New to Advanced NHL Stats

[rkhum]

I am new to advanced NHL stats and a bit overwhelmed.

Can of you recommend the best websites that give a comprehensive overview and the top oh three or five stats?
I saw the first thread but it was overwhelming.

Click to read more...

 

[Cunneen]

I am new to advanced NHL stats and a bit overwhelmed.

Can of you recommend the best websites that give a comprehensive overview and the top oh three or five stats?
I saw the first thread but it was overwhelming.

Hey, welcome to the club. I'll speak from experience, since it was only a few months ago that I was in your position.

So the two sites that provide much of the advanced stats data are behindthenet.ca and stats.hockeyanalysis. Both of these sites provides tons of great data and without those two sites discussions about advanced stats would not be possible.

I'd say the best site (in terms of discussion about advanced stats) is Broad Street hockey. It is the philadelphia flyers site on SBnation. The guys who write on Broad Street Hockey (BSH) are really great and knowledgable about advanced stats. Eric T especially is on the forefront of research in hockey analytics.

There are various blogs that talk about advanced stats. One neat thing is that because the advanced stat community is such a small knit community, you will notice that people often provide links to various blogs that use advanced stats.

There is so much more I could say but I'll stop now. If you have any questions you can email me at [email protected] . What I found is that if you just take the time you will begin to understand everything much better. It takes time, but it is worth it.

Cheers

 

[Cunneen]

I am new to advanced NHL stats and a bit overwhelmed.

Can of you recommend the best websites that give a comprehensive overview and the top oh three or five stats?
I saw the first thread but it was overwhelming.

Most common stats you might see are

1) Corsi. Corsi is a proxy for puck possession, and is shots on goal (including goals) plus missed shots plus blocked shots (shot blocked by opponent).

2) Fenwick. Also a proxy for puck possession, Fenwick is missed shots plus shots on goal (including goals).

3) PDO. PDO is shooting percentage (for a team) plus save percentage (for the same team). Obviously since shooting percentage and save percentage are related PDO will regress or increase towards 1000. PDO measures luck (search PDO for a better explanation.

4) O zone start percentage This is basically the measure of how often a player starts a shift in the offensive zone. (Offensive zone starts/ (o zone starts + D zone starts). Generally, a high O zone start percentage means that the a player is being used in an offensive role, while defensive forwards and defensemen will see lower O-zone start percentages.


There are plenty more but those are four of the main ones.

 

[Kershaw]

Most common stats you might see are

1) Corsi. Corsi is a proxy for puck possession, and is shots on goal (including goals) plus missed shots plus blocked shots (shot blocked by opponent).

2) Fenwick. Also a proxy for puck possession, Fenwick is missed shots plus shots on goal (including goals).

3) PDO. PDO is shooting percentage (for a team) plus save percentage (for the same team). Obviously since shooting percentage and save percentage are related PDO will regress or increase towards 1000. PDO measures luck (search PDO for a better explanation.

4) O zone start percentage This is basically the measure of how often a player starts a shift in the offensive zone. (Offensive zone starts/ (o zone starts + D zone starts). Generally, a high O zone start percentage means that the a player is being used in an offensive role, while defensive forwards and defensemen will see lower O-zone start percentages.


There are plenty more but those are four of the main ones.

This about covers it.

 

[GKJ]

Check out nhlnumbers.com too.

 

[megajake]

Advanced Stats

Can anybody teach me sorta how to use them? Where to get them from? What they mean? Which to look at? Like I find it hard to determine how good a player id defensively if I don't watch them.

 

[Brainiac]

Most common stats you might see are

3) PDO. PDO is shooting percentage (for a team) plus save percentage (for the same team). Obviously since shooting percentage and save percentage are related PDO will regress or increase towards 1000. PDO measures luck (search PDO for a better explanation.

I'm not sure I agree completely with this definition of PDO. I know it's the official definition and all but still.

The average PDO (weighted for #shots) for the whole league will be exactly 1000, that's obvious. But a good team could be consistently over 1000 and a bad team could be consistently under. By definition, it doesn't have to regress to 1000 for one specific team. Or it's not entirely due to luck.

A very skilled team can finish a season with a PDO of 1020 or 1030. It's not luck anymore, IMO. Take an NHL team and send it to the AHL and I would bet my house that its PDO won't 'regress towards 1000'.

What could be considered 'lucky' is to have a very high PDO (like 1050 or 1100) over a stretch because of hot goaltending or unusually high shooting%. But that's it.

 

[hatterson]

I'm not sure I agree completely with this definition of PDO. I know it's the official definition and all but still.

