PDO and Unsustainable Stats

[Beatle17]

Originally Posted by Resolute 14

You don't understand what outliers are then. Case in point, over the last five full seasons, only one team managed a PDO of 1.03 or higher. That is one instance out of 150 team seasons. Only four posted a PDO of 1.025 or higher. Only eight managed 1.02 or higher. These are outliers.

So when people say the Flames' PDO of 1.031 is unsustainable, they do so because they do not believe the Flames are likely to finish at the extreme high end of the scale.

Bluntly, the Flames are not likely to finish at the extreme high end of the scale.

But that does not mean it can't happen. Colorado did it last year under the same circumstances.

So the stats don't prove anything, it is just that "they do not believe". This comment is exactly why people like myself do not have much faith in "advanced" stats. Call them stats that report on past instances and leave it at that.

[Resolute 14]

Originally Posted by Beatle17

So the stats don't prove anything, it is just that "they do not believe". This comment is exactly why people like myself do not have much faith in "advanced" stats. Call them stats that report on past instances and leave it at that.

That is, by definition, what stats are. You're blaming the stat because you don't like the conclusion drawn by people reading them.

[Street Pharmacist]

Originally Posted by Beatle17

So the stats don't prove anything, it is just that "they do not believe". This comment is exactly why people like myself do not have much faith in "advanced" stats. Call them stats that report on past instances and leave it at that.

"Stats" don't believe anything. People do. If stats are useless and can't give you a sense of likelihood of future events then I don't know what to tell you. Your stance is absurd.

[Beatle17]

Now you can see why people who don't use advanced stats as the be all and end all in a discussion choose not to post. Stats used in context, with visual input and knowledge of the process of the professional game, used by people who work in the game are useful. Somebody inputting information in a spread sheet with information from a boxscore and spouting off that it is "unsustainable" or "advanced stats" are the ones who don't have a leg to stand on. I fully believe that stats in context, which I believe every team uses but don't discuss with the public, are valuable. I don't believe that people like SP and Res14 above telling me I don't know how to use them or "get" them ruin the discussion.

And using stats for a team game from season to season, or month to month with a constant changing of the parts (players) changing at a rate of 20% per season (or event) can not prove anything. They can say that the Flames from last year (last month) played at a rate that is too high of a variable from the norm (center) but in no way can they predict that the variable is going to continue. If the team and players all remained static then maybe they could state what the norm is for the team and try to predict what the numbers will show.

Corsi, Fenwick and other stats tell us that the Oilers are better than 10-12 other teams. They aren't period. Possession can't be measured by shots directed towards the net, as someone else mentioned earlier the Sedins could control the play for 2 minutes in the offensive zone and only get 2 grade A+ scoring chances and only the get the same credit for "possession" as Luke Gadzic taking a shot from outside the blueline and a rebound shot from the corner. Which one is more "possession" to you?

The advanced stats followers can present a good argument for themselves but VERY seldom do, they just bash people who don't agree with their basic statement.

[Street Pharmacist]

Originally Posted by Beatle17

Now you can see why people who don't use advanced stats as the be all and end all in a discussion choose not to post. Stats used in context, with visual input and knowledge of the process of the professional game, used by people who work in the game are useful. Somebody inputting information in a spread sheet with information from a boxscore and spouting off that it is "unsustainable" or "advanced stats" are the ones who don't have a leg to stand on. I fully believe that stats in context, which I believe every team uses but don't discuss with the public, are valuable. I don't believe that people like SP and Res14 above telling me I don't know how to use them or "get" them ruin the discussion.

And using stats for a team game from season to season, or month to month with a constant changing of the parts (players) changing at a rate of 20% per season (or event) can not prove anything. They can say that the Flames from last year (last month) played at a rate that is too high of a variable from the norm (center) but in no way can they predict that the variable is going to continue. If the team and players all remained static then maybe they could state what the norm is for the team and try to predict what the numbers will show.

Corsi, Fenwick and other stats tell us that the Oilers are better than 10-12 other teams. They aren't period. Possession can't be measured by shots directed towards the net, as someone else mentioned earlier the Sedins could control the play for 2 minutes in the offensive zone and only get 2 grade A+ scoring chances and only the get the same credit for "possession" as Luke Gadzic taking a shot from outside the blueline and a rebound shot from the corner. Which one is more "possession" to you?

The advanced stats followers can present a good argument for themselves but VERY seldom do, they just bash people who don't agree with their basic statement.

Originally Posted by Beatle17

So the stats don't prove anything, it is just that "they do not believe". This comment is exactly why people like myself do not have much faith in "advanced" stats. Call them stats that report on past instances and leave it at that.

?

You strawman an argument about stats "proving" something, then claim they only elucidate the past insinuating that they have little value.

