NAF Coach Rating System

High & Mighty · Post by **High & Mighty** » Thu Oct 17, 2002 7:55 pm

A question on something that has come up in the High Fan Factor thread. Could anyone in the know explain how the NAF Coach Rating system is calculated?

neoliminal · Post by **neoliminal** » Thu Oct 17, 2002 7:58 pm

I used to live in DC. Where in DC do you live?

-=-=-=-=-=-=-==-=-=-=-

NAF rating system.

Reputation is calculated by a complex formula and isn't for the weak of heart. It's based on the Chess Federation's ranking system with a bit of tweaking stolen from the WOTC system and some additions of our own. Here is the formula as it stands:

Win Probability = 1/(10^(((Opponent’s Reputation – Your Reputation)/150) + (( Opponent's TR - Your TR)/70)))+ 1)

php code looks like this:

Code: Select all

 
   $winprop = 1/(pow(10,((($repagainst - $repfor) / 150) + (($tragainst - $trfor)/70)))+1);

This probability is then used to recalculate each player’s rating after the match. In the equation below, players receive 1 point if they win the match, 0 if they lose, and 0.5 for a draw.

Players’new ratings are determined as follows:

Player’s New Rating = Player ’s Old Rating + (K-Value * (Scoring Points – Player’s Win Probability))

K-Value = 10

Code: Select all

   if ($goalsfor > $goalsagainst) {$winlose=1;} 
    if ($goalsfor < $goalsagainst) {$winlose="0";} 
    if ($goalsfor==$goalsagainst) {$winlose=.5;} 
$newrep = $repfor + (10*($winlose-$winprop));

The result is a change in Reputation based on how tough the opponent was and how likely they were to win. For example, if I'm playing a really tough opponent who has a high reputation and a high TR, then he will be risking more points for a lose than I will, but will gain less from a win, because he is expected to be the winner.

Darkson · Post by **Darkson** » Thu Oct 17, 2002 8:01 pm

Can you show some examples?

(Brain isn't working at the moment - can only just type clearly

)

neoliminal · Post by **neoliminal** » Thu Oct 17, 2002 8:42 pm

Certainly:

http://www.oldboysleague.com/CR.html

Play with this.

High & Mighty · Post by **High & Mighty** » Thu Oct 17, 2002 9:09 pm

THANKS!

neoliminal wrote:I used to live in DC. Where in DC do you live?

I actually live just over the bridge now in Rosslyn, VA, but went to school and now work in Foggy Bottom, DC for 8 years.

Win Probability = 1/(10^(((Opponent’s Reputation – Your Reputation)/150) + (( Opponent's TR - Your TR)/70)))+ 1)

I think you're missing a '(' after the '/' but the idea is if Rep and TR are equal the win percentage is 50% and the better you are on paper than the other person, the higher your chances of winning (with TR difference being weighted more than Rep difference. Correct?)

Players’new ratings are determined as follows:

Player’s New Rating = Player ’s Old Rating + (K-Value * (Scoring Points – Player’s Win Probability))

K-Value = 10

So play to expectations and your Player Rating stays exactly the same. What's the Starting Rating of a player just beginning?

Thanks again!

Darkson · Post by **Darkson** » Thu Oct 17, 2002 9:44 pm

OK, to make sure I'm clear on this.

The ante at the bottom is the points my CR changes if I W/L/D?

The K-value is 10 (as you posted), but what are the TR weight and CR weight and what difference do they make, and as High & Mighty said,whats a starting CR?

christer · Post by **christer** » Thu Oct 17, 2002 10:11 pm

TR weight = 70

CR weight = 150

As far as I know, K = 2*Sqrt(number_of_naf_coaches)

Starting rating is set to 150.0

-- Christer

Darkson · Post by **Darkson** » Thu Oct 17, 2002 10:14 pm

But what are TR and CR weight (not values of)?

christer · Post by **christer** » Thu Oct 17, 2002 10:39 pm

But what are TR and CR weight (not values of)?

winProbability = 1 / (1 + 10^((CRagainst-CRfor)/150) + ((TRagainst-TRfor)/70))

See the 150 and 70 in that formula? That's what the weights are.

-- Christer

Khail · Post by **Khail** » Thu Oct 17, 2002 10:41 pm

Slightly off topic: Hey H&M, nice to see other folks from around here

I moved into the area recently and hooked up with a league in Fairfax. I guess there's a well established league in the Arlington area as well...may start playin with those guys too. Anyway, just thought I'd say hi...heh.

Darkson · Post by **Darkson** » Fri Oct 18, 2002 5:40 am

Ooops!

High & Mighty · Post by **High & Mighty** » Fri Oct 18, 2002 1:49 pm

Khail wrote:Slightly off topic: Hey H&M, nice to see other folks from around here I moved into the area recently and hooked up with a league in Fairfax. I guess there's a well established league in the Arlington area as well...may start playin with those guys too. Anyway, just thought I'd say hi...heh.

Hi. I know I've started to hear about more and more people in the area playing blood bowl. I had a league with some friends, but they moved away to better themselves and and it looks like our league is now kaput.

Fortunately God (his name being Ronald and Galak) created PBEM and I always have plenty of games to play.

Joshua Dyal · Post by **Joshua Dyal** » Fri Oct 18, 2002 4:06 pm

Hey, wouldn't it be more fun to base the ratings on something used in the real sports world rather than chess? Or does the NFL use stuff very similar to this? I'm not really much of a sports can, so I can't tell you how teams are ranked and rated in the the sports world.

Then again, it might just be the aggregate of a bunch of relatively subjective opinions of scouts and analysts -- I really can't say. It'd be fun to have the ratings match the real sports world, though, instead of the chess world.

High & Mighty · Post by **High & Mighty** » Fri Oct 18, 2002 4:13 pm

Joshua Dyal wrote:Hey, wouldn't it be more fun to base the ratings on something used in the real sports world rather than chess? Or does the NFL use stuff very similar to this? I'm not really much of a sports can, so I can't tell you how teams are ranked and rated in the the sports world.

Then again, it might just be the aggregate of a bunch of relatively subjective opinions of scouts and analysts -- I really can't say. It'd be fun to have the ratings match the real sports world, though, instead of the chess world.

The problem is in the NFL, you play a bunch of teams and everyone has a record and then the best people go to the playoffs and one team beats all the other teams. So unless you can get every bloodbowl player together for a tournament, you need some way to compare two coaches who have never played each other.

And plus in the NFL, strength of schedule isn't a factor at all. Everyone plays different teams, and if you're lucky enough to be in a division with the Bengals, you know you're going to have two easy wins each year and get to 10-6. Whereas if you're in a division of all good teams, you may get the same record as the team playing the Bengals, but actually be a better team because you had harder competition. But in the NFL, you're both 10-6 so you're both equal.

Joshua Dyal · Post by **Joshua Dyal** » Fri Oct 18, 2002 4:18 pm

In the NFL yes, but in the NCAA no -- different divisions don't really play each other much except as preseason "exhibition" games. Yet they still get ranked against each other. There's gotta be a way to do it, I think.

But, like I said, I'm not really much of a sportsfan, so I don't know enough of the details to know if it'd work.