The Value of HERO Charts ... I don't see it

[Micklebot]

Where can I find the most recent methods behind HERO charts? I'd like to know the exact statistical model used.

From what I understand, they are just data pulled directly from Puckalytics (or stats.HockeyAnalysis, since it's the same data) and rank ordered for each category.

So for example, CF/60 +/- is CF/60 rel Team (CF/60 on ice - CF/60 off ice). They then rank order all eligible players (I don't know what the min TOI they used is) and make the bar graph based on their rank order.

 

[wgknestrick]

From what I understand, they are just data pulled directly from Puckalytics (or stats.HockeyAnalysis, since it's the same data) and rank ordered for each category.

So for example, CF/60 +/- is CF/60 rel Team (CF/60 on ice - CF/60 off ice). They then rank order all eligible players (I don't know what the min TOI they used is) and make the bar graph based on their rank order.

Exactly.
As I stated in a previous post. They are just a creative "display" tool like a bar chart. The data inside could be anything. You could use them to show any future rate stat.

They can chart (via bar charts) any information and the horizontal scale is effectively "percentiles" , but shown as "lines" or "pairings" to relate them to specific hockey terms.

1st liner > 75% percentile
2nd liner >50%
3rd liner >25%
4th liner >0%
Tanner Glass = 0%

 

[charlie1]

It's a shame there are no published methods.

 

[Micklebot]

It's a shame there are no published methods.

I'm going off memory, but the older version used to have a blurb underneath that explained that it was included the top x players in TOI over the last 3 season (for D I think it was 210, I'd guess it was around 400-500 for forwards), and rank ordered them to get the bar graphs. At the time, it wasn't using relative stats, but instead a usage adjusted stat of his own making (I think), he has a post about UA stats on his blog http://ownthepuck.com/