How Def Stats Are Determined

[Glenn-Philadelphia]

Yeah, defense is always the tougher of the two nuts to crack when developing algorithms for assigning points/skills.

Here is the algorithm that is executed to determine the composite defensive score:

DefComp: IIf([statmassage].[pos]='D',Int(([blkshot])+Int([takeaway])+Int([giveaway])+(([tottoi]/[2010skaters].[gp])*0.85)+([shpct]*80)+((([plusminus]-(1*[esp]))-[teampm])/5)),Int((([blkshot])+Int([takeaway])+Int([giveaway])+(([tottoi]/[2010skaters].[gp])*0.85)+([shpct]*80)+((([plusminus]-(1*[esp]))-[teampm])/5))*0.9))

The main ingredients (with varying amounts used if F or D) are:

seconds played between blocked shots
seconds played between takeaways
seconds played between giveaways
percentage of time played shorthanded
and lastly
player plus minus compared to teams plus minus

If there are some other thoughts on ways to determine a defensive score, I am all ears and hope folks would have some suggestions which they can post here.

Read more: gtgfhl.proboards.com/index.cgi?action=display&board=leaguerequests&thread=702&page=1#ixzz1nnEmcpVl

[Ian-Halifax]

I like the thought of SHTOI going into D rating, but I agree this could negatively affect people who don't see much SHTOI. What if there was just a bonus, like each 20 seconds SHTOI/G = 1 D rating. This probably would only work for forwards, but I don't see anyone getting a huge bonus from it. Six or seven more D for the guys who play on PK the most?

Edit: I'm not set on the 20 seconds = 1 D or anything, just a number I threw out there to explain. More of the fact that playing SH could just be a little bit of a bonus D rating in the formula you are already using.

[Glenn-Philadelphia]

Feb 29, 2012 12:45:21 GMT -5 Ian-Halifax said:

I like the thought of SHTOI going into D rating, but I agree this could negatively affect people who don't see much SHTOI. What if there was just a bonus, like each 20 seconds SHTOI/G = 1 D rating. This probably would only work for forwards, but I don't see anyone getting a huge bonus from it. Six or seven more D for the guys who play on PK the most?

Edit: I'm not set on the 20 seconds = 1 D or anything, just a number I threw out there to explain. More of the fact that playing SH could just be a little bit of a bonus D rating in the formula you are already using.

Understood Ian. That is sort of the way I was looking at it. You may or may not be able to notice from the code snippet above is that D men and Forwards have their own logic associated with all of the values.

What I will do is tinker with some proposed changes after the season to see if I can come up with something that is a little truer. Even guys like Crosby, who in my mind is not a defensive liability, defensive rating should be higher.

Other guys, like Mssr. Daytsuk, who is, as Chris has pointed out 1,000 times, a perennial Selke award nominee, defensive rating may be lower than it should just because he only played a small amount of time of the PP relation to his total time on the ice.

Small tweaks to make it better is what I am looking for, especially on the defensive side of the cards.

[Ian-Halifax]

Luckily for Chris, Daytsuk should have a legit D rating next year whether changes are made or not, he's fighting for the most takeaways in the league.
One question I actually have, does the stat generation figure in stats vs those of teammates too or just the entire league? This is something I have always been curious about.

Edit - Just kidding, just saw it in the formula.

[Glenn-Philadelphia]

Feb 29, 2012 15:11:39 GMT -5 Ian-Halifax said:

Luckily for Chris, Daytsuk should have a legit D rating next year whether changes are made or not, he's fighting for the most takeaways in the league.
One question I actually have, does the stat generation figure in stats vs those of teammates too or just the entire league? This is something I have always been curious about.

Edit - Just kidding, just saw it in the formula.

Next year in an effort to promote league parity, there will be some additional coding added to the engine.

if team = 'Suffolk' then
stat = stat - 2
end if

[Chris-Suffolk]

LOSERS!!! ;D

By the way Ian, Pavel is always among league ( NHL ) leaders in Takeaways....that is one of my main arguments regarding his defensive rating!!!!! Hence, why he has been a Selke Award winner or nominee yearly for the past 5 or 6 yrs......

He just doesnt get the respect in the FHL!

[Scott-New York]

Dustin Byfuglien

2010 NHL stats - 81 GP
20 G 33 A
:05 sec. avg pk toi
-2 (ATL was -33)
72 BS

2011 FHL stats - card 36-32
62 GP
16 G 15 A
+15
75 BS

Tomas Kaberle - 82 GP
4G 43A
:17 SHTOI/G
+4 (BOS +58)
140 BS

25-28
57 GP
0G 10A
-9
89 BS

Cody Franson - 80 GP
8G 21A
:06 SHTOI/G
+10 (NAS +23)
52 BS

32-23
55 GP
5G 10A
+16
60 BS

Anton Babchuk - 82 GP
11G 24A
:30 SHTOI/G
+14 (CAL +2)
156 BS

34-31
52 GP
6G 12A
-8
82 BS

Keith Yandle - 82 GP
11G 48A
:43 SHTOI/G
+12 (PHO +24)
103 BS

31-32
63 GP
7G 19A
+4
69 BS

NOW, I will compare these with one of the best , if not the best overall defensive cards in the FHL this year (this palyer has 2 cards BTW, just noticed that)

Andrew MacDonald - 60 GP
4G 23A
2:23 SHTOI/G
+9 (NYI -33)
160 BS

27-49
56 GP
2G 13A
+9
94 BS


Read more: www.gtgfhl.proboards.com/index.cgi?board=leaguerequests&action=display&thread=702&page=2#ixzz1notfjuGQ

[Phil-Cornwall]

bumping this for Scott's compiled stats above ^^