2012-13 Ratings discussions

[Ville Isopaa]

I guess we might as well open up a new thread for the ratings. I'll give a short review of how the goalie ratings were done. I don't have time to answer why every single goalie is rated exactly how they are. Just look at the stats and you should find 99% of the answers of why your goalie is rated a 71 and not 84.

As there has been a lot of discussion on Brian Elliott I'll use him as an example.

In the past 4-5 years or so, a 41 and 10 game threshold was used to divide the goalies in to starters, backups and depth goalies as the EN wasn't working properly. That 41 game threshold was based on average games played of the past 3 seasons with a weight of 50-30-20% on the different seasons or 70-30% if the player played just 2 seasons. With Elliott playing 55 games in both of the previous seasons, anyone could calculate in january that under the system used in the past he needed to play 27 games to get an average of 41 games and his 38 games would have given him an average of 46,5 games.

After long discussions, the admin team decided to go with a new approach and use the stats of 164 games or for goalies last 2 seasons (i.e. last 164 games of their NHL teams). The split was the same as before with 70-30% weight on the two seasons. With that Elliott played an average of 43,1 games rather than the 46,5 games that he would have averaged in the old system. In the end he would have rated as a starter no matter if we use the 2 or 3 year stats or if we use DU to limit games played for goalies or the 41 and 10 game thresholds to divide goalies in to Starters, Backups and Depth goalies.

Moving on to the base of the ratings. In rating the goalies in the skill categories (SK, AG, RB, SC, HS, RT and PH), the stats used are 70-30% averages of the past two seasons, with the individual stats being Save percentage, Goals Against Average, Shutouts, Wins, Minutes Played, Vezina Votes and Team Defence (measured by ranking the teams by how many shots they allow during the season).

The goalies are then ranked in each category depending on how good they are compared to the 107 other goalies that played during the past two seasons and given a score of 1-107 in every category but Vezina, where only 14 guys recieved votes. The rest were given a 28 score on it instead of 107 which would have put too much weight on the Vezina votes.

To get a final ranking of the goalies in the skill ratings an average score was calculated for the different stat categories to create the final ranking of the goalies. Every position in the ranking equals a drop of 0,5 points in the average skill rating of a goalie compared to the goalie ranking ahead of him. With the help of a handbook on the strenghts of the goalies the rating points were distributed to every goalie to reflect their strenghts and weaknesses. When rating the goalies, the average ratings were raised slightly from last year, as the skater ratings had gone up from last year.

As for the EN ratings they have changed a lot since last year in a hope to limit the unnatural way that starters have been used in the past.
In the NHL the top goalies played 73 games, 3 goalies played 70 or more, just 13 played over 60 games and only 22 played over 50 games.
In the HFNHL we had 6 starters playing less than 70 games. Khabibulin 69 games, Quick 68, Boucher 63, Lehtonen 60, Varlamov 58 and Mason 57. The rest played 71 or more games.

The current EN ratings are still to be tested, so they could go up or down depending on the results of the tests, but the ranking of the goalies compared to eachother in the EN is based on minutes played during the past two seasons and is most likely what it will be in the end.

I'm not going to post the ranking of every stat, but here's how the goalies ranked up against eachother overall. This is not a final ranking of the OV's as there are other things like PS, SZ, EN, DU, etc. that also count towards the OV. This is just the ranking for the averages of the skill ratings.

You will find guys like Rask, Schneider, Holtby, Bishop and Khubodin ranked higher compared to where they would have been in the past. On the other side of the coin there are guys like Roloson, S. Mason and Khabibulin who would have been in the late 20's with the old system rather than in the 45-50 region. Remember that these "bad starters" will still be able to play more games at their highest level compared to guys like Rask, Holtby and Bishop who have lower EN's and will be limited to something close to what they played in the NHL over the past two seasons.

