NAIHL Ratings - Currently Based on 2010-2011 NHL Season

[SimonT]

File Upload.

[NAIHLStars]

File Upload.

Thanks, you are fast.

[Kramden23]

Just wanted to say thanks again for all your hard work!

Next weekend, I will post some insight into how the ratings were put together.

When you get the chance I would be really curious to see how DF ratings for defensemen were put together. There are just a few defensemen who seemed to get "undeserved" high DF ratings so I'm just kind of curious.

Thanks again!

[NAIHLStars]

Intent of this document

To give disclosure of how the NAIHL ratings were derived. I think that is helpful for four primary reasons.

#1: To educate: I’ve been around SIM hockey for a long time and learned a lot about attributes and ratings from guys like Big Al from the fhlsim days and Dmaster of DVHL from both fhlsim and more recently STHS. I did my own ratings when Big Al stopped and then I stopped when DVHL was consistently putting out a quality set of ratings. I am grateful for what I learned from them. I will not do ratings forever (but for now I see no end) and hope that others can see what I have done and be able to continue producing quality ratings for the hockey sim community.

#2: Credibility with commissioners: I’ve run leagues and have helped run leagues. It is often a thankless task with a lot of work. Deciding what ratings to use and why is never an easy decision. Ratings packs vary greatly. In my opinion, the more information that raters can provide; the better it is for the commissioners in making a decision.

#3: Credibility with the ultimate end user, the GM’s: The more GM’s know about the ratings and how they are derived the better. In my opinion, it is critical that the ratings are done at an “arms length” with clear disclosure so that everyone understands that player x or player y is not artificially rated higher because of the raters opinion. The formula’s will not always produce the 100% exact match for a particular attribute, but it gives a consistent approach that hopefully the GM’s can understand and ultimately accept.

#4: Information for the GM’s: If the GM’s understand what information is pulled to derive the ratings, a savvy GM can follow NHL stats and make trades that could benefit their team in the future. Providing this document provides a level playing field for all GM’s in how they are done. Now the talented and high energy GM’s can use this information to their benefit.


General Principles

These are the general principles that are used in producing the NAIHL ratings:

#1: They are done at an “arms length” without manual adjustments for individual players.

#2: They are NHL stat based using the prior year NHL regular season. Playoff stats are not used in the calculation. Only the prior year stats are used. No accommodation is used for a veteran player who had an off year. Only players that played in at least one NHL game are rated.

#3: They follow the tradition set by Big Al and later followed by Dmaster. For things like SC, PA, and DF 80’s are rare and hard to come by. 90’s are even more rare. They are not like modern video game ratings where there are a lot of 90+ ratings.

#4: Consistency: It is important to have consistency from year to year. Therefore the approach will not change from year to year. Small tweaks in the tables (which tell you what 30 goals in the NHL convert to) may be made if the prior year had unexpected results that were inconsistent with the NHL. It should be noted that any tweaks made in the tables, will impact all the players in a similar manner.


General Steps in the Process

#1: Stats are downloaded from the NHL website and dropped into excel. The following tables were downloaded: Skaters: Bios, penalties, real-time stats, time on ice, and shootouts. Goalies: Bios, summary, special teams, shootouts. Each table is given it’s own tab in excel. The NHL website only does 30 rows at a time, so there is a lot of copy and paste special that is done. I’ve found other website that allows you to do more at once, but those websites do not have all the stats that are needed to do a good ratings set and often have names that slightly differ from the NHL set, so it then becomes a hassle trying to use non-NHL website and NHL website stats combined into one. Using just NHL website based stats makes things simpler in the long run.

#2: A combined stats table is made from the individual tables above. You cannot assume the NHL stat tables are in the same row order, so it is imperative that it is checked so you do not get a various stats criss-crossed among players.

#3: Next, most of the accumulated stats are converted to an 82 game basis. In order to avoid anomalies among players with few games the following formula was used: =IF($G2>40,H2/$G2*82,IF($G2>20,H2/$G2*82*0.9,IF($G2>10,H2/$G2*82*0.8,IF($G2>5,H2/$G2*82*0.7,H2/$G2*82*0.3)))) . This basically reduces that 82 game proration by 10% for those between 21-40 games, 20% for 11-20 games, 30% for 6-10 games, and 70% for 5 or less. This stops the 1 game player with 1 goal and 1 assist from looking like a 82 goal, 82 assist, 164 point player.

