You are not logged on | Login | Register  
Rating formula

 
Blankenhor

As I recently stated, I took on the task of making a most needed improvement to the rating system. Hours of mathematical manipulations to eliminate the need of if-then programming code culminated into a single line formula.

Below, the max portion enables an accelerated factor to approach equilibrium of theoretical rating in the initial games.

The 1600 constant effects a spread of the distribution of ratings. In games of greater skill, the constant can be decreased. Games with extensive luck would necessitate a larger constant to create a distribution that is of a range a few hundred away from the median.

The 8 constant is a factor that provides a non-trivial change in rating per game played.

The exponential function provides a probability factor that follows the Gaussian distribution.

In rationalizing the skill involved in placing, and to deal with the well-known 2nd placing agenda, the untempered rating change for a second placer is proportional to the first place finisher by a factor of (-1/PLAYERS) where PLAYERS is the number of players in the game. Subsequent untempered rating changes for 3rd and greater places are incrementally greater in magnitude relative to the 2nd placer with an untempered zero-sum for all places.


Declarations:

OLDRATE = rating at beginning of game
NEWRATE = rating after game played
GAMES = game experience
PLAYERS = number of players in game
PLACE = game result place of player
AVGRATE = average of ratings of players at beginning of game

FACT is the factorial function.
TRUNC is the truncate function.


NEWRATE = OLDRATE + MAX(1, 4-(GAMES^2/200)) * 8 * ((2-PLACE-FACT(PLAYERS-1)/(2*FACT(PLAYERS-3)))/
((FACT(PLAYERS-1))/(2*FACT(PLAYERS-3)))-
(TRUNC((PLAYERS-PLACE+1)/PLAYERS))/((FACT(PLAYERS-1))/
(2*FACT(PLAYERS-3))))*((-1-PLAYERS)*TRUNC((PLAYERS-PLACE+1)/
PLAYERS)+1)*((TRUNC((PLAYERS-PLACE+1)/PLAYERS)*(1/(10^((OLDRATE - AVGRATE)/1600)+1))-(TRUNC((PLAYERS-PLACE+1)
/PLAYERS)-1)*(1-1/(10^((OLDRATE - AVGRATE)/1600)+1))))

[Message edited by Aerobabe]

 
ovechkin

Youre contribution has been submitted to the noble prize committee.

 
Lazlo

That's great. Now calculate the odds of this, or any other recommended improvement, ever being incorporated into the game. Feel free to round off to the nearest millionth.

 
Lumumby

I know that formula Lazlo:

% chance = (100 + (Amount of server break downs per month * 0)) - 100

 
Blankenhor

I welcome anyone wishing to test the formula.

Try setting up a hypothetical scenario of 3 players with an initial rating of 1500 with the following constraints:

Player A:
1st places: 50%
2nd places: 30%
3rd places: 20%

Player B:
1st places: 30%
2nd places: 40%
3rd places: 30%

Player C:
1st places: 20%
2nd places: 30%
3rd places: 50%

Run the formula over the course of 30 iterations and see where the ratings migrate.

 
weakling

I will supply an excel spreadsheet to anyone interested so this new rating system could be more easily understood and appreciated.

The present rating formula mimicks a binomial probability distribution and that is not reflective of placing in this particular game.



 
mongolian

great post Blank, but in lamend's terms. How is this improving the rating system dramatically here? Or Diving off of just not rewarind 2nd place that many points. In your formula, I also was trying to ponder a different new system that basically won't change rating for the average 2-4th place. And my final point is how can this or current system reflect more of a chess-like rating scale in terms of (A) what that rating points corrolates to and (B) how this points can be more evenly achieved.

Aside from my points, I can't see how the important rating is to be changed. Perhaps this is where you need to search before this new formula can be applied.

 
weakling

i do incorporate chess-like ratings when adjusting points after the untempered ratings are calculated (it's two parts in one). However, understanding the dominate game and placement does not translate into the binomial formula that is used here where points are awarded somewhat equally in MAGNITUDE to first and last players and where last and second-to-last can exhibit a large rating difference when in fact if you face an early suicider it speaks little of skill and thus there should be little difference between places from last all the way to second place, with first place deserving the only positive points. second place would get the smallest negative amount of points (relatively small, and this really could help out the situation of second-place finishers, for if the difference between second and third are not that great and realizing a big potential exists in getting first, the dynamics would change in some circumstances).