Rank Player
1 Henrik Lundqvist
2 Jon Quick
3 Pekka Rinne
4 Tim Thomas
5 Mike Smith
6 Roberto Luongo
7 Carey Price
8 Jaroslav Halak
9 Miikka Kiprusoff
10 Antti Niemi
11 Brian Elliott
12 Cam Ward
13 Marc-Andre Fleury
14 Ilya Bryzgalov
15 Jimmy Howard
16 Kari Lehtonen
17 Ryan Miller
18 Cory Schneider
19 Tomas Vokoun
20 Niklas Backstrom
21 Jonas Hiller
22 Jose Theodore
23 Tuukka Rask
24 Johan Hedberg
25 Semyon Varlamov
26 Martin Brodeur
27 Craig Anderson
28 Ondrej Pavelec
29 Devan Dubnyk
30 Evgeni Nabokov
31 Josh Harding
32 Corey Crawford
33 Braden Holtby
34 J. Giguere
35 Michal Neuvirth
36 Ben Bishop
37 Joey MacDonald
38 Scott Clemmensen
39 Martin Biron
40 Curtis Sanford
41 Jhonas Enroth
42 James Reimer
43 Jonathan Bernier
44 Anton Khubodin
45 Mathieu Garon
46 N. Khabibulin
47 Anders Lindback
48 Jonas Gustavsson
49 Robin Lehner
50 Steve Mason
51 Dwayne Roloson
52 Thomas Greiss
53 Jason LaBarbera
54 Matt Hackett
55 Chris Mason
56 Allen York
57 Richard Bachman
58 Peter Budaj
59 Sergei Bobrovsky
60 Martin Gerber
61 Al Montoya
62 Mike Murphy
63 Brent Johnson
64 Cederick Desjardins
65 Brian Boucher
66 Alex Stalock
67 Drew MacIntyre
68 Ray Emery
69 Anders Nilsson
70 Brian Foster
71 Justin Peters
72 Jacob Markström
73 Curtis McElhinney
74 Ty Conklin
75 Marty Turco
76 Rick DiPietro
77 Kevin Poulin
78 Dan Ellis
79 Ben Scrivens
80 Jeff Deslauriers
81 Antero Niittymaki
82 Henrik Karlsson
83 Andrew Raycroft
84 Alex Auld
85 Leland Irving
86 Pascal Leclaire
87 Chris Osgood
88 Peter Mannino
89 Sebastien Caron
90 Brad Thiessen
91 Dustin Tokarski
92 Nathan Lawson
93 Patrick Lalime
94 Mikko Koskinen
95 Mike McKenna
96 Michael Leighton
97 Iiro Tarkki
98 Jussi Rynnäs
99 Yann Danis
100 Mark Dekanich
101 Mike Brodeur
102 Matt Climie
103 David LeNeveu
104 Chad Johnson
105 Timo Pielmeier
106 Shawn Hunwick
107 Thomas McCollum


Now, if someone feels that they have the time to create a better system of rating the goalies and feels they have spare time to create those ratings I think everybody is open to new ideas.

Click to read more...

 

[Fooladelfia]

Nice explanation....but Again

Why Halak has a very low endurance rating 67?

Jaroslav Halak play 46+57+45= 148 games last 3 seasons and he has 67 in endurance

Corey Crawford play 57+57+1= 115 games last 3 seasons and he has 72 in endurance

 

[Ville Isopaa]

Nice explanation....but Again

Why Halak has a very low endurance rating 67?

Jaroslav Halak play 46+57+45= 148 games last 3 seasons and he has 67 in endurance

Corey Crawford play 57+57+1= 115 games last 3 seasons and he has 72 in endurance

Two year averages with 70-30% weights not three with 1/3 weight each. The EN was actually calculated with minutes played rather than games played to be more exact.

Jaroslav Halak played 46+57 = average of 49,3 games or an average of 2911,1 minutes which equals 48,52 games played of 60 minutes

Corey Crawford played 57+57 = average of 57 games or an average of 3253,7 minutes which equals 54,23 games played of 60 minutes

If you look at the goalies around the league you'll find that Halak's 67 is 21st among all goalies in the league and that Kiprusoff and Ward who are rated the highest are rated 85. Depending on how the test work out there could be changes to these levels, but the ranking is based purely on average minutes played, so Halak will be rated the 21st among all goalies in EN.

 

[PasiK]

damn you ville

you really stole Elliott from me, i thought 40games limit was last season only and i always talked that he will be rated lower than he should and suggested even new idea for you that limit should be lower like 30 games.

never heard a word that he will be rated like a started

I should be more careful
Jokerit - Hifk action in the next season between Sharks and avalanche

 

[Fooladelfia]

ok but next time the part that say : Two year averages with 70-30% weights not three with 1/3 weight each. The EN was actually calculated with minutes played rather than games played to be more exact.

