Réalisme des résultats dans vos ligues

This forum is for league commissioners seeking advice when it comes to league related issues; rules, free agency system etc. / Ce forum devrait être utilisé par les commissaires qui cherches des conseils pour les aider à rouler leur ligue; règlements, systême agent libre etc.
Savard
The Accomplished One / L'Accompli
Posts: 119
Joined: Fri May 24, 2019 1:07 am

Re: Réalisme des résultats dans vos ligues

Post by Savard »

@Kitsune: I see in your league the best offensive players also have high DF ratings. Is that on purpose to make the ratings work or is it part of how you created these fictional players? If it's the former, it kind of goes with what I found out, and it is killing me. I mean, yes, I could give McDavid a 85 in DF to make him a dominant scorer, but doesn't that completely shift the balance and a offensively loaded team suddenly is also a defensive powerhose? And how can I make a distinction between him and let's say a Selke candidate who can score AND actually play good defense? Or a defensive/PK specialist? *sigh*

The "inverted SK" phenomenon is also something that is hard for GMs to accept. It could be countered though if the engine wouldnt compare PA, SC and scoring for the decision making process, but instead compare PA, SC and a fixed value. Or instead of a fixed value the inversion of PA or SC or PH. Creativity doesn't come from low skating, but from high PA/SC/PH...
Kitsune
The Addict / Le Drogué
Posts: 555
Joined: Tue Jun 20, 2006 7:23 pm
Location: Toronto
Contact:

Re: Réalisme des résultats dans vos ligues

Post by Kitsune »

It was on purpose .. and DF is a bit of an enigma for me too. I think the way 2.1 ratings should be setup is that your average guy is 50OV-ish, with second liners into the mid 60s, and first liners upper 60s into the low 70s. The Crosby / Mcdavid / etc would be upper 70s into the low 80s. I never did test this though.

.. another thing I noticed is goalie fatigue played a part too. Even if they are at 99.. I noticed a huge jump in the goals given up. I ended up having to disable it. Also, I generated the players in 9 different tiers:

Image

.. goalies I noticed 4 tiers worked best.

Due to the random nature.. its also possible this set may have skewed a bit higher on DF then the previous sets I generated (note the X's are because those are formula based). I tested a few different sets with the same sliders setting and determined this was the best (granted I did notice some odd Farm seasons .. ie the guy with 72 goals and 8 assists in 64 games).
Savard
The Accomplished One / L'Accompli
Posts: 119
Joined: Fri May 24, 2019 1:07 am

Re: Réalisme des résultats dans vos ligues

Post by Savard »

interesting. so you first put each player into one of your tiers, and then manually tweak PH, PA, SC etc. within the HI and LO indicated? (edit: wait. I fogot this is for fictional plyers ;) )

my approach from last year was creating 10 random rosters and then simming auto seasons, looking at the average scoring and other stats per 20 min and then (automatically in Excel) tweaking SC, PA, CK, FG hoping to make them more NHL like. Results were ok sometimes, but some players in certain line combinationss over- or underachieved horribly. High CK or DF ratiings seeminngly playing some part in it. That however was before I learned about PH being some kind of accuracy rating. I really hope to resume my tests this summer after I finished working on our website...

Sorry, I think we kind of kidnapped this thread for 2.1 ratings discussion. Maybe we shoudl create a 2.1 ratings thread where different ideas and findings could be discussed?
Kitsune
The Addict / Le Drogué
Posts: 555
Joined: Tue Jun 20, 2006 7:23 pm
Location: Toronto
Contact:

Re: Réalisme des résultats dans vos ligues

Post by Kitsune »

yeah .. it was generate.. tweak ... update the tiers .. and repeat 20 times. I then spent another 10 test seasons all with the same ratings and sliders and just regenerating the players (as I was using the same Height / Weight / Country / Name.. no macro was required, ratings generate off of vlookup and randbetween). Sim rerate's dont seem geared to 2.1, as they all want to hike SK by +5/the max allowed. Thats why I had to go custom rerates. Sad the league got 0 interest though..
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Réalisme des résultats dans vos ligues

Post by 36Henry »

Savard wrote: Sun Jul 05, 2020 11:14 amThe "inverted SK" phenomenon is also something that is hard for GMs to accept.
Usually that is down to them thinking SK is something it's not :)

I will give a very brief example of how and why I use the "inverted SK" in my ratings.

I will start this off by using two players in this example. One of them is Wayne Gretzky, you may have heard of him. The other one is one of my all-time favorites: Jimmy Mann. If you don't know him, click on the links below as March 28th 1980 is a date everyone should know.

https://www.youtube.com/watch?v=SSla4YgFP3I
https://www.youtube.com/watch?v=Bf70Cpb2u1E

Jimmy Mann played a total of 293 NHL games, scoring 10 goals and 20 assists.
Wayne Gretzky played a total of 1487 NHL games, scoring 894 goals and 1963 assists.

