Ctp Pbem Rankings 12th April 2002

[Gavrushka]

Well I never..... There is one floor in this entire debate.... The rankings are no longer being updated..... So the argument is kind of academic... It is over three months since the results were last calculated...

Keygen the Formula for the new system is at the start of every new rating thread...Calling it new is a misrepresentaion however, it started with Gameleague, apart from the stake was reduced from 20% of your points to 10%, to take into account that in PBEM there are more results per game.

I actually do not think it matters all that much which system is used... I only started using the Gameleague system because I was too thick to understand the other system . BUT I do prefer it because it is easier, and seems to represent ability better...

I think the biggest issue is the players that only want to be rated in games they will score positive at... this kind of defeats the object of having a ratings system... I appreciate the Blackice comments about bad land... But any game creator with 30 minutes of effort can equalise the land, and make the game more enjoyable, fairer, and longer lasting... when games can run for years the 30 minutes should be a worthy investment... SO maybe rules for game creation should be applied too.

But to reiterate, this is just all gibberish unless someone takes the mantle of updating the rankings onboard , and I would strongly suggest you forget about holding a vote as to what system to use, and instead let someone take over the ratings, and use whatever system they see fit.

Oh and by the way, hello everyone!

[quinns]

Hello Gary! You're right about current updates... But why update unless it is backed by a concensus of voters? You, Blackice, and I have all updated the ratings at one time or another. Without a good solid vote to stand on, we were just sort of spinning our wheels. While many appreciated our work, many were not happy with the system for various reasons. At least with a good 51% or more vote, whoever takes over doing the ratings will know that the majority is in support of their work.

Personally, I will be happy to support whatever the vote turns out to be. Though I think that the "Chess-like Points Differential" system is better than the "GameLeague Wagering Points" system, I will fully support whichever one is voted in.

And Keygen... thanks for creating the link to the documentation for the old "points differential" system we used to use!

[blackice]

Keygen the formula is in the first page of this thread.
The rest has been stated if you now have the formulas add it up.
Add new rules or keep the system that needs none.
I also think you see people just want to play add a rule of manditory ranking and lose people.
It is not gavrushka/blackice system gavrushka did the PBEM and I did the online take note they are TWO different systems. It is the new system keep it genetic not personal as I have been requested to do Thanks.

Rankings were being updated until I decided to end it rather than being bumped after volunteering.

This is not chess, games here are rated at the finish more important they are rated per 9 turns chess is not.

That obviously changes the facts and how the system works. The based on GL Not GL system is tried and true for CTP.

It may work in chess too but who cares this is CTP and the new system not only works. It is more fair more to the reality of PBEM CTP. It also does not need more negative rules to make it adapt to a game it was designed for.

So show some people some examples Quinn show them side by side the difference. Show us all why your system for chess is better. Use some examples of current random players. It is your system explain to us all by side by side examples of why your system is better.

That is only fair for an informed vote. Besides you solver and me are probably the only ones who could figure out your system. The new is quite easy and self explanitory ( user friendly ). On that note I think people should fully understand what your system is explain it to us in full detail with examples please.

[Gavrushka]

I got a neuatral suggestion... You are discussing two rating systems which all people see as having Merits and DeMerits... Maybe it is time to devise a NEW system for Apolyton, if everyone (that is interested) posts their ideas of what they would like from a system, then PERHAPS a new, universally accepted Formula can be created.. We do have some Maths capable people here..

I guess in outline a rating system would:



Reward Players more for beating Players with a Higher Score.

Give Higher Rewards the More Players in the Game.

Give diminishing Returns for consecutive Victories.

Reward the Player more for Volume of Games they are in.

Be sufficiently Volatile that a WIN is likely to move them up the Rankings, without a massive change in their score.

Not cripple a player for having one or two bad games.

Have a very marginal impact for gaming against opponents with far lesser scores (but never negative).

Be Transparant and easy to calculate by all.