The average PDO (weighted for #shots) for the whole league will be exactly 1000, that's obvious. But a good team could be consistently over 1000 and a bad team could be consistently under. By definition, it doesn't have to regress to 1000 for one specific team. Or it's not entirely due to luck.

A very skilled team can finish a season with a PDO of 1020 or 1030. It's not luck anymore, IMO. Take an NHL team and send it to the AHL and I would bet my house that its PDO won't 'regress towards 1000'.

What could be considered 'lucky' is to have a very high PDO (like 1050 or 1100) over a stretch because of hot goaltending or unusually high shooting%. But that's it.

I've always been skeptical of this as well, especially on an individual player level.

If you look at the last 6 years combined sorted by descending PDO you see a fairly good representation of the higher levels of talent in the league.

http://stats.hockeyanalysis.com/rat...000&teamid=0&type=goals&sort=PDO&sortdir=DESC

Crosby, Gaborik, the Sedins, Semin, Burrows, Ryan, Thornton, Datsyuk, Ovechkin, Kunitz, Stamkos all in the top 20. Those guys are either elite or played with elite players (Kunitz and Burrows)

Granted the bottom has some surprising names there (Karlsson is the one the really jumped out to me), but given the percentage of elite players that occupy the higher ends of the list, it seems reasonable to conclude that good players *can* maintain a higher than 1000 PDO.

Clearly, like you said, extreme values are a measure of luck (for example Lupul at 1107 last year or Kadri at 1078) but I don't think I've seen enough proof to support the claim that PDO will always regress towards 1000 on a player or team level.

 

[Jyrki]

I've always been skeptical of this as well, especially on an individual player level.

If you look at the last 6 years combined sorted by descending PDO you see a fairly good representation of the higher levels of talent in the league.

http://stats.hockeyanalysis.com/rat...000&teamid=0&type=goals&sort=PDO&sortdir=DESC

Crosby, Gaborik, the Sedins, Semin, Burrows, Ryan, Thornton, Datsyuk, Ovechkin, Kunitz, Stamkos all in the top 20. Those guys are either elite or played with elite players (Kunitz and Burrows)

Granted the bottom has some surprising names there (Karlsson is the one the really jumped out to me), but given the percentage of elite players that occupy the higher ends of the list, it seems reasonable to conclude that good players *can* maintain a higher than 1000 PDO.

Clearly, like you said, extreme values are a measure of luck (for example Lupul at 1107 last year or Kadri at 1078) but I don't think I've seen enough proof to support the claim that PDO will always regress towards 1000 on a player or team level.

Keep in mind though, most of these players played for elite teams so they also had the benefit of strong goalies keeping their PDO up over the years.

I absolutely agree with you that a player's or a team's "true" PDO can be other than 1,000 but over the course of several seasons, most of them do hover near 1,000. I think saying that it will necessarily regress to 1,000 is a bit misleading, but in most cases it's what happens.

http://stats.hockeyanalysis.com/teamstats.php?disp=1&db=200713&sit=5v5&sort=PDO&sortdir=DESC

 

[LiveeviL]

Measure luck and you'll win the Nobel prize, it would not be luck anymore.

 

[Bomber0104]

Stay as far away from them as possible...my advice.

Garbage way to analyze a player, line, and team.

I won't even give CORSI/Fenwick the dignity of calling them "advanced stats".

 

[LiveeviL]

Stay as far away from them as possible...my advice.

Garbage way to analyze a player, line, and team.

I won't even give CORSI/Fenwick the dignity of calling them "advanced stats".

Refined +/-?

 

[Bear of Bad News]

Stay as far away from them as possible...my advice.

Garbage way to analyze a player, line, and team.

I won't even give CORSI/Fenwick the dignity of calling them "advanced stats".

Every statistic, no matter how advanced, has flaws (and so do personal observation and scouting). Generally, when someone says "Statistic A sucks but Statistic B is the greatest", what that means is that they can see the flaws with Statistic A but cannot (yet) see the flaws with Statistic B.

 

[hatterson]

Keep in mind though, most of these players played for elite teams so they also had the benefit of strong goalies keeping their PDO up over the years.

I absolutely agree with you that a player's or a team's "true" PDO can be other than 1,000 but over the course of several seasons, most of them do hover near 1,000. I think saying that it will necessarily regress to 1,000 is a bit misleading, but in most cases it's what happens.

http://stats.hockeyanalysis.com/teamstats.php?disp=1&db=200713&sit=5v5&sort=PDO&sortdir=DESC

Fair enough, but I also see better teams at the top and worse teams at the bottom. Clearly what seperates Pittsburgh, Boston and Vancouver over the last 6 years from the Isles, Columbus, and Jersey isn't just luck.

