VsX similarity scores?

[tinyzombies]

Is there an easy way to do this, or is it as simple as looking at the rankings? My brain not work good.

Click to read more...

 

[ResilientBeast]

Is there an easy way to do this, or is it as simple as looking at the rankings? My brain not work good.

Meaning what? Similarity in peaks?

 

[Hockey Outsider]

Is there an easy way to do this, or is it as simple as looking at the rankings? My brain not work good.

I can probably put that together relatively easily - which players and over how many years?

 

[tinyzombies]

Meaning what? Similarity in peaks?

Not sure- like how they do it in hockey-reference, except their comparisons don't seem right. Not sure if they use adjusted numbers or how they adjust.

 

[Hockey Outsider]

Not sure- like how they do it in hockey-reference, except their comparisons don't seem right. Not sure if they use adjusted numbers or how they adjust.

Hockey-reference.com's similarity scores are based on points shares. The calculation of defensive point shares are so flawed that they likely render that stat useless.

 

[BenchBrawl]

I have no idea what the hell you're asking tiny.

 

[tinyzombies]

I have no idea what the hell you're asking tiny.

I'm guessing VsX Goals7 would only be one component of a similarity score. Wouldn't you include playoff VsX, assists, PIM, etc.?

 

[Johnny Engine]

NHL Similarity Scores | Hockey-Reference.com

Seems like this process would be pretty easy using one kind of VsX, though I'm not sure exactly what you'd find out.

 

[BenchBrawl]

I'm guessing VsX Goals7 would only be one component of a similarity score. Wouldn't you include playoff VsX, assists, PIM, etc.?

All of this was done, except perhaps PIM.

@seventieslord did the playoff VsX thing a few years back, here: NHL Playoff VsX, 1918-2016

Goals 7 years VsX is in that thread: Reference - VsX comprehensive summary (1927 to 2020)

Assists too.

A few pages later there's even PPG VsX.

My point is, instead of asking for an obscurely defined stats, figure out what is it you really want to know, and then we can build a stat for it.

 

[BenchBrawl]

And FTR, I'm increasingly against trying to come up with THE magic system outputting THE magic number that will enable us to compare all players on a single scale. Not that anybody really believed this was possible, but if I'm moving in a direction at all, it is away from this.

I rather just use each tool like a surgical knife for my specific purpose, and keep all of them in mind when comparing players on a global context.

 

[tinyzombies]

It's probably an easy formula to apply ranges to each stat and more of a style comparison than anything. Not sure the utility of it for ATD, but I would find it interesting.

architecte-hockey has player style spider charts which are interesting when putting a line together, but we don't have entry/exit stuff or TOI for older players obviously.

 

[Johnny Engine]

It's probably an easy formula to apply ranges to each stat and more of a style comparison than anything. Not sure the utility of it for ATD, but I would find it interesting.

architecte-hockey has player style spider charts which are interesting when putting a line together, but we don't have entry/exit stuff or TOI for older players obviously.

Ok, so what you're asking for has nothing to do with H-R's similarity scores then.

I don't think it would be overly difficult to graph the relationship between a player's goal-scoring and playmaking measurements, though there'd be a ton of confounding factors that would make that less useful than one would imagine.