Can't be that difficult can it??

[blackice]

You just described the new system

[Solver]

The formula can clearly be built, except for this:

Be Transparant and easy to calculate by all.

But since I already got a program, I'll just have to change the calculation bits should a new formula be presented. That is, of course, what we want. I'm just eager to do an update now, and am only waiting for this formula the, eh?

[Kralj Matjaz]

Reward the Player more for Volume of Games they are in.
WHY? WHY? WHY?

No inflation points, please!

[quinns]

Originally posted by blackice

So show some people some examples Quinn show them side by side the difference. Show us all why your system for chess is better. Use some examples of current random players. It is your system explain to us all by side by side examples of why your system is better.

That is only fair for an informed vote. Besides you solver and me are probably the only ones who could figure out your system. The new is quite easy and self explanitory ( user friendly ). On that note I think people should fully understand what your system is explain it to us in full detail with examples please.

Okay.

Here is the example:

Albert rated 20.
Bob rated 15.
Charlie rated 15.
Dallas rated 15.

GameLeague Based Ratings Calculation:

Power Graph order at turn 9 = Albert, Bob, Charlie, Dallas.

Wagers:
Albert = 10% of 20 = 2 points
Bob = 10% of 15 = 1.5 points
Charlie = 10% of 15 = 1.5 points
Dallas = 10% of 15 = 1.5 points

Total Wagered = 6.5 points.
Percent Distribution = 46%, 29%, 17%, 8%
Wagered Points Distribution = 3.0, 1.9, 1.1, 0.5

After Turn 9 ratings:
Albert = 18 + 3.0 = 21.0 (net change + 1.0 rating points)
Bob = 13.5 + 1.9 = 15.4
Charlie = 13.5 + 1.1 = 14.6
Dallas = 13.5 + 0.5 = 14.0

Albert is playing in four other games exactly similar to the one above, where he plays against 3 much weaker players than himself. So that now, after the first 9 turns of all of his games,

Albert has a new rating = 25 (GameLeague Based)

(20 + 1.0 + 1.0 + 1.0 + 1.0 + 1.0 = 25)

--------------------------------------------------------
Differential Based Ratings Calculation:

Power Graph order at turn 9 = Albert, Bob, Charlie, Dallas.

Albert (20) defeats Bob (15) -- Ratings Points Differential = 20 - 15 = 5

Probability of Lower Player Winning = 1/[2 + Differential**2] (Note: **2 means squared)

Probability = 1/[2 + (5)**2] = 1/27 = 0.04

Because Albert is higher rated than Bob,

Change = 0.04 / 2 = 0.02 points.

[Note: If Bob had defeated Albert then the change would have been (1- 0.04) / 2 = 0.48]

So that Albert gains 0.02 points and Bob loses 0.02 points.

The calculations are the same for Albert defeating Charlie and Dallas. So Albert gains 0.02 points for defeating Charlie and 0.02 points for defeating Dallas.

The total calculations done are: Albert defeats (d.) Bob, Abert d. Charlie, Albert d. Dallas, Bob d. Charlie, Bob d. Dallas, Charlie d. Dallas. We are highlighting on Albert so I will just focus on Albert's rating change.

*** Edit *** (I think this is what Blackice meant about the "loss ratio"):

After Turn 9 ratings:

Albert = 20 + .06 = 20.06
Bob = 15 - .02 + .25 + .25 = 15.48
Charlie = 15 - .02 - .25 + .25 = 14.98
Dallas = 15 -.02 - .25 - .25 = 14.48
*** End Edit ***

In total, Albert gains 0.06 ratings points for defeating the much weaker players in the turn 9 power graph for a new rating of 20.06.

