Login

***luke*** · 09-15-2019, 11:53 AM

Hi,
So this one will be more in depth than my other topics, as I am technically “creating” a new stat, that I pulled from real NHL hockey.

So whats GSVA, well it stands for Game Score Value Added, or in Hockey Terms, using formulas with their Stats, how do they impact the game in 1 stat. So a great article for the NHL GSVA, https://hockey-graphs.com/2016/07/13/mea...ame-score/
This is a great way to get into it(I used it when I was making my own, as well as some stuff from his writing at The Athletic). If you read the athletic Dom Luszczyszyn is a known person in the hockey community there.

Anyway, so a TLDR of the article is that he makes a way to calculate a Game Score to see the way a player impacts the game, either in a single game or through out the season.

So the way he calculates it is this
Player Game Score = (0.75 * G) + (0.7 * A1) + (0.55 * A2) + (0.075 * SOG) + (0.05 * BLK) + (0.15 * PD) – (0.15 * PT) + (0.01 * FOW) – (0.01 * FOL) + (0.05 * CF) – (0.05 * CA) + (0.15 * GF) – (0.15* GA)
Where G = Goals, A1 = Primary assists, A2 = Secondary Assits, SOG = Shots on Goal, BLK = Shots blocked, PD = Penalties Drawn, PT = Penalties Taken, FOW and FOL is obviously faceoff win/.loss, CF and CA is corsi for and corsi against and GF and GA is Goals For and Goals against.

So the way I really converted it to my own purposes, is that I took that formula and had to cut PD, PT, FOW, and FOL(Aaron Wilson didnt calculate them, and I was using his Advanced stats to get it), so instead I have this
Player Game Score = (0.75 * G) + (0.7 * PA) + (0.55 * SA) + (0.075 * SOG) + (0.05 *BLK) + (0.05 * CF) - (0.05 *CA) +(0.5 *AMG) + (0.005 *Steals) - (0.005 * Takeways)+(0.025*HIT)-(0.05*PIM)

So this is my spreadsheet link for it, https://docs.google.com/spreadsheets/d/1...sp=sharing
If you want to look at it in depth there is it, but ill give a bit of an outline here for the top players on here

So if we wanted to test it, we can take the top 10 GSVA players and see if they kinda match up
Philipp Winter 76.86
Mikhail Lokitonov 76.415
DeMaricus Smyth 72.145
Rafe Ulrich 71.065
Anders Christiansen 70.185
Max Mauldin 70.05
Tony Pepperoni 69.9
Dayymo Ralchankinov 69.67
Bobby Watson 69.015
Mike Izzy 67.42

So thats the top 10, so we can tell that it is pretty effective. So one person you might be questioning on why he was able to get high into the top 10 would be Anders Christiansen, a Center on NEW. So if we look at his Case

Anders Christiansen: 14 goals, 21 PA, 11 SA, 46 HIT, 8 PIM, 1079 CF, 847 CA, 385 STL, 458 Turnovers,24 AMG.

So one thing that really stands out is a really high CF to CA rate, a high amount of PA, and high AMG that would make him be higher than some people.

Actually a big theme on this(other than Tony Pepperoni) is that they all have over 23 AMG. Actually in the top 20 of that season, there are only 3 people with under 22 AMG in the top 20.
#20 - Alex Light - 19.99 AMG
#18 - Piotr Czerkawski - 21.51 AMG
#17 - Joshua Woodcroft - 19.19 AMG
#14 - Luke Thomason - 21.98 AMG(basically 22 AMG, so not counted)

So thats for S49, but what about for S48

When going back to S48 the top 10 goes like this

Flacko Lagerfield 81.71
Anders Christiansen 79.76
Joe Kurczewski 79.5
Jason Visser 77.36
Oisin Fletcher 75.945
Lil Manius 72.275
DeMaricus Smyth 69.345
Shooter McGavin. 69.405
Mike Izzy 71.79
Louie Garrett 68.095

With of course, Flacko with his monster year lead in GSVA. This one(there are a lot, so if you pull up the spreadsheet and go to S48 Skaters), has 6 players in the top 20 below 22 AMG, as well as a good trend of most people having 40+ points (2 people who dont have 40+ points have 39 points in Viktor Maruis and Nicholas Williams)

And then finally going back 1 more year in S47 we get this
DeMaricus Smyth 83.64
Piotr Czerkawski 78.065
Lyndis Vakarian 76.845
Mikhail Lokitonov 77.23
Dani Forsberg 73.245
Oisin Fletcher 74.025
Chris York 73.36
Jason Visser 70.795
Louie Garrett 70.135
Viktor Marius 68.805

As Smyth had a monster season, alot with a lot of players up there as well.

