Power Rankings By DL44

[DL44]

Here's my Power Ranking which involves the following categories:



H - Reg/OT Win %
J - Points
L - Point %
N - Last 10
P - Goals/gm
R - Goals A/gm
T - Goal differential
V - 5-on-5 Ratio
Z - Special teams (PP%+PK%)
AD - Shot differential
AF - Faceoff %

Sorry i don't have an end season calculation... just one i did in the middle of the season after about 33-36 gms.

But i got some pretty good reviews and will be picked up by website to be published on an ongoing basis this season...

Any suggestions on how you would weight the categories?

Click to read more...

 

[seventieslord]

T just repeats P and R.
J and L are pretty much just repeats of eachother.
also, whichever one you choose to use is ultimately just a "symptom" of P, R, V, and Z, so you're just doubling down by using it.
not sure AF is relevant.

 

[DL44]

T just repeats P and R.
J and L are pretty much just repeats of eachother.
also, whichever one you choose to use is ultimately just a "symptom" of P, R, V, and Z, so you're just doubling down by using it.
not sure AF is relevant.

I just love the faceoff stat. I consider it one of the more underrated stats that effects possession and critical in some key situations.

J & L do repeat at the end of the season... but proves very effective in season when games played are all over the map... i.e. A team gets props for having points in the bank and also for what they are on pace to do. i.e. going with only one of them would be misleading in numerous cases at any point in season.

T repeating P & R - yeah i guess it does. I started with Goal diff, but wanted to reward strong offensive and defensive teams as well - so i added G and GA... I then i noticed the tiering of teams (we talked briefly about that in the defensive thread) just 30 teams, and could almost group teams of 5 etc... the differential seemed to slightly reward those teams.. and it was easy enough include, and then its a matter of playing with ratios.

As for the symptom line.. Yeah..
G & GA encompass both 5-on-5 and special teams goals... but i don't really touch SP goals... just the % of both units. and since +/- is such a notable on its own, i gave 5-on-5 ratio a category since the majority of game is played at that level...
That's where weighing also plays a factor..

 

[seventieslord]

J & L do repeat at the end of the season... but proves very effective in season when games played are all over the map... i.e. A team gets props for having points in the bank and also for what they are on pace to do. i.e. going with only one of them would be misleading in numerous cases at any point in season.

points can be misleading especially earlier in the season, but why would points% ever be misleading? I say just go with that and forget points altogether.

 

[DL44]

points can be misleading especially earlier in the season, but why would points% ever be misleading? I say just go with that and forget points altogether.

Because points in the bank are always more valuable then what a team MIGHT get...
It's promising, but not guaranteed... they get points for their points pace... but having the points in the bank is always more valuable..

A: 23-10-0, 46pts, 0.697
B: 25-12-0, 50pts, 0.676

No one knows how A will run in their next 4 games... we think they'll run great: so they get rank pts for Point%.
But B already has 50 in the bank... so they score for the points they already have.

Partially redundant or not... its still superior having both part of the formula than just having one of them.

What's con?
an extra column? spreadsheet does the work anyway.

 

[Catamo]

I just love the faceoff stat. I consider it one of the more underrated stats that effects possession and critical in some key situations.

Face off Stat is completely irrelevant as the amount of effect it has on a teams success can already be found in the teams points and goal differential. Including it would be as silly as including Blocked Shots, Hits, Takeaways, etc.

 

[DL44]

Face off Stat is completely irrelevant as the amount of effect it has on a teams success can already be found in the teams points and goal differential. Including it would be as silly as including Blocked Shots, Hits, Takeaways, etc.

EVERYTHING at the end of day is eventually reflected in points/goal diff.
Where's the fun it that?
You must that guy in any power rankings thread that provides a link the league standings...


to your post -Would that really be silly tho?
Why not put all those Real Time stats as group and weighing them appropriately, wouldnt it be further reflective of strengths and weaknesses of teams and therefore a component of a teams overall "power"?

Albeit at fraction the worth of ROW% or points etc...

I'll do what I usually do as crude measure of correlation.... Take the top 5 teams in the league and bottom 5 teams in the league and then try to determine if overall there is any significant correlation with rankings in various categories.

i.e. Blocked shots:3 of the bottom 4 you have the President;s Trophy winner, and the 2 Stanley Cup finalists... So we can regard that stat as meaningless... right? or is it statdium effect... obviously ill have to check over a few years.

 

[Feed Me A Stray Cat]

You're talking about sample sizes that are wayyy too small to be relevant.

 

[seventieslord]

EVERYTHING at the end of day is eventually reflected in points/goal diff.
Where's the fun it that?
You must that guy in any power rankings thread that provides a link the league standings...

haha... exactly!

in all honesty, power rankings are the league standings, with a little bit of extra weight given to the last 10 games. They are fun and little else.

 

[DL44]

First go at it...

I think i may have to separate East and West considering the differences in ROW's already present..

It really IS functioning as 2 different leagues at the moment.

Simplified it from the original. Will add 'Last 10' column after they hit that mark...

ROW%, Goal Diff, Pt, Pt%, 5on5, ST, Sdiff, FO%.