I am surprised this has not been given more consideration.



 
mongolian

a-men. Truly great post and calculator. I agree to 2nd-4th getting relative no gain/loss.

My best guess is that your points so seemingly obvious how much it can help the community aren't being spoken enough as to how the impact effects? I don't mean for you to respond, but respond to a direct challenge why Emiel should go out of his way to use this new improoved system.

 
Hannibal23

Your formula will have the unintended consequence of creating new classes of players: third-placers and fourth-placers. I would much rather just deal with a predictable second-placer.

 
weakling

oh really? how so? a new class of 4th and 5th placers? that's absurd, but i'd like to see how you came up with that conclusion as there'd be very little difference in ratingadjustments among 2nd thru last place.

perhaps you need to reread and grasp the mathematics involved.

 
weakling

I don't know if anyone can truly appreciate what I've done in making a generalized SINGLE LINE (long line) formula to figure out the ratings for players regardless of whether it's for 3,4,5, or 6 player games.

You really have to plug in some numbers and see for yourself what happens.

Try avg rating of players (OLDRATE) = 1500, #players = 6, and plug in 1,2,3,4,5, and 6 for place and see what you get. Just allow # of games played prior = 15 to avoid the accelerated rating change, or, if you wish, let GAMES=0 and see what you get.

 
Lazlo

I'm sure it's a lovely and elegant formula but, since you made no allowance for the type of game being played, it's as worthless as the current rating system.

Playing a game of blind caps should not effect ratings the same way as playing a classic game. Same for different maps, increasing, etc...

It's like comparing chess and checkers as equal weight games.



 
mongolian

I am in favor of any new system supporting that gives the same weight of 2nd place as third. Don't care if both 2nd-3rd mean both places don't lose points, but more incentive to get 1st place should be in store.

The single most helpful rating system would be putting a cap on total # of accounts to be created by players to 3. This really helps the system as people value ratings vs just deleting accounts or making infinite amount of new accounts as the current system rewards.

 
weakling

The formula works for 3 to 6 player games.

For games of more luck such as caps, the 1600 constant could simply be different.

The most notable part of this is that first place is given points whereas the other places receive SLIGHT decreases beginning with second place and the magnitude goes up little from there so you don't get overly punished if someone suicides early on you (not an ordinary measure of skill/lack of skill) or from disconnection.

I'll show you the rating changes per number of players, assuming all ratings going into the game are the same (I use a probabilistic model to temper the points for unequal ratings), and that each person has played at least 20 games in my next post::::::::::::::::::

 
weakling

This is also before a magnitude factor is applied, to effect spread. I put the following in words above but I'll reduce it to numbers herein. Note the conservation of points (net sum = zero). The RATIO of points given with relative points AMONG # of players in game are:

3-player
1st +3
2nd -1
3rd -2

4-player
1st +4
2nd -1
3rd -1.333
4th -1.667

5-player
1st +5
2nd -1
3rd -1.167
4th -1.333
5th -1.5

6-player
1st +6
2nd -1
3rd -1.1
4th -1.2
5th -1.3
6th -1.4

 
weakling

with so little to gain from 2nd over 3rd place, you'll see so much less 2nd placing as the prize is getting to first. the amount of points a 2nd placer gets over a 3rd placer is negligible, except in a 3-player game.

the numbers above are without a magnitude factor (I use 8 I think in the formula, but it could be made to 4 or whatever the owner decides), and thus the actual points, barring rating differences and accelerated ratings for first few games would be multiplied by such factor to effect a non-trivial rating change.



 
mongolian

Yes, i love that formula weakling. The only problem is you can't do a straight point net/loss. You have to consider each player's rating + your system. The other thing is more +/- will therefore need to happen, as in, each game shouldn't just be worth +6 points because of ratings.

 
weakling

??????????????

straight what? the player ratings will affect the points somewhat, as you would see if you bothered to test the formula.

i mentioned that the above numbers are ratios and made it clear that the magnitudes could be altered. In my forumla, ALL THE POINTS ABOVE WOULD BE MULTIPLIED BY 8. Did you look at the formula, read about it and even read in my recent posting about the magnitude change to affect a non-trivial rating change?

What you are asking for in attempting to shoot holes in my formula probably already exists. please look it over carefully. I'll send you an excel spreadsheet so you don't have to waste your time copying the formula...though it's not that difficult. cut and paste it in a spreadsheet and see for yourself what happens!

 
weakling

along with the factor of 8 is a factor relating to the probability of one winning which is the exponential factor....the factor is roughly 0.5 +/- a little depending on one's rating versus the average of ratings involved.