V1 age, V2 birthdate: NHL stats downloads a birth date in one cell. The date cell will look like this, “Apr 26 '77” Use the “data text to columns” function in excel to split the 1 cell into 3 cells. This will make your year 77. use another column to be +1900+ the year column to make the year 1977 (for V2). For V1 then have another column be +2010-your 1977 column.

#4: Tables are created that convert NHL stats into NAIHL attributes. The lookup function is then used to reference the NHL stat to the table. (example =LOOKUP(G3,Tables!$O$4:$O$49,Tables!$P$4:$P$49)*0.



Skater Ratings

There are many different ways to determine what a particular attribute should be and how to use stats to calculate it. It is not an exact science. Attributes like scoring is fairly easy to do with little debate as goals scored would be a good stat to use for scoring. Others like defense are more debatable and could be done many different ways. The purpose of disclosing what is used is not to spur debate on other alternatives. It is simply to disclose what is being used.


CK – Checking: 80% from a table that uses NHL hits and 20% from the ST rating. The thought behind the ST rating being 20% is that a check will be more effective from a larger player. However, 80% still comes from the # of hits. Some data points from the table for the 80% part from NHL hits are: 40 hits = 60, 90 =70, 140=80, 215=90

FG – Fighting: 20% from the discipline rating, 10% from the strength rating and 70% from a % of major penalty minutes to major plus minor penalty minutes.

DI – Discipline: 100% from pro-rated penalty minutes. Sample data points on the table are: 10 PIM=90, 30 PIM=80, 65 PIM=70, 100 PIM=60, 120 PIM=50, 150 PIM=40, 180 PIM =30.

SK – Skating: This is one that is difficult to quantify from stats. For this stat, we look at what other ratings have used and use an average and if necessary curve it up or down.

ST – Strength: 80% from weight + 20% height. Sample weight data points are: 180 lbs=60, 200=70, 220=80, 240=90.

EN – Endurance: 100% time on ice per game with separate tables for forwards and defense. Sample data points are: Forwards 7m=60, 12m=70, 17m=80, 22m=90. Defense 10m-60, 17.5m=70, 22.5m=80, 27.5m=90.

DU – Durability: V2: 100% GP, V1 70% GP + 30% TOI. Sample data points are: 20GP=50, 40GP=60, 56GP=70, 66GP=80, 76GP=90, 78GP=95

PH – Puck Handling: 20% passing + 20% scoring + 60% ratio of giveaways to points. Sample data points on the giveaways to points ratio table are: 0.38=90, 1.0=80, 2.0=70, 4.0=60

That is enough for tonight ...... I will pick up with FO tomorrow

[NAIHLStars]

A continuation of the previous post on skater ratings. Goalies will be posted tomorrow.

FO – Face Offs: 50% # of face offs taken + 50% face off %. Sample data points are: 350 FO = 60, 750=70, 1250=80, 1750=90. 45% =60, 50% =70, 55% = 80, 60% = 90.

PA – Passing: 100% assists. This year we made the D table much closer to the forwards. Common data points are: D 10a=60, 22.5=70, 40=80, 60=90. Forwards 12.5=60, 27.5=70, 45=80, 65=90

SC – Scoring: 100% goals. Again, this year we tightened the range between the tables. Common data points are: D 5g=60, 10g=70, 25g=80, 40g=90. Forwards 10g=60, 20g=70, 35g=80, 55g=90

DF – Defense: 40% SH TOI + 30% +/- + 30% blocked shots + takeaways. Sample defense data points are 40% SH TOI/game 0.5m=60m 1.75m=70, 3.25m=80, 4.25m = 90, +/- -22.5=60, -7.5=70, +12.5=80, + 32.5=90. BS+TA 100=60, 150=70, 175=80, 200=90. Forward data points are: SH TOI/game and +/- Same as the D. BS+TA 30=60, 77.5=70, 102.5=80, 127.5=90.