WEST
1. Sharks
2. Blues
3. Hawks
4. Ducks
5. Wings
6. Wild
7. Oilers
8. Canucks
--
9. Avs
10. Yotes
11. Preds
12. Stars
13. Kings
14. Jackets
15. Flames

EAST:
1. Lightning
2. HAbs
3. Bruins
4. Pens
5. Sens
6. Devils
7. Isles
8. Leafs
--
9. Rangers
10. Canes
11. Jets
12. Sabres
13. Panthers
14. Flyers
15. Caps

Here's the 2 conferences together which is unimportant this yr... but the formula takes into consideration the whole league.

1 *SAN JOSE - 3.15
2 *TAMPA BAY - 4.05
3 *ST. LOUIS - 5.40
4 *MONTRÉAL - 5.45
5 *CHICAGO - 5.50
6 *BOSTON - 7.15
7 *PITTSBURGH - 7.30
8 *OTTAWA - 8.15
9 *ANAHEIM - 8.50
10 *NEW JERSEY - 12.00
11 *NY ISLANDERS - 13.10
12 *DETROIT - 15.65
13 *MINNESOTA - 16.00
14 *EDMONTON - 16.55
15 *VANCOUVER - 16.60
16 *TORONTO - 16.85

17 *COLORADO - 16.85
18 *PHOENIX - 17.00
19 *NY RANGERS - 17.95
20 *CAROLINA - 21.30
21 *NASHVILLE - 21.40
22 *WINNIPEG - 23.75
23 *BUFFALO - 24.60
24 *FLORIDA - 24.65
25 *DALLAS - 24.80
26 *PHILADELPHIA - 25.20
27 *LOS ANGELES - 25.95
28 *COLUMBUS - 26.60
29 *CALGARY - 28.20
30 *WASHINGTON - 29.95



Obviously the formula is extremely sensitive to a couple wins here and there with it only being 2 weeks in... Will try it again after another few weeks when there is a LAST 10 column to speak of.

 

[LTIR Trickery]

I'd love to see you work on this as often as possible to see if we can find trends with playoff teams between the #12-24 range of games.

 

[Wizeman*]

First go at it...

I think i may have to separate East and West considering the differences in ROW's already present..

It really IS functioning as 2 different leagues at the moment.

Simplified it from the original. Will add 'Last 10' column after they hit that mark...

ROW%, Goal Diff, Pt, Pt%, 5on5, ST, Sdiff, FO%.

WEST
1. Sharks
2. Blues
3. Hawks
4. Ducks
5. Wings
6. Wild
7. Oilers
8. Canucks
--
9. Avs
10. Yotes
11. Preds
12. Stars
13. Kings
14. Jackets
15. Flames

EAST:
1. Lightning
2. HAbs
3. Bruins
4. Pens
5. Sens
6. Devils
7. Isles
8. Leafs
--
9. Rangers
10. Canes
11. Jets
12. Sabres
13. Panthers
14. Flyers
15. Caps

Here's the 2 conferences together which is unimportant this yr... but the formula takes into consideration the whole league.

1 *SAN JOSE - 3.15
2 *TAMPA BAY - 4.05
3 *ST. LOUIS - 5.40
4 *MONTRÉAL - 5.45
5 *CHICAGO - 5.50
6 *BOSTON - 7.15
7 *PITTSBURGH - 7.30
8 *OTTAWA - 8.15
9 *ANAHEIM - 8.50
10 *NEW JERSEY - 12.00
11 *NY ISLANDERS - 13.10
12 *DETROIT - 15.65
13 *MINNESOTA - 16.00
14 *EDMONTON - 16.55
15 *VANCOUVER - 16.60
16 *TORONTO - 16.85

17 *COLORADO - 16.85
18 *PHOENIX - 17.00
19 *NY RANGERS - 17.95
20 *CAROLINA - 21.30
21 *NASHVILLE - 21.40
22 *WINNIPEG - 23.75
23 *BUFFALO - 24.60
24 *FLORIDA - 24.65
25 *DALLAS - 24.80
26 *PHILADELPHIA - 25.20
27 *LOS ANGELES - 25.95
28 *COLUMBUS - 26.60
29 *CALGARY - 28.20
30 *WASHINGTON - 29.95



Obviously the formula is extremely sensitive to a couple wins here and there with it only being 2 weeks in... Will try it again after another few weeks when there is a LAST 10 column to speak of.

Its very difficult to predict how good teams are using some mathematically formula, as you are finding out.

 

[DL44]

Always modifying..added disciple, just cuz i could (PPO-TSH).
At about the Quarter poll now...
Teams have enough gms so a couple get shaved off for the Last 10 column to have a slight effect... going forward, it will have the largest effect...

(Into account - Last 10, pts, pt%, ROW%, GDiff, 5on5, ST, SoGDiff, FO%, disciple)

West

1. Hawks, 3.80
2. Canucks, 5.45
3. Ducks, 5.45
4. Wings, 9.15
5. Sharks, 10.25
6. Stars, 10.95
7. Coyotes, 11.70
8. Preds, 12.25
9. Blues, 13.15
10. Kings, 17.60
11. Oilers, 17.60
12. Wild, 17.80
13. Avs, 18.85
14. Flames, 20.85
15. Jackets, 22.55

East

1. Bruins, 4.95
2. Devils, 6.40
3. Canes, 6.95
4. Pens, 9.20
5. Leafs, 9.25
6. Habs, 9.55
7. Sens, 10.15
8. Rangers, 10.85
9. Lightning, 12.20
10. Flyers, 15.90
11. Jets, 17.85
12. Islanders, 19.95
13. Sabres, 20.90
14. Capitals, 21.15
15. Panthers, 21.15