Thus the gross magnitude would be roughly 4, so that's about 24 points for first place in a 6-player game. I think this is fine as too wild of swings should be curtailed. Second place in a 6-player game would be about -4 with sucessive places a SLIGHTLY larger magnitude, those being approximately -4.4 for 3rd, -4.8 for 4th , -5.2 for 5th and -5.6 for 6th.

The points awarded are conserved so if you do well like take first place more than 1/6 times in a 6-player game, you will move up. If you are only a second placer, you can't move up.



 
mongolian

no..i misread that part. Didn't realise you were applying that to your formula above.

PS - looks like a made a new enemy ;p Just here to try to help the game weakling, sorry if you took my comments personally.

 
weakling

Evidently it appears I have to list the rating changes per formula given above as not everyone comprehended my mathematical sentences.

Here's what the rating points would be if the formula is used, albeit without a modest factor based on rating differences of those playing, but the factor of which is secondary:

6-player 5-player 4-player 3-player
1st +24 1st +20 1st +16 1st +12
2nd -4 2nd -4 2nd -4 2nd -4
3rd -4.4 3rd -4.67 3rd -5.33 3rd -8
4th -4.8 4th -5.33 4th -6.67
5th -5.2 5th -6
6th -5.6

Note, there'd be less wild swings are there exists now. Playing a single game should not alter a rating more than about 2%. Note the insignifcant points between 2nd and 3rd place relative to the difference between 2nd and 1st place which would remove lots of purposeful 2nd placing, not to mention points would be taken away for 2nd placing. Also, getting eliminated quickly for whatever reason won't take 60-100 points away from you as can happen in the present system (outrageous penalty, especially if simply faced with a stalking suicider or suffering from server disconnection).

The ratings could be stored as real numbers and reported as integers.

 
weakling

i apologize for the formatting...i didn't type it that way.

 
Spir-An

6-player 5-player 4-player 3-player
1st +24 1st +20 1st +16 1st +12
2nd -4 2nd -4 2nd -4 2nd -4
3rd -4.4 3rd -4.67 3rd -5.33 3rd -8
4th -4.8 4th -5.33 4th -6.67
5th -5.2 5th -6
6th -5.6


 
Spir-An

Only slightly better. You use [ code ] and [ /code ].

 
weakling

6-player     
5-player      
4-player      
3-player<br><br>
1st&nbsp;+24&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;1st&nbsp;+20&nbsp;
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;1st&nbsp;+16&nbsp;&nbsp;&nbsp;
&nbsp;&nbsp;&nbsp;&nbsp;1st&nbsp;+12<br><br>
2nd&nbsp;-4&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;2nd&nbsp;
-4&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;2nd&nbsp;-4&nbsp;
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;2nd&nbsp;-4<br><br>
3rd&nbsp;-4.4&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;3rd&nbsp;-4.67&nbsp;&nbsp;
&nbsp;&nbsp;3rd&nbsp;-5.33&nbsp;&nbsp;&nbsp;&nbsp;3rd&nbsp;-8<br><br>
4th&nbsp;-4.8&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;4th&nbsp;-5.33&nbsp;
&nbsp;&nbsp;&nbsp;4th&nbsp;-6.67<br><br>
5th&nbsp;-5.2&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;5th-6<br><br>
6th&nbsp;-5.6
&nbsp;

[Message edited by Aerobabe]

 
Spir-An

Fucking hell, talk about breaking the bloody forum.

 
whirldpeas

ABSOLUTELY FABULOUS!



 
mongolian

Final standings should calculate ratings, but here are the key elements the new rating system should take in. None of which current or your system weakling is taking in competely.

- 1st place = awarded most points (depending upon ratings + total # players

- 2nd place = most cases should result in no loss of points nor gain. Very little gain if any, see forumula below.

-3rd - 6th place would all be negative points lost but they ladder a system of loss. See below.
--------------
If all accounts are 1500 in a 6 player game it would look like: 1st +25, 2nd + 5, 3rd 0, 4th, -5, 5th, -10, 6th - 15
--------------

The major flaw of your system weakling is despite I like the fact 2nd & 3rd get same points, 2nd and 6th shouldn't be geting the relative same penalty.