For an earlier question about some surprises on some unexpected high DF d-men it is probably coming from high SH TOI/game and a combination of blocked shots and take aways.

EX – Experience: This is one we lift from other ratings that have already been done.

PS – Penalty Shots: 75% shootout stats + 25% average of the PH/SC rating. This took some time to do, because not every player took a penalty shot, so getting the data to line up was a trick. I used this formula to pull names from shootout data onto the sheet “ =LOOKUP(B3,shootout!$A$2:$A$314,shootout!$A$2:$A$314) “. Then in the next column a true/false test to see if there was a match “=IF(BE3=B3,1,0) “. The next column to pull data from the shootout table =LOOKUP(BE3,shootout!$A$2:$A$314,shootout!$L$2:$L$314) . Next a formula to use the data if there was a match and use a 50 if no match =IF(BF3=1,BG3,50). For shootouts points were given for the following: a goal = 2 points, a game deciding goal = 1 point, a shot = 0.5 points. Sample data points on the table were: 3 points = 60, 10.5=70, 18=80, 26.75=90.

LD – Leadership. Another one we pirated.

PO - Our league does not use this. We just set them all at 70.

[tank7]

great summary, and thanks again for your work. Using your ratings in my league! Thanks.

[NAIHLStars]

We have issued a revision to our ratings, which should be available shortly when Simon updates the downloads. The adjustments that were made are:

V1: Adjusted DF ratings for forwards (see below).

V2: Fixed the PS and EX fields being criss crossed from the original english version instructions for V2.
Cleared the bad hyperlinks.
Adjusted DF ratings for forwards (see below).

The adjustment for DF for forwards was made as there was a general observation about the ratings being lower than in the past. The adjustment were:

Old Adjust New
57 1 58
58 1 59
59 2 61
60 2 62
61 3 64
62 3 65
63 3 66
64 4 68
65 4 69
66 4 70
67 5 72
68 5 73
69 4 73
70 4 74
71 4 75
72 3 75
73 3 76
74 3 77
75 2 77
76 2 78
77 1 78
78 1 79

[NAIHLStars]

Here goes information on the goalie ratings.

Goalie Ratings

Goalie ratings are more difficult then skater ratings in terms of matching specific attributes against a NHL statistic. For example goals are easy to correlate with scoring. What NHL stat can be used for hand speed or agility or reaction time? Also how do you rate the GAA of a goalie playing 10 games as a back up versus a NHL 65 game starter? Another problem with many rating packs is that an average of many NHL stats is used and then many of the goalie attributes are the same for the player across the board. Simon created 14 different attributes for goalies and it was my goal to make each attribute unique and not have across the board numbers for goalies. What was created below is not perfect, but it will give ratings based on NHL stats with unique attributes for each goalie that should make things interesting for Gm’s.

Concept #1: Generally the goalies who play more games are judged to be the best in the game and more highly regarded than goalies that play a split role in 40 games per year or a backup role of 20 some games per year. However, many of the stats like GAA or Save % can be the same or similar. Therefore, for each of the tables, the tables were broken out with the following ranges: Games played 66+ from 70 to 95, 57-65GP 68-91, 49-56GP 66-87, 41-48GP 64-84, 33-40GP 62-80, 25-32GP 60-76, 16-24GP 58-73, 8-15GP 56-70, and 1-7GP 54-67.

Concept #2: An overall average or rating needed to be calculated, but we did not want to fall into the trap of having every attribute for a goalie being the same. Therefore, we calculated an overall average and then for each attribute used 30% of the overall average and 70% of a unique statistic.

SK – Skating: For skating, we used the overall composite for this rating. The components that we used for this was from shots against per game, winning percentage with a bonus for shutouts, save percentage, short handed save percentage, and goals against average.

DU – Durability: This stat was based on games played.

EN- Endurance: The following formula was used for endurance: =ROUND(MAX(O3,(70+O3)/2),0) where column O was the durability calculation. The second part of the formula bumped up the endurance rating for goalies with durability under 70. The concept or thought was that even a goalie that only played a few games would have an endurance capable of completing a game or two.

SZ – Size: This was taken from 50% height and 50% weight.

