked + Shots on net)
(CF%)Corsi for percentage: CF/(CF+CA)*100%
(FF)Fenwick for: Shots on net + Shots missed for
(FA)Fewnwick agaisnt: Shots on net + shots missed against
(FF%)Fenwick for percentage: FF/(FF+FA)*100%

 

[PM]

Woo hoo! Fancy stats time! Someone posted this article the other day (CarolinaCanuck?) on fancy stats and I thought it was a good read so here it is:

http://hockeyanalytics.com/2008/01/the-ten-laws-of-hockey-analytics/

 

[bossram]

war-on-ice is THE best advanced stat site

 

[Fat Tony]

What is Corsi/Fenwick TL;DR

(CF)Corsi for: Shot attempts for (Shots missed + Shots Blocked + Shots on net)
(CA)Corsi against: Shot attempts against (Shots missed + Shots blocked + Shots on net)
(CF%)Corsi for percentage: Shot attempts for - shot attempts against
(FF)Fenwick for: Shots on net + Shots missed for
(FA)Fewnwick agaisnt: Shots on net + shots missed against
(FF%)Fenwick for percentage: (Shots on net for + Shots missed for) - (shots on net against + Shots missed against)

CF% should be - CF/(CF+CA)*100%
FF% should be - FF/(FF+FA)*100%

 

[WTG]

CF% should be - CF/(CF+CA)*100%
FF% should be - FF/(FF+FA)*100%

**** your right.

 

[Bleach Clean]

Couple of sites (and particular articles) to add to this thread:

http://www.nucksmisconduct.com/2015/5/6/8549779/canucks-player-autopsy-nick-bonino

Let’s be honest, he’s not a true second line Center! Sir Trevor & associates anointed Nick as their 2nd line Center immediately after his trade to the team… Was it because they expected him to be better than his previous season or was it because there wasn’t a better option?

http://canucksarmy.com/2012/8/19/toiqoc-canucks

There are metrics that do a reasonably good job of measuring offensive contributions, GVT and Vukota for example, and it has been proven that better players drive on-ice shooting percentage. Still, many of the existing metrics we rely on arguably fall short of capturing the full value elite offensive players bring to their respective clubs.

http://www.broadstreethockey.com/2011/3/15/2046512/simplify-scoring-drop-secondary-assists

So now you're calibrated -- something with a correlation coefficient (R2) of 0.22 fluctuates about as much as goal scoring does. A correlation of 0.22 actually means quite a bit of fluctuation (just ask Alex Ovechkin), but I think we all believe that scoring goals is a talent, so this tells us that we should consider a correlation near or above 0.22 to show that something is a result of a player's talent.

http://nhlnumbers.com/2015/4/16/playoff-predictions-with-a-basic-shot-attempts-model

The percentages are based on the probability of the home team winning each game and series. So the model predicts four relatively clear series wins for Ottawa, Pittsburgh, St. Louis and Vancouver.

 

[The Stig]

Can be helpful, but I dont believe theyre something you should build a team around

 

[Wilch]

Can be helpful, but I dont believe theyre something you should build a team around

In my opinion, the advanced stats is more or less, a filter and a form of standardization at the amateur scouting level.

Properly evaluating players is extremely time consuming and requires attention.

With the help of advanced stats, it is much easier for scouts to zone in on certain players like Brayden Points and Damon Seversons while ignoring the Patrick Whites and MacKenzie Stewarts.

At the professional level, it is a tool that help scouts determine player's performance and is a strategic metric.

It is, and never was, something that's meant to be a be-all and end-all solution to building a team.

 

[Verviticus]

Can be helpful, but I dont believe theyre something you should build a team around

this just illustrates a fundamental misunderstanding of the concept. that's like saying "goals are helpful, but you dont build around them" or "save % is helpful but you cant build with it". everything is a piece of a larger puzzle and what we have now are more useful than what we had five years ago (repeat until stone age)

 

[Taelin]

this just illustrates a fundamental misunderstanding of the concept. that's like saying "goals are helpful, but you dont build around them" or "save % is helpful but you cant build with it". everything is a piece of a larger puzzle and what we have now are more useful than what we had five years ago (repeat until stone age)

Like most stats, you have to use them in context. I think advanced stats are useful when used as evidence for support. Also helps with recognizing flukes vs. the real thing.

 

[Addison Rae]

leggggo analytics squad

 

[Verviticus]

Like most stats, you have to use them in context. I think advanced stats are useful when used as evidence for support. Also helps with recognizing flukes vs. the real thing.

they're useful for all sorts of things. if we limited their use to simply support for an argument, people never would have made the connection that shooting % and sv % tend to regress. arguing with them is one thing but its way more interesting to look at numbers that don't mesh with what i perceive and question where the flaws are

 

[Addison Rae]

the human eye is naturally bias we see things the way we want to see them. for example y2k Canucks and myself likely don't see the same things when watching the Sedins play, it's conformation bias. the eye test is obviously important but it's dumb to suggest that you think x player is good when all objective information points to the contrary.

there's two extreme stances in this debate, there's the extreme pro advanced stats and the extreme anti advanced stats, both sides are fairly flawed IMO. people that use advanced stats as a be all and end all aren't using them properly, everyone that knows anything about hockey can tell you that Shea weber is better than Patrick Wiercioch, the thing is advanced stats would tell you differently. advanced stats are best used when pilled in with the eye test. people who ignore advanced stats completely are dinosaurs, the information presented towards us gives us a better understanding of our game is totally ridiuclous to dismiss them without properly debunking them. every succssfull organization in the league usses them, and it's no wonder that teams like the Kings and Blackhawks are always near the top in CF%

keeping the puck 200ft away from your net is the best form of defense, it's why id rather use a guy like Sedin or Thornton against other teams best players than a guy like Kesler or Backes. Sure, Kesler and Backes are visually good defensively but who gives a **** Sedin and Thornton are much more effective, and last time I checked the premise of the game was to allow less goals than the opposition, and sedib and Thornton are better at that.

puck possession correlates to goals, look at the beat CF players in the league and look where they rank in the GF% department, obviously there's outliers but for the most part puck possession correlates to goals for.

