Jump to content

Goalies and Regression: S63-S66


Recommended Posts

VHL Goalies and Regression S63-S66

 

VHL goalie TPE skill statistics and game statistics were gathered from Seasons 63 through 66. A total sample of 68 goalies (including bots) were collected. Due to constraints in regression analyses, bots have to be included to meet minimum sample size requirements to ensure accuracy in the results. Although the tests have meet sample size requirements, it provides further limitations in the analyses due to the distinct differences in some TPE skills when comparing bots to actual goalies. The most appropriate thing would be to gather more actual goalies, however, I am far too lazy to do so. It takes long enough to collect and align four seasons worth of TPE and stats. In other words, take these findings with a grain of salt.

 

Also, I added an additional variable called “Group”. This variable distinguishes between goalie and bot. This will help clarify the analyses and if this is ever significant in the results, we can determine how much of the variation in the regression model comes from whether it’s a goalie or a bot.

 

Since most of us are unfamiliar with the goalie stats, I’ll list them out here as the independent variables (i.e., TPE skills) used in this analysis:

·         Skating (SK)

·         Size (SZ)

·         Agility (AG)

·         Rebound Control (RB)

·         Style Control (ST)

·         Hand Speed (HS)

·         Puck Handling (PH)

·         Reaction Time (RT)

·         Penalty Shot (PS)

·         Experience (EX)

·         Leadership (LD)

 

One more thing I want to mention about goalie stats: these will never be accurate. All of this is speculation since there is there is so much more to a game than just the goalie. For example, predicting wins from a goalies stats. A stat may be highly correlated with wins, but the goalie could have a real good team in front of him. Or for example, predicting shots against, or saves – this depends on how much offense the other team throws at you – if it’s a team full of Micheal Gary Scott’s your shots will be low against that goalie, but if it’s a team of Palo’s, the shots will be very high. This will affect things like GAA and SV% as well. Keep an open mind when it comes to this data.

Nevertheless, let’s get this underway.

 

Wins

 1AHZ7CY.png

 

The model obtained a significant statistic however there was no relationship between any TPE skills and Wins. The closes stat was Skating. It is approaching significance but statistically, there is no meaningful relationship.

 

Losses

I won’t post this one as the model was insignificant.

 

Save Percentage

WfnWQnZ.jpg

 

This model was statistically significant. Only one TPE skill showed a relationship to SV% and that was Size. Theoretically, the higher the Size skill of a goalie, the more likely they are to have a higher SV%. This makes complete sense, the larger a goalie can make themselves when players are shooting, and the more likely they are to stop more shots.

 

GAA

This one is disappointing because the model was statistically insignificant. I’m going to violate the laws of stats and look at it anyway. This also means any relationships observed are null but fuck it.

w7wH3wy.jpg

 

The significant (but null) relationship here is between Size and GAA. This one is a bit backwards though. The smaller your size stat, the higher your GAA. This contradicts the SV% finding. So ignore this one completely since it’s null and violated stats laws anyways. No relationship to see here.

 

 

Before we waste any time on further stats, the same patterns have emerged for a lot of the other tests. The main problem we have is bots in the data set. So I am going to go back and go through the same tests you just saw, and more, while taking bots out of the data and violating the sample size requirements.

 

Wins

j3KsFLE.jpg

 

This model proved to be statistically significant. Remember we’re looking at goalies only now, with no bots included.

The one TPE relationship we can see is between Size and Wins. The higher your size stat, the more wins you are likely to have.

 

Save %

ViJH1vU.jpg

 

This test obtained a significant model. Unfortunately, there is no significant relationships between TPE skills and SV%.

 

GAA

vdEY7Qi.jpg

 

This model was significant. Here we have two relationships between GAA and TPE skills. First is Reaction Time. The higher the reaction time, the higher the GAA. And yes read that again. You want a lower GAA. So this is basically saying if you’re faster, you’ll let more goals in on average. Next, the more significant relationship is between Rebound Control and GAA. This one is a good relationship. The higher your Rebound Control, the lower your GAA. Less rebounds = less goals, Hockey 101.

 

Assists

wfX9PV9.jpg

 

This one I did for fun. Not the results we all expected. The model obtained a significant stat. The only significant relationship to Assists was…..SIZE! Weird. Thought it would be Puck Handling if anything.

 

Shutouts, Shots Against, Penalty Shot SV%, Penalty Shot GA

These models obtained a significant statistic, yet there was no significant relationships found. I won’t display the tables. Nothing is even close. There most likely is not enough goalies to show any effects.

 

Judging from these highly reliable models, there are three main TPE skills that truly matter for goalies in the past 4 seasons:

·         Size

·         Reaction Time

·         Rebound Control

If you’re a goalie, pump these stats up. You heard it here first from me first.

 

Extra Stats for Fun

Here, I did a similar thing as my lasts stats article. I used a Confirmatory Factor Analysis (CFA) to see which the most common builds for goalies are. Basically this is telling us what the most common TPE skills which are the focus of our goalies.

a68uvqM.jpg 

 

Primary Goalie TPE Skills:

·         SK, SZ, AG, RB, SC, HS, RT

Secondary Goalie TPE Skills:

·         PH, EX, PS, LD

 

Conclusions From All of These Statistics

Well first of all, it speaks to the similarity of all the goalies that are made. There is not much variation when you look at our active goalies. Their TPE is pumped into the same 7 categories. This means we don’t really have enough data to understand any effects of these TPE on outcomes such as GAA and SV%. If we go back into further seasons, this may change, but as far as patterns go in recent seasons, goalies are the same. Another conclusion is that Size matters. Size showed two relationships within S63-66 goalies (no bots) where it helped explain some positives. So one might say to put more TPE into Size if you’re a goalie. 

 

Goalies are weird.

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...