al Furqaan Cricket Power Rankings - Page 2

**Zunaid** · #26 August 27, 2009, 09:07 PM

Al - go back 3-5 years. Have all team be given the same normalized rank 100 - then apply your algorithms to compute the rise and fall of rank. I would posit that that will give you a more accurate rendition of today's ranking. This would be similar to how FIDE ELO ratings were used.

al Furqaan · #27 August 28, 2009, 12:37 AM

Quote:

Originally Posted by Zunaid
Al - go back 3-5 years. Have all team be given the same normalized rank 100 - then apply your algorithms to compute the rise and fall of rank. I would posit that that will give you a more accurate rendition of today's ranking. This would be similar to how FIDE ELO ratings were used.

that would take quite a while...and frankly is beyond my limited scope. if anyone has a quickfire way to do it, i'm up for learning how.

at any rate, i've arbitrarily pre assigned ranks, and after a couple years or so, the formulas should set things straight - if they aren't already. although i doubt very much i'll be able to update the charts. perhaps every few months.

at any rate, a couple of people have already disagreed with my rankings, and offered counter-reasons why. but of course i stand by my rankings for the reasons already outlined above.

Isnaad · #28 August 28, 2009, 04:09 AM

How about a wier update very soon???...

auntu · #29 August 28, 2009, 05:35 AM

Quote:

Originally Posted by Tigers_eye
So whoever beats Ind in a test series will be crowned the new Test World Champ. Can we do it in February? What an opportunity for us.

Good point Mijan bhai. Hope Mash back and Ash get back his shots he played in last Test series against them.

Beamer · #30 August 28, 2009, 11:52 AM

Al always throws a curve ball when you are least expecting! Kudos ma man..

*AsifTheManRahman* · #31 August 28, 2009, 11:59 AM

Is batsman A from Ireland better than batsman B from Bangladesh in ODIs because the former averages more than the latter having played most of his games against non-Test playing ODI nations?

I know this is about team rankings, but some iftar for thought...

*AsifTheManRahman* · #32 August 28, 2009, 12:05 PM

In other words, is it fair to judge an ODI batsman by his average given that the ICC has made a total mess of things by not having ODI tours between Test playing and non-Test playing ODI nations in the FTP?

al Furqaan · #33 August 28, 2009, 05:12 PM

Quote:

Originally Posted by Isnaad
How about a wier update very soon???...

just haven't had time to keep up with it...i plan on doing it one day though, iA.

al Furqaan · #34 August 28, 2009, 05:15 PM

Quote:

Originally Posted by Zunaid
Al - go back 3-5 years. Have all team be given the same normalized rank 100 - then apply your algorithms to compute the rise and fall of rank. I would posit that that will give you a more accurate rendition of today's ranking. This would be similar to how FIDE ELO ratings were used.

i mulled over your post, Doc, and i ended up agreeing with you. however 3-5 years is rather a long time and just as arbitrary as any other period of time. so i've decided to start from the end of the last world cup. lets see if i can get this done before school resumes...

however, afghanistan are a new inclusion and bermuda are an exclusion. i'm thining of ways to include afghanistan without upsetting the balance. i'll prolly have to way completed matches into the equation somehow.

Rifat · #35 August 28, 2009, 05:18 PM

3-5 years?

how about post World cup 2007...?

**Zunaid** · #36 August 28, 2009, 05:23 PM

Quote:

Originally Posted by al Furqaan
i mulled over your post, Doc, and i ended up agreeing with you. however 3-5 years is rather a long time and just as arbitrary as any other period of time. so i've decided to start from the end of the last world cup. lets see if i can get this done before school resumes...

however, afghanistan are a new inclusion and bermuda are an exclusion. i'm thining of ways to include afghanistan without upsetting the balance. i'll prolly have to way completed matches into the equation somehow.

In the FIDE rating system - unrated players are given a rating of 2200; and their ratings may go up or down based on games that these players play in "rated" tournaments. A rated tournament will have a sufficient number of rated players to be statistically significant in using the results to compute the rating changes.

As an example, the top player today probably has a rating of ~2800. There will be only a few super GMs with ratings above 2600.

Bangladesh's GMs are rated in the high 2400s and low 2500s. Not a very good rating.

- Z

al Furqaan · #37 August 28, 2009, 06:49 PM

Quote:

Originally Posted by Zunaid
In the FIDE rating system - unrated players are given a rating of 2200; and their ratings may go up or down based on games that these players play in "rated" tournaments. A rated tournament will have a sufficient number of rated players to be statistically significant in using the results to compute the rating changes.

As an example, the top player today probably has a rating of ~2800. There will be only a few super GMs with ratings above 2600.

Bangladesh's GMs are rated in the high 2400s and low 2500s. Not a very good rating.

- Z

all teams go in with the same 100 points as of april 29, 2007 (right after last world cup)...afghanistan are the only team messing up the list, but they will also debut with 100 points, and will start accumulating points from their first match of world cup qualifiers - even though these matches aren't "official" ODIs.

because of this, i have modified the ICC's stance a bit. i am counting all the world cup qualifying matches as official ODIs, even though they are not. this allows me to maintain integrity in my rankings formulas, and avoids that glitch. in order to do so, i have also to assign point values to "unranked" teams, such as uganda. in order to keep things somewhat fair, all teams ranked below #16 will be given an equal point value, to be determined by me when i get to that point. actually, 25% of the points of the #16 team sounds fair, yet not too insignificant...will be subject to modification as the time arises.

**Zunaid** · #38 August 28, 2009, 08:04 PM

Here is another thing I suggest. At the point where you start computing the rating changes, apply a multiplying coefficient to the rating changes that is > 1. You reduce the co-efficient to 1 once you reach today's date. This will allow quicker separation of the teams but as you collect more and more data you do not need to have that multiplier.

al Furqaan · #39 August 29, 2009, 01:09 AM

one thing i haven't factored in, and i'm interested to see is if my system (correctly) penalizes teams for winning dead rubbers - in other words, is a 2-3 outcome valued less if you win just the last 2 matches after already losing the series vis a vis a 2-3 series which was decided on the final match.

and i just crunched those numbers...

the results don't really make sense, but its a small difference: if a team wins a series 2-1, losing the dead rubber they get penalized marginally. if they win the first game, lose the 2nd, and win the decider, they get rewarded most highly. if they lose the first game, and win the last 2 to steal the series, they get an intermediate reward. again the difference is about a single percentage of total points in the simulation i ran - could be bigger or less depending on actual ranks of actual teams involved.

the logic of the formula is that you get more points for winning a dead rubber, since you've beaten a "stronger" team. ok, it kinda makes sense now. now point harping on this potential shortcoming...its a paradoxical situation i guess.

One World · #40 August 29, 2009, 11:38 AM

Have to say its compelling. When it matures up, if you can add up a performance monitor which will judge the pitch condition, umpire quality and weather forecast along with your power parameters to assimilate probabilistically the match outcome over by over you may actually put it in the web with google ads.

It may become a very popular tool for T20/IPL supporters.