Proposal: Statistical Analysis

[91Kadri91*]

 

[4thline]

Now this is interesting. With this along with passing and shot metrics and a large enough data set I'd think there could be some promising potential in quantifying and evaluating different skillsets of players, comparing the progression of each players conversion rates- shot attempts to successful shots, shots to chances, chances to goals. Get a more reliable track on volume generation vs. quality/conversion ability.

 

[91Kadri91*]

I think the idea to compare pass efficiency with goal scoring is great, but the charts supplied are done in bad faith.

The author of the chart suggests that he expects his rsquared to change as the sample size is increased...but not significantly.

And yet, when the sample size is increased merely from 3 to 5 (which is also not reasonable) the rsquared changes considerably from 0.99 to .88. How can he/you possibly conclude that it wouldn't change significantly when n=5 turns into n=30?

Sorry for the late response.

The extra two teams he is using do not have every game measured, so he's comparing incomplete data (passing data from some, but not all, of said team's games) with complete data (the GF of a team up to that point in the season). The two teams' passing efficiency would change as the number of games measured changed, but their goal for percentage would stay the same, meaning that the current correlation is an inaccurate assessment (since the samples are not identical) of validity.

The author showed the correlation of the three teams with complete tracking because that's a more reasonable conclusion (all of the games used for GF measures were tracked). The author's not going from three complete team sets to a five team set, he's going from three complete team sets to five partially completed sets (and comparing it to five complete team sets).

That's my interpretation, at least.

 

[91Kadri91*]

Now this is interesting. With this along with passing and shot metrics and a large enough data set I'd think there could be some promising potential in quantifying and evaluating different skillsets of players, comparing the progression of each players conversion rates- shot attempts to successful shots, shots to chances, chances to goals. Get a more reliable track on volume generation vs. quality/conversion ability.

Agreed, it's progress.

This is sort of related (in terms of player classification), and fairly interesting, if you (or anyone) is interested: http://donttellmeaboutheart.blogspot.ca/2015/06/clustering-nhl-forwards-using-k-means.html

 

[The Legend]

No.

My issue is the inferences being made based on shot-based analysis.

There's a whole sub-forum dedicated to this junk.

Rather than littering this forum with this amateur analysis, perhaps you should be taking it there.

http://hfboards.mandatory.com/forumdisplay.php?f=241

I'm sure it will be a hit.

I thought the OP's post - if you bother to read it - was informative, well-reasoned, and - most importantly - well-researched.

If you have a real issue with it, you a) can ignore the thread or - and apologies if this is crazy talk - b) contribute some meaningful analysis of why it's junk, what statistics you prefer, how the statistical principles were applied improperly - literally ANYTHING beyond just another negative comment in a stats-based thread.

Because it's tired.

 

[91Kadri91*]

ohhh baby, i'm going to have to take some time to peruse the op later. Im sure some of this info will go a loong way in helping a bettor. Thanks for the resource list

 

[91Kadri91*]

I listed Chris Boyle among the 'people to follow', but I thought I'd show y'all some of his work. It's fantastic:

Terms like "defensive defenseman" or "shutdown defenseman" tend to evoke images of a staunch defender, tough, big, blocks shots, clears the net and is willing to sacrifice for the team. These players tend to be defined as specialists on one side of the puck. Their forte is to defend and they are elite at this skill.

At issue with this type of thought process is the separation of offense and defense in a free-flowing game. It is understandable where these differentiations emerge when we take into account other major sports in North America. In baseball, offense and defense are two separate entities. You can be an elite defender and have zero offensive capabilities because they both exist in a vacuum. One does not blend into the other. Football is the same with very little crossover outside of defensive turnovers that lead to touchdowns. Even basketball has defined offensive/defensive possessions with a 24-second shot clock.

Hockey doesn’t work in the same manner. Outside of special teams, the game operates in a continuous flow. In order to be an effective player, you must excel on both sides of the ice. If you are poor at one aspect, you must overcompensate by being dominant in the other one.

http://www.msgnetworks.com/shows/hockey-night-live/chris-boyle/is-keith-yandle-a-good-defenseman.html

Most of it pertains to the New York Rangers, but it's still really interesting.

 

[Menzinger]

I'll have to give Boyle a follow.

The good news is I think the current Leafs management group have understood the decline and fall of the "shut down" defensemen.

If you can't skate or can't move the puck stay off the ice.

 

[91Kadri91*]

 

[91Kadri91*]

 

[91Kadri91*]

Fixed.

 

[WilliamNylander]

Saw Burch retweet this one today.

Nylander producing in the SHL last year at a 45 point NHL pace is insane. Already second line production as an 18 year old kid. I wouldn't be surprised if he's able to score at a 50+ point pace next year for the leafs.

 

[Menzinger]

Yep - those numbers are absolutely insane.

Kid is going to start his Leafs career with a bang.