 
Magician

can somebody lock this topic please :-)

just give this site a total reset and we will notice who the best player is very soon

 
copisa

whahahaha lekkus ;)

 
weakling

In order to help prevent the 2nd placing phenomenon, 2nd place and 3rd place NEED to have very little difference in points awarded (or taken away).

The penalty for getting disconnected should not hurt a person unduly, nor should there be a big detriment to a player if he/she is taken out early by chance or from a purposeful suicide.

Being first place is paramount and anything less is still one thing:

LOSING.


Whether you lose fast or lose slow, I have accomodated a reasonable rating formula, with 6th place getting penalized roughly 50% more than 2nd place.

It may simply be out of past experiences, not mathematically founded in the basis of a game, that 2nd place should receive some sort of positive award, but I contend it is unfounded for this particular game.

If one plays well, he/she will still have a rating much higher than peers as placing first more often than 1/(number of players in a game) will effect an increase in rating.

If one plays for second and consistently gets it, well, too bad, as that player would lose rating points by not obtaining the real prize, and that prize is to come in #1.**************



 
Dom_Threat

Well, gee, this is some pretty complicated mathematics......I sometimes have trouble counting up my armies on the screen........



 
mongolian

Reasons why dominate would use a new rating system:

#1 Higher ratio of points won and lost between each place. Effect: Rewards players finishing high, penalizing low placement. Not such the best idea for penalizing low placement becuase most players are disconnecting or didn't even play longer then 5 minutes to recieve these low positions.

#2 Less points awarded to the middle places, starting at 2nd place. Effect: penalizing 2nd-4th with the same amount of points lost/won. In theory, more people should then be trying for 1st. Con: This new rating calculator, also destroys the fact a 2nd place player was just as bad as a 4th place player. After seeing some 4+ hour games, these are the scenarios why the current system is more justified.

------------
Conclusion:

Despite a new calculator should reward 1st place with more points and water-down the points lost from 2-6th, this doesn't seem heavy enough issue for a rating change. Again, I am in favor of a slight rating formula change, but in reguard to things needed to be changed in the game, this is bottom of the totem poll of needed changes.

 
weakling

The rating schedule is at the foremost of things needing change for this site. It is along with the connectivity issue, the largest issues. Teaming is secondary to these.

As a mathematician and have made many rating formulas for different games, I stand behind the SIMPLE math involved in determining points for places, and some of which takes into account the connectivity issue for placing poorly in a game, as it makes no sense to have 100 points taken away from a connectivity issue (5% of one's rating).

I have thus far seen no valid reason to indicate anything less than robustness of the formula I presented. I am not in favor of the present system whereby your rating could change by 5% in a single game. That is simply not reasonable for any rating system.

If I presented the formula in a several step manner, and a different formula for each of the 4 different nubmer of playered games, it'd look very simple, but the real work was in combining them into a single line formula, and that's why the truncation and factorial functions are involved, otherwise you wouldn't see those functions.

I'll be happy to forward anyone who requests, an excel spreadsheet of the formula.



 
mongolian

Before new rating:
------------------
- Research & Investing. The game is no longer being devoloped with better features.

- New Redesign. Dominate game still could be branded into a more "dominate game" & website.

- New programming. Could be new game options, cleaner GUI.

- Better payment plans. Change payment plan to 19 euro or life time membership. Offer more benefits, change settings, etc.

- Advertising. There are ways to do next to budget advertising.

-----------------
The above are general topics that need to upgrade, but the goal with upgrades is focusing on things that bring in more people which then equate to more money. Rating formula is not something a new user is going to buy into. Current users aren't going to buy membership directly because of it either. Having a 6 month rating reset (and keeping top 50 records) is something that would validate people to want to play more then just a new formula.



 
Dom_Threat

Um....Yeah - !

 
LompeToni

I have taken a closer look into the formula. With the 1600 constant my rating after infinite games(took percentages against 1500's)would be somewhere in between 2500-3000. I win 40%. I agree with weakling, if you dont want anybody to secondplace, the player who becomes second should not be rewarded, otherwise that nevers stops. Pity you became second, win next game. This formula is nice balanced. I like this chess-like rating adjustments. THIS IS IT!!!!!

 
weakling

One quick note to those who received my spreadsheet: I was testing the formula and may have changed some cells, notably some may have an ABS function contained in the formula and a cell reference that does not make sense. The correct formula to be in the spreadsheet that reflects the formula provided at the beginning of this forum topic can be effected by the following:

To Test the formula, enter these values in cell # (do not enter the quotes):

1) A3, "3" (for 3-person game).
2) B3, "1" (for first place).
3) B4, "2" (for second place).
4) B5, "3" (for third place).
5) C3, "1500" (rating for first place player).
6) C4, "1500" (rating for second place player).
7) C5, "1500" (rating for third place player).
8) C7, "=average(C3..C5)" to get avg ratings.
9) D3, "=C3-$C$7" to get rating - avg rating.
10) copy above into cells D4 and D5*.
11) E3,E4, and E5, "1" (# of games played).
12) F3, "=$C3+MAX(1,4.005-E3^2/200)*2*((2-B3-FACT($A$3-1)/(2*FACT($A$3-3)))/((FACT($A$3-1))/(2*FACT($A$3-3)))-(TRUNC(($A$3-B3+1)/$A$3))/((FACT($A$3-1))/(2*FACT($A$3-3))))*((-1-$A$3)*TRUNC(($A$3-$B3+1)/$A$3)+1)*((TRUNC(($A$3-B3+1)/$A$3)*(1/(10^((D3)/1600)+1))-(TRUNC(($A$3-B3+1)/$A$3)-1)*(1-1/(10^((D3)/1600)+1))))"
13) copy above formula into cells F4 and F5*.

*: After entering, just pull down with mouse at lower right of cell to copy to cells beneath.

You may change C3 to C5 to see new ratings at F3 to F5.

The above can simply be modified for a 4-, 5-, and 6-player game by changing cell A3 and by entering additional rows following the same format.

 
weakling

I originally suggested using a factor of 8 to make a sizeable (non-trivial) rating change but I figure a factor of 2 to 4 would be more appropriate.

Play with the formula and tell me what you think!



 
chickenshi

WTF
:/

EKK!



 
Dom_Threat

Enough. The current rating system is poor, but a fixed-rate point system (also allowing for some priority privilege based on level) would be preferable to another complex scoring method.



 
Dom_Threat

Don't get me wrong - I don't by any means discourage innovation, and perhaps your system is quite sound. The difficulty in introducing it to the game makes it unworkable in practice.

Cheers.

 
pakezel

Just play the game as it is, it's not that hard to win.

 
weakling

Well, the formula is not that complex and could easily be adopted.

It may look complicated only because I made it a one-line formula, but if parsed for 3,4,5, and 6-pllyaer games, and with a different formula for first place, then you;d see how simplistic it is.

The reason behind it is manifold:

1) present system awards in a binomial fashion that is not reflective of the game

2) it discourages second-placing

3) it does not alter one's rating by 5% in any one game that is completely preposterous as can happen in the present system

4) being it's more robust, it would allow a high rated player to play among low-rated players without worry of getting killed in the rating.

Just follow the simple directions above and you can see for yourself.

 
pakezel

I bet you have a hard on while typing this shit.

 
Elitest

This is all fine and great, but here is my issue with this:



Second placing will always be a part of the game. I'll come out and say it, I would rather take second place than third.



Which is great, and we can reduce this all we want. But, as a question to everyone who's ever played a game with people rated over 1800, what would the game come to? Nobody would attack anybody, and we would have yet another string of endless games.

I, for one, think that if the game were to change at all, it should be amended to ensure people continue to attack, and the games don't come to a stand-still.

But having said that, I also think that everyone should suck it up. I build my accounts up, get them to a point where I can't play games anymore because they all come to a stand-still, and let my account reset. Lets all do the same. =o)

 
pakezel

Perhaps you are playing the wrong settings, try open map caps and use 1 or 2 bots to speed things up. Makes for best games ever, no sitting back for a zillion hours like reg is all about.

 
weakling

"I bet you have a hard on while typing this shit."

Nope, but doing the math provides a most wonderful cost-free high.

 
Elitest

Bots? Thanks, but I'd rather play with a little skill required.

 
Morgion

I definitely think we need to adopt this new system. Second-placers are very close to completely ruining this game! I've only been here for 100 games, but, I CONSTANTLY encounter a scenerio in which there is one big player, and the other small player turns on me and kills me. On MANY occasions, both of us could have fought the big player, and the game could have continued.

We need some system in which second place is not very well rewarded. I beg you to adopt this rating system!!



 
mongolian

Coming back from a leave of absence, I whole heartly agree to this system. I for one have almost quit this game just for the simple fact the game is currently broken seeing so many people 2nd place.