One major theme that goes on(which is a weakness of GSVA), is that it is mostly skewed towards offensive productivity, and more two way players and defensive defenseman are negatively affected by this by lower scores. One obvious case is Mike Izzy, who has been a very consistent Darr nominee who has been hovering around the top 10, when he should probably be in the top 5 or so perennially.

Now on how to grade goalies, Dom’s model is pretty simplistic just using
Goalie Game Score = (-0.75 * GA) + (0.1 * SV)

I took that and did a bit more, as I had more adv stats(Ty Wilson again) so my formula goes like this

So my formula goes like this
Goalie Game Score = (-0.75 * GA) + (0.1 * SV) + (0.1*GSAA) +(0.1*QS) - (0.1*RBS) +(0.05 * RC%)

So if we want to test to see if it worked, we go to S48

The top 5 are as follows
Johannes Leitner 58.878
Kata Vilde 48.675
Cedric Robinson 45.0895
Beaujeaux Biscuit 39.7205
Sebastien Primeau 39.5795

So I think it's good to say that it's pretty accurate for this season. Calgary and NEW were the 2 powerhouses that had a strong goaltenders. Robinson was a good goalie is a weaker division + good defense around him. Biscuit is behind a strong WKP team, and Primeau has high TPE and is a good goalie in his own right.

Going to S48 you get this
Benjamin Blue 56.9905
Geezus Kryyst 47.495
Beaujeaux Biscuit 42.7215
Cedric Moreau 40.9275
Aleister Cain 37.375

As Blue had a monster season(every other season it seems like), as well as a strong performance by Kryyst. Biscuit is doing his usual, and Moreau had a decent Jets team around him, and Cain had a dominant NOLA team as well.

Going back into S47 we get this
Tommy Tuck 50.375
Jay Bae 48.605
Johannes Leitner 44.286
Kata Vilde 43.607
Beaujeaux Biscuit 39.3355

So that makes sense as Tuck was at his peak, and Jay Bae helped carry Chicago into the playoffs.

Now you may be asking on why I am going back 3 seasons for everything. So what I do is that I list everyone that is on every teams current rosters and pull the GSVA in the past 3 seasons(if they don't then they are just N/A for that season, giving longer players more weight). From that, I weigh each 3 seasons and weigh their TPE as a factor to give their Impact on the Game(IG) Score that I made for this.

So if you’re confused think of it as this
GSVA = Their Game Score for 1 season
IG = Impact on the game over 3 seasons + weighted TPE

So how do I weight the Impact on the game? And what do I get from that.

So lets take Dermot Lavelle for example, from Buffalo Sabers Stampede.

So the formula is this
IG =(0.1*Latest Season)+(0.05*iferror(2nd Latest season,0))+(0.03*iferror(3rd latest season,0))+(0.002*TPE)

So the reasoning for this, is that the latest season counts more as its recent one you played. The 2nd season the points is cut in half as you are 1 season removed, and then the 3rd season is cut by 66% percent as it is now 2 seasons removed. And it is .002 from the TPE, so 2k TPE is = 4 IG, and 1K = 2 IG. My reasoning is that if someone with Higher TPE would be more consistent throughout their game, so they should be a boost in IG a bit.

Now that being said, I am looking to further improve my formula and my ADV stat as time goes on, so I can get a more accurate reading of all the variables.

