[V]

Originally Posted by ricardodw

So in your mind:

I wanted to start with this, because this is so important. I do not have pre-conceived ideas in my mind that I try to use stats to verify. That's not how statistics works. I think that's the cardinal rule you're breaking in this thread, and your mind is so made up that you appear to be unable to accept the validity of the other viewpoints that you're seeing in this thread.

1) hitting and blocking shots and winning or losing puck battles is a waste of effort? All hockey players should be striving to increase their shots on net and scoring stats? The team with the top two scorers (TB) should win at least most of the time. aside: only 19 of the top 30 scorers last year were on teams in the playoffs.

No, you're misreading things. They aren't a waste of effort at all. They are an indicator that the other team has the puck. This has been mentioned a couple of times in this thread, but you're not seeing it, so maybe I can flesh out the argument for you.

In all the time that people have tried to understand what drives wins there has only been one fundamental answer. Goals. (Throwback to mc in 2006 for those that remember)

The problem is that goals are a rare event, so the advanced statistics world has had to understand what drives goals. And the consensus answer is possession. Possession drives goals which drives wins. It is an undeniable truth that has been demonstrated over and over again. You need to have the puck to score. And you need the other team to not have the puck to keep them from scoring.

This means that metrics that drive possession can be used to predict success. Mind you, you can't just cherrypick a stat that you like and show that it will project success. You need to use valid statistical methods. But that doesn't remove the fact that possession metrics drive wins.

Now think about your gritty stats. Hits, blocked shots and takeaways all increase dramatically when a team does not have the puck. Giveaways are actually an indicator of possession, because you need to have the puck to give it away. Hits and blocked shots are actually more intuitive, because they do not guarantee you got back posession. Just because you hit someone, doesn't mean they gave up the puck.

So to make a long story short, no, hitting and blocking are not a waste of time. But what would be even better would be to not have to hit or block shots because you always have the puck.

2) There is no statistical way to show that a player is soft relative to his peers and some players contribute to a team's effort in playing more physical?

I think I just showed you a statistical way. Of course, it's bunk. I also think that it's a narrative that mainstream media likes to use that has little relevance to what's actually happening out there. That's not to say that being soft doesn't affect results. I only mean to say that if a player has attributes that will increase or decrease the potential for success, it will be shown in the real metrics that matter.

3) the leader in the take-ways ( average of the top 30 Def - in 2012 40/yr) needs to weighted to be the equivalent of the leader in hits ( Average of the top 30 hitting D Men 190/yr) and blocked shots (top 30 average 160).

Not at all, I was merely trying to add statistical relevance to your model. It makes no sense to add absolute values when the scale of those values don't match. If there are 10x more hits on average than the takeaway/giveaway differential, by adding the absolute values you are saying that hits are 10x as important as the differential. Which, to be completely honest, is perfectly fine, but it's not what you were getting at in your initial analysis. You were just adding values. So I decided to add some statistical relevance by at least normalizing the values so they could be viewed as equals.

There's nothing saying you couldn't add weighting to each value. In fact, that's exactly what a regression model would do for you in the method I had originally proposed you try.

A hit, a blocked shot and a Take-away and a give away are all hockey events. Why in the world would you feel a need to make a takeaway and give away worth 5 hits? or 4 Blocked shots?

That's not what I'm doing, though. I am normalizing values so that the value is no longer an absolute value, but its distance from the mean relative to its standard deviation. So now it's not "Player A has tons of hits!! He must be gritty!" but instead we're looking at how many hits he has relative to the rest of the league. This isn't rocket science.

weighting for an explanation

Clever.

You know, I thought about this earlier, and it's too bad that people like you who try their hand at stats get slapped down so vehemently, like a lot of comments in this thread, my first one included. I think it's great that people are trying to quantify the nuances in the game. I think that your methods are misguided, but I'm hoping that this post at least gives you a little insight into some statistical methods that you might want to try for future endeavours. There's a lot of good stuff out there worth reading as well, so I strongly encourage you to read into it and get better at this. We need more people interested in stats.

But then I remember that you're the guy that brings up Bouwmeester in every gd thread. So you deserve no sympathy.