Advanced Stats Explained

[#freejame]

6 hours ago, That Aud Smell said:

Very interesting.

 

This is going to be soooooooo ***** awesome. It’s going to be an amazing tool for coaches of all levels. I can’t wait to see. 

[SwampD]

1 hour ago, #freejame said:

This is going to be soooooooo ***** awesome. It’s going to be an amazing tool for coaches of all levels. I can’t wait to see. 

None of us are going to see any of it. They are going to sell it all to EA Sports for a $h!t ton of money to make NHL 2020 more lifelike.

[#freejame]

23 minutes ago, SwampD said:

None of us are going to see any of it. They are going to sell it all to EA Sports for a $h!t ton of money to make NHL 2020 more lifelike.

Ha as if EA gave two shits about the NHL franchise 

[LTS]

1 hour ago, SwampD said:

None of us are going to see any of it. They are going to sell it all to EA Sports for a $h!t ton of money to make NHL 2020 more lifelike.

 

1 hour ago, #freejame said:

Ha as if EA gave two shits about the NHL franchise 

NHL 20 is already out.. so it'll have to be NHL 21. ?

[WildCard]

 

[Taro T]

19 hours ago, That Aud Smell said:

Very interesting.

 

Except that yellow 1st down line was an outgrowth of the Fox Sports "glowing puck."  So, the NHL was 1st.  (Their NFL broadcast stole the robots idea from their NHL broadcast, too.)

[That Aud Smell]

The charts below depict, in some part, what the Buffalo skaters are doing at 5v5. As I understand it, this metric is intended to express what you can expect out of a skater and various combinations when it comes to shots (on goal, I presume) over the course of 60 minutes of TOI play (as distinct from game-play time). 

It's interesting to see just how good the LOG line and its parts have been. It's also interesting to see that Eichel -- with his top-flight quality of competition -- is just getting by in terms of 5v5 shots expected.

 

[Taro T]

8 hours ago, That Aud Smell said:

The charts below depict, in some part, what the Buffalo skaters are doing at 5v5. As I understand it, this metric is intended to express what you can expect out of a skater and various combinations when it comes to shots (on goal, I presume) over the course of 60 minutes of TOI play (as distinct from game-play time). 

It's interesting to see just how good the LOG line and its parts have been. It's also interesting to see that Eichel -- with his top-flight quality of competition -- is just getting by in terms of 5v5 shots expected.

 

Interesting chart.  Who's that way over on the left all by himself? ?

[North Buffalo]

 

[triumph_communes]

Ripping on the WOWY thank god

[matter2003]

On 10/10/2019 at 10:06 PM, SwampD said:

None of us are going to see any of it. They are going to sell it all to EA Sports for a $h!t ton of money to make NHL 2020 more lifelike.

Yeah...except teams will still average 10 shots a game too many, let in wacky goals way too much, etc...they add tons of "cool new features" but don't bother fixing things that have been broken for years, just like Madden where im sure teams are still able to return muffed punts for TDs and teams always have all 3 Timeouts at the 2 minute warning each half and all of the teams use their timeouts in the exact same way...

[Thorner]

15 hours ago, matter2003 said:

Yeah...except teams will still average 10 shots a game too many, let in wacky goals way too much, etc...they add tons of "cool new features" but don't bother fixing things that have been broken for years, just like Madden where im sure teams are still able to return muffed punts for TDs and teams always have all 3 Timeouts at the 2 minute warning each half and all of the teams use their timeouts in the exact same way...

NHL THREEEEEEEES!!111!!

[Stoner]

On GR this morning Jeremy reported that some analytics site has final record predictions for NFL teams. The Bills are to finish 11-5. "It's math," Jeremy said, contrasting with Howard's 13-3 eye test.

Can we at least agree that it's hogwash to say math can predict a football team's record? How many smart people have gone broke thinking that way?

[TrueBlueGED]

1 hour ago, PASabreFan said:

On GR this morning Jeremy reported that some analytics site has final record predictions for NFL teams. The Bills are to finish 11-5. "It's math," Jeremy said, contrasting with Howard's 13-3 eye test.

Can we at least agree that it's hogwash to say math can predict a football team's record? How many smart people have gone broke thinking that way?