I've played high lvl games, 1800 rated games and low rated games. It has nothing to do with the type of player, there is not enough incentive for people to try for 1st place.

THIS SYSTEM NEEDS TO SUPPORT WINNING 1st PLACE~!

 
whirldpeas

So, let's get a form email going. One hundred of us could send an email to emiel (hmmm, kinda rhymes), with subject: "Please adopt this new rating formula for the sanctity of the game".

In the body, simply write "Please take a moment and review the webpage: http://www.dominategame.com/website.php?page=discussion&id=5323&t=1196960944"

So who had emiel's email address?

 
whirldpeas

And while we're at at, we can do a quasi-resetting of the current ratings, effected by a simple equation:


New rating = 1500 + 0.5*(Current rating - 1500)

This reduces the magntitude of the ratings from 1500 in half.

By doing this, it would reduce the impact that the old rating system had. I figure this would be a compromise from fully resetting ratings and give new players a better chance of moving up the ranks relative to established players.

 
Spir-An

He used to have a page on www.copernica.nl and I'm sure I got an email address off there but I can't find any mention of him on that site now.



 
mongolian

I don't think its a good for the general public to know his email.

Instead, Aero has his email and I have Aero's email. We can all forward a link to a website where everyone has signed they want the new rating change. Or even simpler, create a new thread on this website where people verbally agree they want the rating changes.

 
jar4me

Who cares what you think.

Because you are an admin now, you think we are suddenly interested in your opinion?
Get back to not posting your comments here.



 
mongolian

I have new admirer now! *waves to jar4me*

 
weakling

Resetting current ratings to 1/4 to 1/3 of magnitude from 1500 might actually be better so established players wouldn't get too much of a headstart on the new rating formula.

The formula as it exists today rewards second and sometimes third placing in a positive way for 6-player games and that would be undone with the new rating system. Thus, ratings in the current system are unduly expansive in range relative to what the new system would effect by these rewards. However, the factor of 1600/2000 could be altered to duplicate the current range after the rating resetting.

 
fr1ends4ev

is the game gonna enter public domain or am I missing something here?

 
weakling

So, when does the petition to emiel begin?



 
xbc

bump



 
mongolian

Just so you are aware, a list of requested changed were brought forth to Emiel. He actually did implement the easier stuff. A new rating formula was on the list too, but as you might have noticed hasn't been used. The good/bad is atleast we did get more upgrades out of it, bad being that we could definately have used more to push the game forward. That's a product of not much money coming in or out of the game I suspect and no desire to implement otherwise.

 
CharlesVIl

Your reply is over two years late Mongolian. Keep up the good work

:)



 
xbc

care to detail the upgrades mongo?

 
sixsixsix

i am back! blankenhor=weakling=whirledpeas and others...

just seeing how this games functions....damn suiciders! too bad there wasn't a function of giving people -5 points who eliminate a player by taking at least 4 of his countries to make it happen before the 5th turn. so early = the suicider will get killed as well.

anyway, greetings to mongoloid and chicken and the rest

 
toom

This change is long overdue. The game is not amenable to a binomial distribution of points.

 
tuuttuut

why not simple; take the number of 1st places (no other places), divide that by the number of games you play at a minimum of 20 and thats it; no second places whatsoever



 
lankyhp3

thats a stupid idea. it will only boost the amount of accounts people make

 
breadbaby

everybody should try to play with crazygenius , instasuicide on him and see him whine in the chat , lots of fun!

 
whirldpeas

bringing to the forefront since i've heard many talk about the rating system here. scroll down 3/4 the way and go right to the relative proportion and suggested actual rating points added/subtracted though it would still be impacted by a players rating per the average rating of the players in the game....that's where you'd have to see the formula in action and using the spreadsheet as you can make from copying cells i post here, you can witness it.

 
whirldpeas

Note the beginning part:

NEWRATE = OLDRATE + MAX(1, 4-(GAMES^2/200))*[~]

Games/Multiplier per the max function:
1,4-1/200=3.995
2,4-4/200=3.98
3,4-9/200=3.955
4,4-16/200=3.92
5,4-25/200=3.875
6,4-36/200=3.82
7,4-49/200=3.755
..
15,4-225/200=2.875
..
20,4-400/200=2
21,4-441/200=1.795
22,4-484/200=1.58
23,4-529/200=1.355
24,4-576/200=1.12
25,1
>25,1

This is all to get a new player's rating equilibrated. This could be modified to:

NEWRATE = OLDRATE + MAX(1, 2-(GAMES^2/600))*[~]

if it is felt the prior was too aggressive.