Although, I suppose you generally aren't looking at PDO on elite players as much as you are on guys who appear to be performing above expectations and those would much more likely be in the 1000 range.

Either way, I think a statement like "Obviously since shooting percentage and save percentage are related PDO will regress or increase towards 1000" can be a reasons why people get turned off of 'advanced' stats because 1.) it's not actually obvious; and 2.) it's not entirely accurate

If it was 100% luck then you can conclude that Crosby is the luckiest player in the league and Nate Thompson is the most unlucky. However, it's much more likely that Crosby is really good and Nate Thompson just isn't.

 

[Roof Daddy]

New to them as well, but becoming more interested. I find the zone starts important, especially if you relate them to Corsi and +/-.

Basically, if you have a guy starting in his defensive zone more often, yet he/his line are out shooting their opponents and are + players, he/the line must be doing something right.

 

[tarheelhockey]

I'm glad to see this conversation about PDO. That stat bothers the hell out of me for the exact reasons being expressed here.

 

[Brainiac]

Stay as far away from them as possible...my advice.

Garbage way to analyze a player, line, and team.

I won't even give CORSI/Fenwick the dignity of calling them "advanced stats".

I wouldn't go as far as saying 'garbage', but it is true that hockey is a much more 'chaotic' game than baseball or basketball. Thus, it is a game less amenable to statistical analysis.

Advanced stats are still interesting IMO. You just have to keep in mind that they don't carry the meaning they do in baseball, for example.

As somebody else pointed out, you can sort players by PDO and the talent still comes out on top. But it is true that the rest of the team had a part to play in that number.

 

[MasterofGrond]

Stay as far away from them as possible...my advice.

Garbage way to analyze a player, line, and team.

I won't even give CORSI/Fenwick the dignity of calling them "advanced stats".

Points and +/- are all that matter, brah.

 

[tarheelhockey]

As somebody else pointed out, you can sort players by PDO and the talent still comes out on top. But it is true that the rest of the team had a part to play in that number.

But that's an indictment of PDO itself. In theory, with luck being random, if you sort players by PDO it should come out like a randomized list.

 

[Brainiac]

But that's an indictment of PDO itself. In theory, with luck being random, if you sort players by PDO it should come out like a randomized list.

Not sure what you mean there? When you sort out by PDO, it clearly doesn't give a randomized list.

Or maybe that's what you meant? - i.e. PDO cannot be a measure of luck.

 

[Bear of Bad News]

Yes, that's precisely what he meant.

 

[tarheelhockey]

Or maybe that's what you meant? - i.e. PDO cannot be a measure of luck.

Yes, that's precisely what he meant.

Yes, exactly.

The PDO rankings for 2013 (among players >20GP) show 4 Penguins in the top 6. Three of them are Crosby, Kunitz and Dupuis.

The fact that such a thing would happen strongly suggests that PDO does not measure what it purports to measure.

 

[Master_Of_Districts]

Yes, exactly.

The PDO rankings for 2013 (among players >20GP) show 4 Penguins in the top 6. Three of them are Crosby, Kunitz and Dupuis.

The fact that such a thing would happen strongly suggests that PDO does not measure what it purports to measure.

It's true that variation in PDO among individual players is a combination of both talent and luck.

Of course, this is also true for every other hockey statistic that's ever been conceived of.

 

[TheDevilMadeMe]

I'm not sure I agree completely with this definition of PDO. I know it's the official definition and all but still.

The average PDO (weighted for #shots) for the whole league will be exactly 1000, that's obvious. But a good team could be consistently over 1000 and a bad team could be consistently under. By definition, it doesn't have to regress to 1000 for one specific team. Or it's not entirely due to luck.

A very skilled team can finish a season with a PDO of 1020 or 1030. It's not luck anymore, IMO. Take an NHL team and send it to the AHL and I would bet my house that its PDO won't 'regress towards 1000'.

What could be considered 'lucky' is to have a very high PDO (like 1050 or 1100) over a stretch because of hot goaltending or unusually high shooting%. But that's it.

Agreed. PDO is probably the best example of what isn't uncommon among advanced stats, where the creator and proponents of the stat think it does more than it actually does.

 

[TheDevilMadeMe]

It's true that variation in PDO among individual players is a combination of both talent and luck.

Of course, this is also true for every other hockey statistic that's ever been conceived of.

But considering PDO purports to be a measure of luck, isn't this basically acknowledging that it doesn't show the one and only thing that it's supposed to?