Do you think that bookies don't use math when setting lines, including for season-long bets? 

[Curt]

2 hours ago, PASabreFan said:

On GR this morning Jeremy reported that some analytics site has final record predictions for NFL teams. The Bills are to finish 11-5. "It's math," Jeremy said, contrasting with Howard's 13-3 eye test.

Can we at least agree that it's hogwash to say math can predict a football team's record? How many smart people have gone broke thinking that way?

Math can attempt to predict such things, just as anyone who watches games can.  Neither will be correct all the time.  Which will be right more often, I can’t say for sure.

[SwampD]

Just a point that needs to be made. It’s not math. Math comes up with formulas that are correct 100% of the time.

This is applied math, and it’s wrong quite often.

[WildCard]

Just now, SwampD said:

Just a point that needs to be made. It’s not math. Math comes up with formulas that are correct 100% of the time.

This is applied math, and it’s wrong quite often.

The math is right, how people interpret it varries. It's not wrong 

[SwampD]

2 minutes ago, WildCard said:

The math is right, how people interpret it varries. It's not wrong 

You are missing the point. It’s not math.

[Thorner]

41 minutes ago, SwampD said:

You are missing the point. It’s not math.

Message/messenger. The message is (hypothetically) pure stat (which, contrary to popular belief, can still have inherent bias/failure, as with anything and everything in life). Messenger will interpret and apply as they see fit. 

Edited October 21, 2019 by Thorny

[triumph_communes]

Sports statisticians take 4% Rsq and promote it as the metric to explain everything.  Sports stats are bad.  Journalists who don't understand the basics about stats then take these things and make it even worse.

[Thorner]

Just now, triumph_communes said:

Sports statisticians take 4% Rsq and promote it as the metric to explain everything.  Sports stats are bad.  Journalists who don't understand the basics about stats then take these things and make it even worse.

Who's more foolish, the fool or the fool who follows him? @DarthEbriate

[LTS]

14 hours ago, WildCard said:

The math is right, how people interpret it varries. It's not wrong 

It's not the interpretation of the result that is the problem.  So, yes, the math is right.  You punch in variables and the output is predictable.  The key point is deciding what variables and in what way they are used in any given formula.  

So, it's interpretation of information prior to the submission of that information to the formula.  

Always my biggest problem with my statistical business classes. They always focused on the output.  I get the output to say a lot of different things if I don't have to explain the inputs.  Fiddle around with numbers until you get something you like and then go.  

DISCLAIMER: I am not insinuating that analytics is that way. I understand that more work goes into verifying that the outputs of any set of statistical inputs correlates with what is happening on the ice.

[DarthEbriate]

23 hours ago, PASabreFan said:

On GR this morning Jeremy reported that some analytics site has final record predictions for NFL teams. The Bills are to finish 11-5. "It's math," Jeremy said, contrasting with Howard's 13-3 eye test.

Can we at least agree that it's hogwash to say math can predict a football team's record? How many smart people have gone broke thinking that way?

That's not math. 11-5 is 6. That's math.

13 hours ago, Thorny said:

Who's more foolish, the fool or the fool who follows him? @DarthEbriate

Fancy stats and applied math are true, from a certain point of view. They're all based in fact (hopefully), but what you do with facts and which ones you use for your analysis, and what message you want to share all shape your truth. (And the answer is: the situation and who is viewing/making a decision on the fool and the fool's follower determines who is more foolish. It's based on your point of view.)

[That Aud Smell]

23 hours ago, PASabreFan said:

On GR this morning Jeremy reported that some analytics site has final record predictions for NFL teams. The Bills are to finish 11-5. "It's math," Jeremy said, contrasting with Howard's 13-3 eye test.

Can we at least agree that it's hogwash to say math can predict a football team's record? How many smart people have gone broke thinking that way?

I think you've posited the question incorrectly.

I think the question is: Does a sound analytical model more accurately predict a team's future results than a sportscaster's best guess(es)?

No one has a crystal ball, obviously.

I'm also - all of a sudden - intrigued by how "small sample size" factors into NFL modeling.

[Brawndo]

https://www.seattletimes.com/sports/hockey/nhl-seattle-analytics-guru-alexandra-mandrycky-details-advanced-statistics-for-the-average-fan/