Suppose we were to create these two players in the Sim and we do that as follows (this is based on a simplified version of the decision making formula):

Jimmy Mann - 50 SK, 50 PA, 50 SC
Wayne Gretzky - 99 SK, 99 PA, 99 SC
STHS Manual wrote:So, a players with 99 in the 5 skills has the same % of chance to shot to the net that a player who has 50 in the 50 skills.
Applied on our examples above we get:

Chance of a shot:

Jimmy Mann - 50 (SC) / 150 (SK+PA+SC) = 33%
Wayne Gretzky - 99 (SC) / 297 (SK+PA+SC) = 33%

Chance of a pass:

Jimmy Mann - 50 (PA) / 150 (SK+PA+SC) = 33%
Wayne Gretzky - 99 (PA) / 297 (SK+PA+SC) = 33%

Chance of the player skating with the puck:

Jimmy Mann - 50 (SK) / 150 (SK+PA+SC) = 33%
Wayne Gretzky - 99 (SK) / 297 (SK+PA+SC) = 33%

Here it becomes important to note that skating with the puck means just that. The player will go full Rico Fata (never go full Rico Fata), bury his head and start skating around with the puck until someone takes it away from him. He wont be skating around looking for someone to pass to, or to get himself into a position to shoot the puck. Skating in the STHS is the most accurate definition of "dead-puck" there has ever been.

In other words, skating is in fact the antithesis of a creative play.

Thus, we can define creativity as the act of either passing or shooting the puck. This is key.

Chance of a creative play:

Jimmy Mann - 100 (PA+SC) / 150 (SK+PA+SC) = 66%
Wayne Gretzky - 198 (PA+SC) / 297 (SK+PA+SC) = 66%

In other words, these two players are equally creative(!) using these ratings. Now obviously, PH and the level of PA and SC will make Gretzky more adept and accurate with his plays, but this doesn't change the fact that these two players are rated to have the same level of creativity (or hockey sense), which is rather poor when measured in terms of realism.

So, how does inverted SK factor into this? Well, if we want to make certain players more creative than others, we can do that using SK.

In this example we tweak just one thing, the SK of Wayne Gretzky from 99 to 10.

Jimmy Mann - 50 SK, 50 PA, 50 SC
Wayne Gretzky - 10 SK, 99 PA, 99 SC

Chance of a shot:

Jimmy Mann - 50 (SC) / 150 (SK+PA+SC) = 33%
Wayne Gretzky - 99 (SC) / 208 (SK+PA+SC) = 48%

Chance of a pass:

Jimmy Mann - 50 (PA) / 150 (SK+PA+SC) = 33%
Wayne Gretzky - 99 (PA) / 208 (SK+PA+SC) = 48%

Chance of the player skating with the puck:

Jimmy Mann - 50 (SK) / 150 (SK+PA+SC) = 33%
Wayne Gretzky - 10 (SK) / 208 (SK+PA+SC) = 5%

Chance of a creative play:

Jimmy Mann - 100 (PA+SC) / 150 (SK+PA+SC) = 66%
Wayne Gretzky - 198 (PA+SC) / 208 (SK+PA+SC) = 95%

So suddenly, we now have a player in Wayne Gretzky who will be an offensive threat nearly everytime he touches the puck in the offensive zone. Meanwhile, Jimmy Mann will remain rather scrublike in terms of creativity (but if you invite me to join a 1979-80 NHL retro league, I'm picking Mann over Gretzky every time).

Things are, obviously, a bit more complicated than this, but in terms of the reasoning behind inverted SK, this example illustrates things rather well. Plus, it gives us a chance to talk about Jimmy Mann.

Image
Image
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Réalisme des résultats dans vos ligues

Post by 36Henry »

As for the "DF influencing offense" debate: No it doesn't.

There are other factors at play behind these stats:

Code: Select all

Joueur------------STATS--------------COTES----------------C----------LW-----------RW-----------D----------D------
John Tavares:     68GP-43G-42A-85P / 92SK 85PA 91SC 72DF (Tavares  - Kuznetsov  - Scheiffele - Reilly   - Slavin)
William Karlsson: 66GP-36G-44A-80P / 80SK 76PA 78SC 82DF (Karlsson - Richardson - Burakovsky - Theodore - Demers)
Zach Parise:      68GP-31G-31A-62P / 82SK 76PA 78SC 70DF (Konecny  - Parise     - Foligno    - Dunn     - McNabb)
----
Alex Ovechkin:    65GP-27G-21A-48P / 92SK 83PA 95SC 67DF (O'Reilly - Ovechkin   - Landeskog  - Carlson  - Hamonic)
Sidney Crosby:    68GP-25G-33A-58P / 95SK 97PA 85SC 71DF (Crosby   - Pavelski   - Dadonov    - Doughty  - Burns) 
Image
SimonT
STHS Owner / Propriétaire du STHS
Posts: 14756
Joined: Sat Oct 08, 2005 4:18 pm
Location: Montreal, Canada
Contact:

Re: Réalisme des résultats dans vos ligues

Post by SimonT »

I'm starting to think 36Henry real name is Jimmy Mann. ;)

I want to quote a phrase in the manual about 36Henry comment about SK.