Every gm should know that before no?? Cause I don't think Ville had traded you Elliott for Morrow a 4th and a 2nd.


BTW. Nice job for the ratings in general.

 

[Hossa]

Ben Bishop's ratings seem bizarrely high though, all things considered. Players who played more last year with better numbers are higher.

 

[Ville Isopaa]

Ben Bishop's ratings seem bizarrely high though, all things considered. Players who played more last year with better numbers are higher.

There seems to have been a type-o in the Bishop stats. He should rank 58th not 36th. I'll correct the ratings tomorrow.

Great catch!

Yes maybe but what's the Typo? i guess not size cause hes 6 foot 8 inches.

If you must know the goals against for last year was 12 not 22, which it was supposed to, ranking him too high in both GAA and SV%

 

[Fooladelfia]

There seems to be a type-o in the Bishop stats. I'll correct that tomorrow. Great catch!

Yes maybe but what's the Typo? i guess not size cause hes 6 foot 8 inches.

 

[Fooladelfia]

Question.

Stamkos has 97 in scoring and 156 goals in 3 years
Crosby has 94 in scoring and score 91 goals in 3 years

I know he play less games... but If a guy play 3 games during the NHL year and score 3 goals will he be rated at 99 in scoring....how forwards ratings work?

 

[Hossa]

There seems to have been a type-o in the Bishop stats. He should rank 58th not 36th. I'll correct the ratings tomorrow.

Great catch!

Makes sense. At first I only saw it as a disparity between Bishop and Bachman, but a closer look suggested Bishop was an anomaly, and apparently an error, because unlike other goalies who played small numbers of games, Bishop has never posted particularly impressive numbers.

 

[Fooladelfia]

Makes sense. At first I only saw it as a disparity between Bishop and Bachman, but a closer look suggested Bishop was an anomaly, and apparently an error, because unlike other goalies who played small numbers of games, Bishop has never posted particularly impressive numbers.

OK thats why you ask me to trade you Bishop for Bachman....If I have say yes no discussion about Bishop ratings will be going on right now....wow....I know now how it work.....

 

[Ohio Jones]

Question.

Stamkos has 97 in scoring and 156 goals in 3 years
Crosby has 94 in scoring and score 91 goals in 3 years

I know he play less games... but If a guy play 3 games during the NHL year and score 3 goals will he be rated at 99 in scoring....how forwards ratings work?

Skaters' ratings are based on the player's last 164 games, with the most recent 82 games counted at 70%.

 

[Ohio Jones]

OK thats why you ask me to trade you Bishop for Bachman....If I have say yes no discussion about Bishop ratings will be going on right now....wow....I know now how it work.....

Are you saying this is the first time you've been approached to make a trade based on ratings? Or that people only question their low ratings and other teams' high ratings? Because the former seems naive, and the latter seems, well, um, yeah: naive.

Shouldn't you have been the one asking why Bishop's ratings were so unusual?

 

[Fooladelfia]

Are you saying this is the first time you've been approached to make a trade based on ratings? Or that people only question their low ratings and other teams' high ratings? Because the former seems naive, and the latter seems, well, um, yeah: naive.

Shouldn't you have been the one asking why Bishop's ratings were so unusual?

Ben Bishop next big thing.
Ok amigo too long your question and Hedman say hi.

 

[Ohio Jones]

Ben Bishop next big thing.
Ok amigo too long your question and Hedman say hi.

Cute

 

[Fooladelfia]

Cute

That's cute ...imagine me....

 

[Ohio Jones]

That's cute ...imagine me....

They clearly take after their mother. Thank God.

 

[Hossa]

OK thats why you ask me to trade you Bishop for Bachman....If I have say yes no discussion about Bishop ratings will be going on right now....wow....I know now how it work.....

Two things. First, I asked about Bishop months ago, because my quota of Sens players is low (particularly as Nick is swallowing up every Sens prospect around). The interest was always there. And second, while I had noticed the ratings, it's Bishop's cheaper contract that I really wanted. The inflated ratings would have just been a bonus.

Not that I'd apologize if it was purely for the ratings though...