Spreadsheet link: https://docs.google.com/spreadsheets/d/1...sp=sharing

So if we go to Dermot Lavelle
Dermot Lavelle 1525(TPE) , 54.945(S49) , 60.215(S48) , 60.17(S47)

So the weight of his TPE is 3.05, S49 is 5.49, S48 is 3.01, and S47 is 1.8, so when you add them all up, you get 13.35 IG.

So from everyone's IG, I add up everyone all of the IG from Forwards, Defense, and Goalies, and compare each one.

So for Forward IG, it goes like this

TEAM - Forward Avg
New Orleans Specters - 118.3618
Manhattan Rage - 99.76085
Calgary Dragons - 95.91095
West Kendall Platoon - 95.1706
Tampa Bay Barracudas - 94.98595
Buffalo Stampede - 91.47395
Edmonton Blizzard - 91.43
New England Wolfpack - 90.93625
Los Angeles Panthers - 88.6117
Chicago Syndicate - 84.6826
San Francisco Pride - 82.5971
Winnipeg Jets - 79.4414
Toronto North Stars - 77.17645
Minnesota Chiefs - 69.80505
Hamilton Steelhawks - 69.0113
Texas Renegades - 54.7958

So first off, I think Manhattan is a bit higher than they should be, but the data is skewed slightly because of their monster season this season, which mostly came from their Forwards, as well as acquiring Herb Roberts as well. Texas and Hamilton doesn't surprise me being last(along with Minny) as they don't have a ton of great forwards, or many great combination of forwards to drag them up as well.

For Defense AVG you get this from the teams

West Kendall Platoon - 57.4514
New Orleans Specters - 54.6569
New England Wolfpack - 54.32225
San Francisco Pride - 53.44425
Buffalo Stampede - 50.9818
Calgary Dragons - 49.5898
Edmonton Blizzard - 48.8506
Winnipeg Jets - 46.90505
Manhattan Rage - 45.45
Texas Renegades - 45.2288
Tampa Bay Barracudas - 44.87835
Minnesota Chiefs - 43.3528
Chicago Syndicate - 37.568
Hamilton Steelhawks - 35.29175
Los Angeles Panthers - 34.0183
Toronto North Stars - 32.02875

One thing that I am surprised by is NOLA being as high as they are. One major reason is the acquiring of Bob Bergan who has 12.6866 IG over the past 3 seasons. From that, NEW had a really good defensive core, as well as Pride(they have all 6 defenseman over 1k, with 3 over 1.4k). It also makes sense that the Rage would be lower(around middle of the pack), as their offense really carried them to where they were. It makes sense that TOR would be as low as they are, as they only have 1 player over 10 IG, with Zander Rys. They also dont have any D with over 1k in TPE.

For goalies we get this
Calgary Dragons - 12.208558
New England Wolfpack - 11.913314
Tampa Bay Barracudas - 11.252471
Manhattan Rage - 11.167164
Hamilton Steelhawks - 10.345628
Buffalo Stampede - 10.175642
Los Angeles Panthers - 9.894379
Minnesota Chiefs - 9.451472
New Orleans Specters - 9.408274
West Kendall Platoon - 9.311495
Texas Renegades - 9.210449
San Francisco Pride - 9.005767
Edmonton Blizzard - 7.81353
Toronto North Stars - 5.2309
Chicago Syndicate - 4.94284
Winnipeg Jets - 4.07306

So, I think its safe to say the top 3 is helped by a strong goalie having monster seasons,with Kata Vilde on a great Calgary team. Hamilson has a strong goalie in Krryst. It makes sense that the jets IG for goalie to be so low, given that Leblanc is low TPE, but its looking better for Leblanc, as his GSVA is on a trend to move upward. One thing that doesn't sit right with me in where MAN is on this list, they are high because the way I do it, is that I sum the 2 goalies, and the sum of their two goalies were 11.16 IG. From Crawfling and Larson who both had 4.X and 7.X IG, so I think next season I need to think of something different for goalies. This also happens for SFP, so I need to find a way to make it balanced.