 

[Kurtz]

I listed Chris Boyle among the 'people to follow', but I thought I'd show y'all some of his work. It's fantastic:

Hah, I actually sit 10 feet away from Chris at work. I'll pass on your compliments. He works for Sportsnet and MSG, which is why lots of his stuff is on the Rags.

 

[91Kadri91*]

How good is Kuznetsov? Pretty darn:

Some people here in America don’t like Russian style. They say it’s boring, all you do is skate. Nobody fight. Blah, blah, blah. But I like to see when team possess the puck for two minutes and then wait for guy to shoot in the open net. Here, some fans always yelling “Shoot it! Shoot it!” when you cross blue line. But watch how much Chicago holds onto the puck in the playoffs. They don’t have many Russians, but they play the Russian style. I’m happy to see it working in NHL. To me, that’s the best way to play hockey. That’s amazing.

What makes that surprising is that the Capitals only generate 54.3 Shot Attempts per every 60 minutes that he's on the ice. Kuznetsov is an extremely creative player who appears to prefer physical (rather than theoretical) puck possession over taking low quality shots. That might explain why the team doesn't generate more SAT while he is on the ice and why over 50% of the shot attempts that they do take are scoring chances (30.14).

 

[91Kadri91*]

 

[91Kadri91*]

Almost everywhere you look, Drouin is better than team average on the Lightning, but there is one exception: shooting. Drouin has just six goals in his NHL career over 95 games including the playoffs, and a big reason why is he just doesn’t shoot very often. In fact the only regulars on the Lightning who shoot less often at even strength than Drouin are defencemen Matt Carle and Andrej Sustr.

However, in every other area of the game, Drouin is actually quite excellent. Only Nikita Kucherov is involved in more scoring chances per 20 minutes played than Drouin is on the Lightning, and no forward makes as many possession driving plays.

Those two statistics alone tell us that individually Drouin is one of the best players on the Lightning at creating scoring opportunities and driving possession, but going deeper we find that he’s also above average in important areas such as exiting the defensive zone with control, entering the offensive zone with control, and removing possession from the opponent with successful defensive plays, along with getting the puck back with loose puck recoveries.

 

[91Kadri91*]

Hah, I actually sit 10 feet away from Chris at work. I'll pass on your compliments. He works for Sportsnet and MSG, which is why lots of his stuff is on the Rags.

Wow, that's really cool! His work is great; he deserves all of the praise he has (and will) received.

 

[X66]

Great thread.

I actually love threads like this because I know it would make those analytic's detractors heads explode(largely because they don't understand them).

 

[91Kadri91*]

I can't edit my original post at the moment, but a few more 'who to follow' suggestions:

https://twitter.com/Stringhama?lang=en
https://twitter.com/SeanTierneyTSS?lang=en
https://twitter.com/Prospect_Stats?lang=en
https://twitter.com/Josh_Khalfin?lang=en

 

[91Kadri91*]

 

[Stand Witness]

This is what I was saying about his production.

He shot the puck just over once per game. Like 76 shots in 70 games. You aren't going to score like that and it really isn't even enough to provide a sample size on his SH% too. With a 5.2 SH% (with only 76 shots) you can expect that to go up to around 10-11% which would double his goal production with the same amount of shots. Next you would expect with more development/confidence/new system he will be given chances to or told to shoot more.

A player of his caliber/potential should be shooting the puck somewhere between 150-200 times a season, minimum. That will be the ultimate solution to his goal scoring. A bit more luck and a lot more chances.

He has already been a positive possession player in his NHL career too which supports that his shot totals should and can increase.

The kid is going to be a star. He just needs to be given the chances.

 

[Stand Witness]

Great thread.

I actually love threads like this because I know it would make those analytic's detractors heads explode(largely because they don't understand them).

Even as a student who is currently taking a statistics class and taken calculus in the past I still feel a little over whelmed by some of the content (likely due to some of the lack of explanation for some charts).

I couldn't even imagine what someone with no statistical background would feel like reading through some of the graphs.

 

[91Kadri91*]

This is very interesting (and surprising):

It appears that the likelihood of generating a rebound opportunity does not increase with sustained zone time. There is an uptick right around the minute mark but the sample at that point is small (12 rebounds in 113 USAT).

So here’s the takeaway: Possession for possession’s sake, or possession that abstains from taking a “lesser” scoring opportunity for the purpose of retaining control of the puck, may have value; but that value doesn’t appear to be related to offense. To put it another way, there is no inherent boost to shooting percentage as the amount of time spent in the offensive zone increases.

http://hockey-graphs.com/2015/12/01/rebounds-extended-zone-time-and-the-quest-for-more-offense/

I suggest reading the whole article; I only posted the conclusion.

 

[91Kadri91*]

Ladies and gentleman, the long awaited replacement for CHLStats is here: http://www.ohl.prospect-stats.com/ES.html