AG – Agility: 30% was taken from the composite rating. 70% was taken with the following formula =MIN(90,AE3*2-AH3) where AE is the composite rating and AH is the SZ rating. In essence, it worked opposite SZ with the assumption that 2 goalies with the same stats, but unequal size would mean the smaller goalie would need higher agility to match the larger goalie. For example 2 goalies with an equal composite of 75, but sizes of 80 and 70 would have the SZ 80 goalie with a AG of 70 and the SZ 70 with an AG of 80.

RB – Rebound Direction: 30% from the composite rating and 70% from a derived stat of 60-shots against per game/60. In essence a stat using inverted shots against to work in an excel lookup function. The thought behind a correlation of shots against and rebound direction is that a goalie with better rebound direction/control will generally give up less shots per game.

SC – Style Control: Wow, what is style control? We used the 30% composite and derived a new stat for the remaining 70%. The stat combined possible points from a W-L record and added in a bonus for shootouts. Possible points took (w+l+OTL) *2. Points won took W*2+OTL*1. Then 1 bonus point was added for each shutout. The total points was then divided by possible points to derive a percentage that was applied against a table. In essence a combination of winning percentage and shutouts gave a goalie style points.

HS – Hand Speed: 30% from the composite and 70% from save percentage. The thought was a high save percentage would correlate to hand speed.

RT – Reaction Time: 30% from the composite and 70% from special team save percentage when the goalies team was short handed. The thought was high reaction time was needed for a higher save percentage when short handed.

PH – Puck Control: 30% from the composite and 70% from goals against average. The thought was that keeping the puck out of the net is the ultimate in puck control.

PS – Penalty Shots: 30% from the composite and 70% from shootout save percentage.

The above it not perfect, but it does provide for unique attributes for each goalie with statistics that can somewhat correlate with various attributes. In the end, the SK rating provides as a composite that is used in 30% of the other ratings to keep some sort of overall tendency.

In the end, Ryan Miller had the best composite, Bryzgalov was #2, and Brodeur #3. This amazingly mirrored the Vezina voting for 2010.

[SimonT]

File Available on Website.

[Kramden23]

Just wanted to point out two minor errors I noticed.....

Zach and Johnny Boychuk have their positions reversed, and Colin Wilson is listed as 26 years old when he is actually set to turn 21 in October.

[shamrocks]

Great work on the ratings and providing the relevant information. Glad to see you dedicating your time to not only your own league but for leagues that use your ratings. Keep up the great work.
Thanks.

[shamrocks]

Just wondering about a few Goalie ratings. Chris Mason is rated 70OV while Jimmy Howard who had an outstaning year with similiar numbers only has a 50Ov and Howard had a better GAA. Other goalies I'm concerned about are:
Howard, Lehtonen, Niemi. Their numbers just look to low compared to other netminders.
Thanks for hearing me out.
Other from this the ratings are wicked and I appreciate the time and effort put in doint them.

[NAIHLStars]

Just wondering about a few Goalie ratings. Chris Mason is rated 70OV while Jimmy Howard who had an outstaning year with similiar numbers only has a 50Ov and Howard had a better GAA. Other goalies I'm concerned about are:
Howard, Lehtonen, Niemi. Their numbers just look to low compared to other netminders.
Thanks for hearing me out.
Other from this the ratings are wicked and I appreciate the time and effort put in doint them.

What file are you using? I'm showing Jimmy Howard no where near an OV 50.

[shamrocks]

Just wondering about a few Goalie ratings. Chris Mason is rated 70OV while Jimmy Howard who had an outstaning year with similiar numbers only has a 50Ov and Howard had a better GAA. Other goalies I'm concerned about are:
Howard, Lehtonen, Niemi. Their numbers just look to low compared to other netminders.
Thanks for hearing me out.
Other from this the ratings are wicked and I appreciate the time and effort put in doint them.

What file are you using? I'm showing Jimmy Howard no where near an OV 50.

Shit sorry, not sure who i was looking at. Howard is showing to be 68 which makes sense. But Lehtonen is a 56 though, is that because he only played 12 games last season?

[tank7]

are there gonna be ratings for 2010-2011?