Code: Select all

* When a player has the puck, he needs to make a decision of what to do with it. There are 3 results possible: skate, pass or shot. It doesn’t matter if the player skills are at 50 or 99 for this formula. The player’s skills are compared between each other’s for this formula.
Source : http://sths.simont.info/ManualV3_En.php ... reviations
-SimonT
Forum Administrator / Administrateur du Forum
STHS Owner / Propriètaire du STHS
English V2 & V3 Manual - Manuel V2 & V3 Français
Savard
The Accomplished One / L'Accompli
Posts: 119
Joined: Fri May 24, 2019 1:07 am

Re: Réalisme des résultats dans vos ligues

Post by Savard »

Yeah, I am aware of how the decision making process works and why a low SK rating results in a creative player. However as 36Henry says, GMs think SK is what it in reality isn't. That's a problem. And in my opinion it isn't the GMs' problem but it the engine's problem. For every single rating the logic is "the higher the better", except for SK. That is very hard to explain to GMs. And simply put isn't very logical.

If from the decision to skate there never happens anything "creative", and it is simply "delaying" the next pass or shot at the risk of being stripped of the puck, there are probably two ways to kind of fix this:

1. Make the act of skating more important to offensive play. (I don't know enough about the engine to be more specific)

2. Do not factor in SK into the decision making process. SImply put: good skating doesn't equal uncreative play. Instead of using SK, the engine could for example always use a fixed number like 60 as a placeholder for what SK is used for today (in the decision making process only). Or - probably much better - the fictional SK would be dependent on SC and PA, in a formula like 100-(SC+PA)/2. That way a good PA and SC automatically results in more creative play. Good skaters are't punished for a high SK.

I can already see Simon rolling his eyes ;) I know you don't want to start over with a new engine, so I do have a third option that may be implemented quite easily:

What if the SK in the decision making process (and only there) is inverted. So a player with a 90 in SK has 90 used in whatever formula uses SK (avoiding hits for example), but when SK is used for the decision making process, the engine "inverts" it using a formula liek 100-SK so in our example 10 is used. That doesn't fix all the problems I have with the concept, but the ratings would look a lot more consistent, SK can be part of a OV formula if someone wants to use it, GMs aren't confused etc. I have no idea how hard this would actually be to implement or what side-effects I didn't consider, but maybe this could be a simple way of improving the 2.1 engine.
SimonT
STHS Owner / Propriétaire du STHS
Posts: 14756
Joined: Sat Oct 08, 2005 4:18 pm
Location: Montreal, Canada
Contact:

Re: Réalisme des résultats dans vos ligues

Post by SimonT »

Any change of the simulation engine is no go for me. It would mess up all the current rating. You'll have to live with what you got.
-SimonT
Forum Administrator / Administrateur du Forum
STHS Owner / Propriètaire du STHS
English V2 & V3 Manual - Manuel V2 & V3 Français
Savard
The Accomplished One / L'Accompli
Posts: 119
Joined: Fri May 24, 2019 1:07 am

Re: Réalisme des résultats dans vos ligues

Post by Savard »

I wouldn't want you to change 2.1 obviously, but you could instead create a 2.2 with said changes. Then people have the choice.

But I don't want to push you too hard. STHS in most aspects is a great product at a ridiculously low price (you could easily triple it imo) but the rating/engine relationship (for 2.1) is badly flawed, which is frustrating.
36Henry
The Accomplished One / L'Accompli
Posts: 118
Joined: Wed Dec 15, 2010 1:58 am

Re: Réalisme des résultats dans vos ligues

Post by 36Henry »

How about just explaing to managers how SK works? It's not rocket science.
Image
Savard
The Accomplished One / L'Accompli
Posts: 119
Joined: Fri May 24, 2019 1:07 am

Re: Réalisme des résultats dans vos ligues

Post by Savard »

That would be too easy ;)
Kitsune
The Addict / Le Drogué
Posts: 555
Joined: Tue Jun 20, 2006 7:23 pm
Location: Toronto
Contact:

Re: Réalisme des résultats dans vos ligues

Post by Kitsune »

.. I guess I would of been better to launch with 1.5? oy.
SimonT
STHS Owner / Propriétaire du STHS
Posts: 14756
Joined: Sat Oct 08, 2005 4:18 pm
Location: Montreal, Canada
Contact:

Re: Réalisme des résultats dans vos ligues

Post by SimonT »

Kitsune wrote: Wed Jul 08, 2020 4:33 pm .. I guess I would of been better to launch with 1.5? oy.
All the simulation engine use the same logic. The formula is a little bit different but the logic is the same on all engine.
-SimonT
Forum Administrator / Administrateur du Forum
STHS Owner / Propriètaire du STHS
English V2 & V3 Manual - Manuel V2 & V3 Français
Savard
The Accomplished One / L'Accompli
Posts: 119
Joined: Fri May 24, 2019 1:07 am

Re: Réalisme des résultats dans vos ligues

Post by Savard »

I really wonder if there is a way of creating 2.1 ratings without "sacrificing" the SK rating. Maybe if the SK, PA, SC trio is kept in a more neutral area (70 max) with PH being the decisive factor offering a bigger range of outcomes? Anyone ever tried this?
Post Reply