 

[Fan.At]

Just looked over the ratings and 2 guys that stood out for me were Garnett Exelby and Ryan Potulny... for as little games they have played in the NHL in the past 2 seasons, they seem to have rather high ratings - can anyone explain?

 

[Fooladelfia]

Two things. First, I asked about Bishop months ago, because my quota of Sens players is low (particularly as Nick is swallowing up every Sens prospect around). The interest was always there. And second, while I had noticed the ratings, it's Bishop's cheaper contract that I really wanted. The inflated ratings would have just been a bonus.

Not that I'd apologize if it was purely for the ratings though...

Bishop he's finish

 

[Dr.Sens(e)]

Just looked over the ratings and 2 guys that stood out for me were Garnett Exelby and Ryan Potulny... for as little games they have played in the NHL in the past 2 seasons, they seem to have rather high ratings - can anyone explain?

Ville actually mentioned these two guys in an e-mail discussion amongst the ratings group, and they are good catches.

The reason their ratings are still NHL-fringe worthy, is their last 82 games were pulled from the last 3 NHL seasons. Typically for players with low games played (<100), the games were in the more recent seasons. For Potulny and Exelby, the majority of the games (or all, in Exelby's case) were from the 09-10' season.

Now because we take the last 82 games (or 164), a guy has played during the three years, a guy like Potulny is still a 15 goal scorer, because even with his 0 points in 10 NHL games over the last two years, the weight of his 15 goals in 65 games or whatever it was in Edmonton, still drive the bulk of the weight of his rating. That said, his rating was reduced (as were all players) with fewer than 82 games over 3 years, so his rating is still lower than in past.

That said, this type of scenario probably justifies a bigger penalty or reduction, but there are only a few players who fall into this category, and given they are marginally rated anyway, we're likely to adjust the rating formula next year rather than try and go back and change it to account for just a few players like this.

 

[Dr.Sens(e)]

Question.

Stamkos has 97 in scoring and 156 goals in 3 years
Crosby has 94 in scoring and score 91 goals in 3 years

I know he play less games... but If a guy play 3 games during the NHL year and score 3 goals will he be rated at 99 in scoring....how forwards ratings work?

As Douglas noted, 70% of the SC rating is based on goals per game over a players last 164 games, with 70% of this weight on his last 82 games (and 30% on his 82-164). Another 30% of the SC weight is on goals per 60 minutes (70% weight on 1-82, and 30% on 83-164).

So it doesn't really matter how many games a player played in one season or another. As long as they played 164 games over the last 3 seasons, we based on their last 164 games (for things like SC - DU definitely gets impacted for missed games obviously).

And by this calculation, Crosby is quite close to Stamkos as a goal scorer. But a pretty decent gap after those two, as I recall.

 

[Hossa]

Has anybody else noticed that a fair number of players who registered more assists than goals have higher SC than PA ratings? I know I have a few on my team that fit the bill.

Obviously this makes sense at a certain point because, ceteris paribus, assists are easier to come by than goals, therefore I can see why PA and SC would be rated with slightly different scales relative to NHL production. However, one of the theories as to why scoring has been low in years past is that our old ratings were PA-heavy, and therefore few players actually had better SC than PA ratings.

It's something to watch for in our test seasons.

 

[Fooladelfia]

Can we counter some ratings? Cause look like Nate Thompson play with out skates...the guy dosen't skate...

 

[Ohio Jones]

Has anybody else noticed that a fair number of players who registered more assists than goals have higher SC than PA ratings? I know I have a few on my team that fit the bill.

Obviously this makes sense at a certain point because, ceteris paribus, assists are easier to come by than goals, therefore I can see why PA and SC would be rated with slightly different scales relative to NHL production. However, one of the theories as to why scoring has been low in years past is that our old ratings were PA-heavy, and therefore few players actually had better SC than PA ratings.

It's something to watch for in our test seasons.

Test seasons so far seem to make sense production-wise. One caveat, however, is that simming multiple seasons with the same ratings and settings does tend to produce substantial variations in the performances of players and teams. Randomness remains a significant factor in the sim - even with morale turned off. (In fact, morale may not not have been as much of a factor as we thought).

There are definitely no guarantees that you'll win on any given night, or on the season as a whole. But hopefully smart roster and line manager will improve your chances.