What is the expected response here...

[Beatle17]

Originally Posted by Street Pharmacist

?

You strawman an argument about stats "proving" something, then claim they only elucidate the past insinuating that they have little value.

What is the expected response here...

It's not a strawman. The stats guys quote them as the final answer, when they aren't. If someone such as myself questions the stats then we are hit with the usual "well you don't understand them so you shouldn't be questioning them" rejoinder and are put down as below the discussion.

Don't act like a dick to people who question the "facts" that you state, accept that maybe the advanced stats don't tell the whole story and not everyone agrees with your "opinion".

[Resolute 14]

If nothing else, Beatle17, I am impressed at how quickly you nailed yourself to a cross there. You make one post demonstrating that, yes, you do not understand statistics at all, then rush right into the obviously pre-planned whine against advanced stats. I am honestly impressed.

Also, bonus points for commenting on the value of using stats in context, then turning around and using a stat out of context to try and claim the group you dislike is arguing the Oilers are better than "10-12 other teams".

[Beatle17]

Originally Posted by Resolute 14

If nothing else, Beatle17, I am impressed at how quickly you nailed yourself to a cross there. You make one post demonstrating that, yes, you do not understand statistics at all, then rush right into the obviously pre-planned whine against advanced stats. I am honestly impressed.

Also, bonus points for commenting on the value of using stats in context, then turning around and using a stat out of context to try and claim the group you dislike is arguing the Oilers are better than "10-12 other teams".

Ok, you want to attack a person again because they question you so I will ask you a few simple question that you should be able to answer.

Which NHL teams have the best Corsi/Fenwick/PDO numbers? Are they the best teams because of these stats, or are the stats in these categories better because of the players/coaching/philosophy of the teams style of play? Did these stats assist the teams to acquire the players (Pouliot on the Oilers) or did the teams acquire the players because they fit the system they want to play? Do all the players on the team, regardless of roles on the team, have the same ratings, and if they don't why not? Can having a better 4th line (Bruins v Oilers as example) change the overall team performance in the stats?

If the stats you want to use are so great they should answer every one of these questions for every player on each team. If they can't then they are a tool to use IN CONTEXT with the other tools I mentioned earlier.

[Resolute 14]

Originally Posted by Beatle17

Ok, you want to attack a person again because they question you so I will ask you a few simple question that you should be able to answer.

How is that any different than your rant? Pot meet kettle.

As to the rest, you have merely asked a series of inane questions, but in the interest of giving you what you claim to want...

Which NHL teams have the best Corsi/Fenwick/PDO numbers?

You're a big boy. You can find that out yourself from the myriad of websites that offer it.

Are they the best teams because of these stats, or are the stats in these categories better because of the players/coaching/philosophy of the teams style of play?

Stats don't make anything better or worse. They merely reflect what has happened or been counted within their defined sample. Though in these examples, they do correlate to varying degrees of precision to the overall standings.

Ultimately though, you aren't asking a question about Corsi, Fenwick or PDO directly, but are asking what things can affect them. That was the discussion we were already having before you sidetracked everything by rushing head on into your "I hate advanced stats guys" posts.

Did these stats assist the teams to acquire the players (Pouliot on the Oilers) or did the teams acquire the players because they fit the system they want to play?

You'll have to go poll the teams themselves if you want the answer to this.

Do all the players on the team, regardless of roles on the team, have the same ratings, and if they don't why not? Can having a better 4th line (Bruins v Oilers as example) change the overall team performance in the stats?

No context. What are you referring to by "rating"?

As to the 4th line question, better players lead to better stats? Goodness gracious! What a revelation! Thank god you are here!

If the stats you want to use are so great they should answer every one of these questions for every player on each team. If they can't then they are a tool to use IN CONTEXT with the other tools I mentioned earlier.

There is no single stat that can tell the entire story. All they can do is give you information and allow you to incorporate that information (or dismiss entirely, as is your choice). Yes, people could choose to look at one stat (or grouping) and draw a conclusion is not appropriate given the stat used. You yourself did that so disingenuously with your Oilers comment.

And congratulations. You've come in with all this bluster to say the exact same things regarding context as I have been saying for the past several weeks in these threads. Go you.

[octothorp]

For those interested in stats, I found this article interesting, regarding how a weighted balance of goals and shots produces a better indicator of future goals than either goals or shots on their own (1.0 for goals and 0.2 for shots, in terms of a 41-game-sample predicting an 82 game sample).

http://tangotiger.com/index.php/site...tial-aka-tango

[theinfinitejar]

I actually think PDO was originally or best used as a statistical second opinion on players. It allows a quick look at a guy who had an uncharacteristically bad or good season to see if they really broke out or really declined or if you should expect him to revert to his previous form. Really useful to find undervalued guys or know when to stay away from a mirage. It's only a measure of luck if you have a baseline to compare it to.