Albert is playing in four other games exactly similar to the one above, where he plays against 3 much weaker players than himself. So that now, after the first 9 turns of all of his games,

Albert has a new rating = 20.30 (Differential Based)

(20 + 0.06 + 0.06 + 0.06 + 0.06 + 0.06 = 20.30)

Edit: Without doing the exact calculations for the other players, I can estimate their new ratings assuming that order of the 5 games
-----------------------------

Conclusion:

In the first example with GameLeague Based calculations, Albert has managed to go from a rating of 20 to a rating of 25 by playing in 5 games against players with ratings no higher than 15. This is unfair in my opinion.

In the second example with Differential Based calculations, Albert has gone from a rating of 20 to a rating of 20.30 by playing the mass number of weaker players. This is fair in my opinion.

[Gavrushka]

Aaargh! Doesn't everyone see that everyone has an opinion on which system is best, therefore neither system can work! a new one is required based on the consensus of what everyone (who is interested ) thinks... If you vote for one system or the other, people who lose the vote will see it as a defeat and may refuse to be rated.. So why not get MAd Maths Man J Bytheway to devise a new system, which everyone has input into... There are no absolutes here, just subjectivity, and differences are irreconcilable - so just create a new system.. It really is the only way all people will be at least moderately happy.

Anyone see that awful news bulletin yesterday about the cat that was force fed Petrol by some kids and then thrown into a Barrel?? It ran round and round the barrel like a thing possessed until it just suddenly stopped...

Was it dead I here you all shout?

Nope.... it had just ran out of petrol.....

[blackice]

He won them he should be rewarded quinn. It is fair.
You will note even the losers in the first example gain points for playing a stronger opponent. This in effect off sets the win of Albert doesn't it? Of course it does...

Albert's new scores would also go up each game before the calculations. So would the pot then. While Albert gained more per turn so would his opponents. The losers playing stronger players are in a win win scanerio. They too would go up again off setting Alberts gains. Take note too it is doubtfull three players will ever have the same score.

Albert has a new rating = from 20 to 25
Bob 1.5 x 5= 7.5 from 15 to 22.5
Charlie 1.1 x 5= 5.5 from 15 to 20.5
Dallas .05 x 5= 2.5 from 15 to 17.5

It is win win scenario the rankings are ever moving this has been proven with the new system.

Quinn could you show what the lossers lost in the second example please. Then show Albert losing to weaker players.

After all that has been the complaint not the points gained in either system for winning. Now does everyone see why Klair may have never moved from first place over all? At least in our life time no one made enough points per turn to do it. And she only played ONE GAME!

Here is Albert losing to weaker players in the new system. Notice the points distribution is the same. The higher ranked player always helps the lower ranked in a game by adding more to the pot. This is why this system is fair and promotes higher ranked playing lower ranked.

After Turn 9 ratings Albert losing:

Dallas = 13.5 + 3.0 = 16.5 (net change + 1.5 rating points)
Bob = 13.5 + 1.9 = 15.4 (net change + .4 rating points)
Charlie = 13.5 + 1.1 = 14.6 (net change - .4 rating points)
Albert = 18 + .5 = 18.5 (net change - 1.5 rating points)

Note Albert lost to lower ranked players and pays the price. Because Albert put so much in the pot most all the weaker players gain points. Win win scenerio.

Except for the losers Charlie a mear .4 loss for being beaten by same ranked players. Albert as it should be he lost to weaker players. He would more than likely go down in the rankings for this one loss but gain on his next five wins. Also note this system tells people to play more games to gain more points. This is the way it should be it also does this without the need for a play three games minmum rule. Less rules is good right? I say so. This way you do not force people to play more games if they do not have he time or the inclination to do so.

This system moves players if Albert won his next 4 games the rankings would be very close Albert would still be ahead in the rankings but so would the other players. Win Win scenerio.

Now lets see an example of the old system and the losses each player takes...Sorry for the request but that is the whole point here...

As said the ratio as with the blackice example is 2 to 1 or better. You win two and lose one you drop like a lead balloon...As you can see in the new system that is simply not the case. You lose big time for losing to lower ranked players. You would then gain on the rest of your games for winning them.