 

[vanuck]

Some more links I know of:

http://www.war-on-ice.com
Much like Extra Skater before it, provides player and team data as well as for individual games. Probably the most useful if you're looking for a general starting point with regard to advanced stats.

http://www.naturalstattrick.com
Also for individual games, also lets you see how specific players did when on the ice against certain players.

http://www.puckalytics.com/superwowy.html
An enhanced form of WOWY, which lets you see how well a guy did with and without certain teammates.

http://ownthepuck.blogspot.ca/
Usage Adjusted Corsi which takes into account factors like quality of teammates/competition, offensive zone starts etc.

 

[carolinacanuck]

thanks for this thread.

i'm a stats newbie. time to do so homework so i can keep up with the young'uns.

my problem is i hated stats in college, took it 3 times before i passed it. i click on some of these links and i'm like...why am i here, i don't know what any of these percentages mean, and now my brain hurts.

can we have an 'eye test' thread?

 

[Verviticus]

feel free to ask questions. thats honestly why i wanted a thread hanging about. it can be hard to learn from just a single website or primers and guides

 

[vanuck]

feel free to ask questions. thats honestly why i wanted a thread hanging about. it can be hard to learn from just a single website or primers and guides

Agreed, a thread where you can ask people stuff about specifics is probably better.

Also, thanks to Mr. C for making this thread. I guess I'll throw in a question of my own: On behindthenet, is it better to use Corsi Rel QoC or raw Corsi QoC to evaluate a player's competition? Likewise with Corsi Rel QoT vs. Corsi QoT for teammates.

 

[Jay Cee]

Yay! Corsi thread. I guess people didn't like mine enough

 

[van22]

Good resources for advanced stats
[...]
http://chlstats.pythonanywhere.com/
-CHL advanced stats

[...]

Nice - was looking for a site for CHL stats.

But...I would be very weary with their estimated TOI though. It's determined using statistics based on who was on the ice when a goal was scored - they do highlight that there is a margin of error but they probably should set an upper and lower limit as it affects their estimate of /60 by creating wide discrepancies when trying to compare players (the players with the higher estimated TOI look to have worse /60 stats than they actually should).

For example McDavid and Bennett are probably closer to 20min per game and closer to each other in TOI than the eTOI stats that are being used would suggest. Bennett has an estimated TOI of 34min, McDavid 27min.

 

[Tiranis]

Nice - was looking for a site for CHL stats.

But...I would be very weary with their estimated TOI though. It's determined using statistics based on who was on the ice when a goal was scored - they do highlight that there is a margin of error but they probably should set an upper and lower limit as it affects their estimate of /60 by creating wide discrepancies when trying to compare players (the players with the higher estimated TOI look to have worse /60 stats than they actually should).

For example McDavid and Bennett are probably closer to 20min per game and closer to each other in TOI than the eTOI stats that are being used would suggest. Bennett has an estimated TOI of 34min, McDavid 27min.

McDavid does play around 22-25 minutes some games from what I've seen. It's not uncommon for it to be that high for star players in junior. That said, I doubt anyone averages above 25 minutes and even that might be a stretch.

It's just not very good at handling the high scoring players, especially when they're on an otherwise ****** team. Bennett is basically the most extreme example.

 

[Intoewsables]

this just illustrates a fundamental misunderstanding of the concept. that's like saying "goals are helpful, but you dont build around them" or "save % is helpful but you cant build with it". everything is a piece of a larger puzzle and what we have now are more useful than what we had five years ago (repeat until stone age)

Yeah, no kidding. What's wrong with having as much information at your disposal as possible? Most of the stuff available to the public is just based around shots and shot attempts; it's not some sort of voodoo magic.

On a side note, this thread randomly made me realize how much I miss reading mc79's stuff. Does anyone know of an archive out there with all/most of his posts?

 

[Tiranis]

Does anyone know of an archive out there with all/most of his posts?

Nope. This is the best I've seen http://archive.is/www.mc79hockey.com

 

[DL44]

Here's a zone start question....

They only count starts off of face offs correct?

Is there a place we can find the actual raw number of zone starts... and not just a percentage?

Feel with about 60 total zone starts to be handed out per game for all ES / PP / SH situations between 3 different pairings, that zone start % is one of the most abused or misinterrupted stats around relative to it's actual significance.. ( also neutral zone starts I believe is also excluded, so there go 20-33% of the 'handed out' zone starts)

i.e. I have a feeling for some players that one single zone start could effect their percentage by 5-10%...

Part B .. is there a place to see zone start % as a percentage to the total team' s starts?

 

[Verviticus]

Here's a zone start question....

They only count starts off of face offs correct?

Is there a place we can find the actual raw number of zone starts... and not just a percentage?

i know i've seen a tally on some sites, ill put it here if i find it again

i.e. I have a feeling for some players that one single zone start could effect their percentage by 5-10%...

probably most of them, there arent that many starts in any given game

Part B .. is there a place to see zone start % as a percentage to the total team' s starts?

that i've never seen before