[Jay Random]

Originally Posted by theinfinitejar

I actually think PDO was originally or best used as a statistical second opinion on players. It allows a quick look at a guy who had an uncharacteristically bad or good season to see if they really broke out or really declined or if you should expect him to revert to his previous form. Really useful to find undervalued guys or know when to stay away from a mirage. It's only a measure of luck if you have a baseline to compare it to.

But the thing to look at in that case is shooting percentage. Knowing what a goalie’s save percentage was while a particular skater was on the ice tells you very little about the skater.

[theinfinitejar]

Originally Posted by Jay Random

But the thing to look at in that case is shooting percentage. Knowing what a goalie’s save percentage was while a particular skater was on the ice tells you very little about the skater.

A lot of people look at GF/60 and GA/60 when evaluating players. Obviously your GA/60 and on-ice goal-differential is going to be affected by your on-ice Sv%.

[SebC]

Originally Posted by theinfinitejar

I actually think PDO was originally or best used as a statistical second opinion on players. It allows a quick look at a guy who had an uncharacteristically bad or good season to see if they really broke out or really declined or if you should expect him to revert to his previous form. Really useful to find undervalued guys or know when to stay away from a mirage. It's only a measure of luck if you have a baseline to compare it to.

Originally Posted by Jay Random

But the thing to look at in that case is shooting percentage. Knowing what a goalie’s save percentage was while a particular skater was on the ice tells you very little about the skater.

More accurately, it's a measure of a player's "luck" in +/-, where player PDO is the PDO of his team while he's on the ice.

[Jay Random]

Originally Posted by theinfinitejar

A lot of people look at GF/60 and GA/60 when evaluating players. Obviously your GA/60 and on-ice goal-differential is going to be affected by your on-ice Sv%.

Which is why GA/60 and on-ice differential can only be used to compare teammates. The skill of the goalie cancels out, since both players are playing in front of the same goalie. Which is to say that the save percentage factors itself out of the equation automatically.

GA/60 and on-ice differential still aren’t very good tools, but we haven’t got any good metrics for individual defensive ability. It’s the toughest statistical problem in hockey, and we are a long way from seeing a solution to it. I applaud the stats guys for trying to tackle that problem, but if any of them think they have it solved, they are fooling themselves.

The rock-bottom problem, in my opinion, is that we just don’t have the necessary quality of data to even begin a proper analysis. I looked at the problem some years ago, when I was futzing around trying to write a hockey simulator for my own amusement. I broke down all the events in a hockey game into seven basic categories:

1. Change of possession (passes, turnovers, loose pucks)
2. Change of zone
3. Player substitution
4. Shot attempted (on goal or wide)
5. Shot blocked or deflected
6. Save
7. Stoppage (penalty, penalty shot, icing, offside, faceoff after goal, etc.)

Each event occurs at a particular time and location on the ice, which could in principle be tracked.

(You will notice that I didn’t list ‘goal’ as an event. That’s because if a shot is on goal and no save is made, it’s a goal by definition. I could have counted goals as events instead of saves, but a save is the result of the goalie doing something, and a goal is the result of the goalie failing to do something, and it’s my philosophy that you should look at measuring things that happen, not things that fail to happen.)

The trouble is, #1 and #2 are by far the most frequent events in a hockey game, and they are not tracked. If they were, we wouldn’t need Fenwick or Corsi as proxies for possession, because we could measure possession directly. I understand that the NHL plans to track these things by chipping the players, but it does not plan to share that data with third parties. That means that for the foreseeable future, we simply will not have the data to really tackle hockey as a statistical problem.

[ramizle]

TLDR I didn't learn anything because I didn't read anything. This page is 20 swipes deep. Get a room already you guys.

[Jay Random]

Originally Posted by ramizle

TLDR I didn't learn anything because I didn't read anything. This page is 20 swipes deep.

And you just made it longer to tell us this. Bravo.

[ramizle]

Originally Posted by Jay Random

And you just made it longer to tell us this. Bravo.

Tough crowd. I was trying to make a joke.

Why so angry? The Flames are a treat to watch and they are winning! I assume you're a Flames fan.

Edit: and Edmonton is no good

[Jay Random]

Originally Posted by ramizle

Why so angry?

I’m not angry. I had the idea that you were.

[ramizle]

Originally Posted by Jay Random

I’m not angry. I had the idea that you were.

Naw, I'm giddy like a prepubescent male watching titanic for the first time.

Anyways, on topic. Can we argue that our luck stat, yes I'm calling it that because as mentioned PDO is a stupid name, is actually sustainable as we've consistently sustained high numbers for a while now..?