It is not so with the old system your loss is so great even if you win the next two games because they are weaker players you would drop 4 ranking spots or more for winning! It did not promote playing weaker players the new system does...
The reality of the situation is you will more than likely have weaker players playing in a game in PBEM.

That is the way it should be you should be awarded for winning. Albert playing weaker players will gain points but the points he gains are limited because he plays weaker players. He still wins though! just not as much...

gav

This is a new system and it works great! the whole Idea here is to compare them side by side and let people vote with an informed vote.

[quinns]

I cannot convince you Ice, and I never will. All the logic in the world will not matter to you. You know and love GameLeague's system, and that is that. The above explanation was not really meant for you, Blackice, but for the others who may want to vote on the issue.

I would agree with Gavrushka, though, about another system if one is presented. Yes, J Bytheway is the most qualified of all of us. But I do not know if he is willing to do it. Are you able to find time to do this Mad Maths Man JBTW?

[blackice]

why won't you post the loss ratio quinn?

The above explanation was not really meant for you, Blackice, but for the others who may want to vote on the issue.

It is only fair people get a complete view of the two systems not half. Post the whole situation...I find it highly suspect when you do not post what people complained the most about the unfair win loss ratio...Complete the example you did it for the new system why did you not for the old one? At best is misleading the people here... nothing personal it was probably an oversight...

It is not gameleague's system it is a variation of it adapted specifically to pbem

I like it because it is better others do too obviously. It rewards winning. It is fair for all players including losers of games.

What if Paulino1 or Yoda joined here. They would be ranked as lower players. We all know they are not that at all. So why should a higher ranked player get wiped out in the rankings by a couple of turn 9 rounds playing better lower ranked players? The old system did just that. The new ones accounts for newer good players and does not heavy handedly wipe out higher ranked ones that lose to better lower ranked players.

The idea here quinn is to have a honest display of both systems to convince others here that your system or the new one is better. Not to convince me that will never happen. The key differences between these systems is the loss factor of the old system. I want the other players to see that, they need to see that in order to get the complete picture.

I really see no need to have a Third system as stated. This system is fair and new.

Lets just post the actual facts and let people decide so how about it Quinn post the loss ratio in these examples so people can see the whole picture and decide.

Lets get on with it shall we...

[Kralj Matjaz]

Blackice, half of post about ladder system is yours and you'll only one promoting Gavrushka/Blackice system. You make your point.
JBTW can you came out with third solution?

[blackice]

Kralj and? What are you telling me to shut up? That is not going to happen sorry...

I am trying to inform the people here so they can make an informed descision. What have you posted for details so people can make an informed decision?

Last time It is not my system

I find it odd you have read everything but missed that again?

JBTW can you came out with third solution?

Why do you need a third system you have not seen the comparision of these two yet?

It is getting clear a few of you do not want people to see the comparison at all let alone vote on it. Humm wonder why?

On top of that we could see them side by side and start a new thread with the complete story of the two rankings and then offer the choices.

New
Old
Third system to be designed yet

Give the people the chance to see the true pictures then the chance to vote. There are a lot more people here than us five...let them have a chance to vote on all three options. Some people seem to want to just over rule the rest with a third option. Get the true picture out and give the others and opportunity to see them and vote...

Too much politics just get on the the full picture that was omitted. Well whats wrong why not post it Quinn?
You did not offer a complete picture of your system you did of the new one.

Be fair post it lets go...

[J Bytheway]

I've been thinking, believe me, but I can concieve of no system which will fulfill many of the requirements stated.

Personally I think the fact that in the wagering system winners can lose points and losers gain them is a much more serious flaw than the problem with the old system (which I still don't understand, frankly...).

I'm going to look up the rules on the differential system and get back to you - I think for a start the power used might be reduced to offset the problem. The calculations as described above make no sense since if the two players had the same score the probability of one winning would be 1/(2+0)^2 = 1/4, it should be 1/2...