So from that, these are the top 10 for IG in each position, like Forwards, Defense, and goalie

Forwards - IG
Mikhail Lokitonov 17.192
DeMaricus Smyth 16.829
Mike Izzy 15.882
Flacko Lagerfield 15.752
Piotr Czerkawski 15.669
Anders Christiansen 15.517
Viktor Marius 15.379
Manuel Gotze 15.357
Jason Visser 15.303
Robert Phelps 15.291

Defense - IG
Tigole Bitties 13.715
GOD McZehrl 13.138
Otis B. Driftwood 12.971
Bob Bergen 12.687
Ensio Kalju 12.356
Ludwig Koch Schroder 12.220
Jon Toner 12.211
Brady McIntyre 12.130
Tor Tuck 12.125
Rex Kirkby 11.860

Goalie - IG
Kata Vilde 12.209
Johannes Leitner 11.913
Benjamin Blue 11.252
Geezus Kryyst 10.346
Cedric Robinson 9.828
Sebastien Primeau 9.451
Aleister Cain 9.408
Beaujeaux Biscuit 9.311
Walter Hobbs 9.210
Evgueni Marmeladov 7.822

So looking at all of the top 10, I think its pretty accurate holistically when it comes to it. It makes sense on why people are in the top 10 for forwards,as they are all players that are having monster seasons, and a great 3 season career. For Defense, it also makes sense that Bitties is #1, and then McZ, as they are 2 great players. For goalies, Kata Vilde is a premier goalie, and Leitner and Blue had monster seasons as well.

So my final thing is this Tableau comparing the teams, F vs D, D vs G, and F vs G.
https://public.tableau.com/profile/luke....ublish=yes

You can sort by conference there so you can look at them all in each division.

Conclusion:
From that, the conclusion is that this would be the this would be the rough standings for next season using IG
Team - IG
New Orleans Specters 182.4
West Kendall Platoon 161.9
Calgary Dragons 157.7
New England Wolfpack 157.2
Manhattan Rage 156.4
Buffalo Stampede 152.6
Tampa Bay Barracudas 151.1
Edmonton Blizzard 148.1
San Francisco Pride 145.0
Los Angeles Panthers 132.5
Winnipeg Jets 130.4
Chicago Syndicate 127.2
Minnesota Chiefs 122.6
Hamilton Steelhawks 114.6
Toronto North Stars 114.4
Texas Renegades 109.2

And this would be the split for the divisions
NorthWestern GS All
Calgary Dragons 157.7
Edmonton Blizzard 148.1
Winnipeg Jets 130.4
Minnesota Chiefs 122.6

Southwestern GS All
New Orleans Specters 182.4
San Francisco Pride 145.0
Los Angeles Panthers 132.5
Texas Renegades 109.2

Atlantic GS All
West Kendall Platoon 161.9
New England Wolfpack 157.2
Manhattan Rage 156.4
Tampa Bay Barracudas 151.1

Great lakes GS All
Buffalo Stampede 152.6
Chicago Syndicate 127.2
Hamilton Steelhawks 114.6
Toronto North Stars 114.4

So with SimonT being simonT, it probably wont be like that, but from what I can tell, this is pretty accurate of what it should be.

Next season after seeing how this season goes, ill try to see how I can improve my method and get more accurate results.

***bluesfan55*** · 09-15-2019, 12:04 PM

luke i love that you're bringing analytics into the SHL
but i don't love the fact that those analytics tell me how shit i am

ty to @High Stick King @EvilAllBran and @Ragnar for the sigs

awils13 · 09-15-2019, 12:48 PM

Good stuff Luke

C9Van · 09-15-2019, 01:04 PM

I love it Luke. You completely lost me after the second sentence but it looks important and factual so you I love it.

*Leppish* · 09-15-2019, 05:02 PM

Code:
Minnesota Chiefs - 69.80505

Nice

~~mahi01~~ · 06-15-2020, 07:13 AM

Go to Roku page roku activation record enter Roku com activation code appeared on Roku TV. My Roku com activate not working use new Roku code.

***luke*** · 06-15-2020, 08:30 AM

06-15-2020, 07:13 AMmahi01 Wrote: Go to Roku page roku activation record enter Roku com activation code appeared on Roku TV. My Roku com activate not working use new Roku code.

Did you try turning it off and turning it back on again

Navigation

Extra Menu

About us