 
Sudz

Moot topic as all development was supended, even before the OP

The ratings formula is, and will forevermore be:


Change=32*(1-(1/(Math.pow(10.0,((OpponentRank-YourRank)/400.0))+1)))

And ran against each player in the game.

 
whirldpeas

You use rank and not rating? Where is the placing in the formula? Is the opponent rank suppose to be the placing? I am guessing for a 5-player game this algorithm would have to be iterated 10 times?

If in a 5-player game, with point of interest on an 1800 player and the average of the ratings of the 5 players are at 1400, what would be the new rating of the 1800 player if:

finished 1st?
2nd?
3rd?
4th?
5th?


You'd have to use a formula to compute these and that formula is lacking the placing. Would you try to calculate the above with the formula you provided? If the rating going into a game is superfluous then the rating formula is far from robust. Separately, why can't there be a better rating system - what is the hold up-time, money, or? I would do it for free. Thank you.

 
Sudz

"what would be the new rating of the 1800 player if:

finished 1st?
2nd?
3rd?
4th?
5th?"

As shown, that would depend on the rating of the other players.

finished 1st = + points from 2nd through 5th
2nd = + points from 3rd through 5th AND - points from 1st
3rd = + points from 4th through 5th AND - points from 1st through 2nd
4th = + points from 5th AND - points from 1st through 3rd
5th = - points from 1st-4th



"Separately, why can't there be a better rating system - what is the hold up-time, money, or?"

Because the owner stopped caring years ago.

 
whirldpeas

Well, would the owner allow an alteration?

I'd like to see one time this system according to the formula you posted. I am a mathematician and model real life stuff, but this is unclear to me from seeing the formula per the wording.


so let's try this for simplicity, ratings prior to the game:

player1 = 1800
player2 = 1200
player3 = 1300
player4 = 1300
player5 = 1400

player 1 wins and the rest finish per the order shown above. I'd like to see where the numbers go.

 
Sudz

whirldpeas:"I am a mathematician and model real life stuff, but this is unclear to me from seeing the formula per the wording."

I believe 2 things to be true after reading that:

1) You, sir, are a poor mathematician and/or a poor programmer*.
2) You, sir, would have ZERO chance of recoding the server and have ANYTHING working afterwards.



*poor in both instances meaning "of a low or inferior standard or quality"

 
Sudz

"player1 = 1800
player2 = 1200
player3 = 1300
player4 = 1300
player5 = 1400"

If player 1 wins, the result for player 1 is 1807
If player 1 is 2nd, the result for player 1 is 1775*
If player 1 is 3rd, the result for player 1 is 1743*
If player 1 is 4th, the result for player 1 is 1711*
If player 1 is 5th, the result for player 1 is 1679*



*assumptive that all other players finish in numerical order (i.e., 2nd wins and 3rd is next in line, then 4th...)

 
whirldpeas

???

Poor, poor poor? Wow. Thanks a lot. Is this the immaturity that starts from the top here? How do you think I came up with the formula posted above? You want to make many more inane unqualified remarks? Do I have to show you more of the math and programming I did to receive a mature reply? What's up with the childish behavior?

Now why did the response show player 1 with different scenarios? Read the English written question again please. There is one and only one scenario posed: player 1 finishes first, player 2 finishes second, and so on. I suppose this is where I am suppose to interject some blasting of your lack of intelligence to match you? I prefer to be above that. Hope you can try to do the same.

 
Sudz

It's not an "inane unqualified remark"; it's the truth.

An average 12 year old could understand the formula I posted. I know this because an average 12 year old wrote it (and did so in javascript).

I wrote what I wrote because I was too lazy to do a "mathematician's" work for him and only ran the formula for player 1. You have the formula and you have examples of the results, if you cannot do the rest of what you want then you qualify by earlier comments.

You could TRY to attack my intelligence; however, given that you cannot keep up with an average 12 year old, I doubt you'll succeed (especially when it comes to code and/or math).

 
Sudz

*you qualify my earlier comments

 
Sudz

Because I was bored....

"player1 = 1800
player2 = 1200
player3 = 1300
player4 = 1300
player5 = 1400

player 1 wins and the rest finish per the order shown above."

player1 win = 1807
player2 2nd = 1264
player3 3rd = 1314
player4 4th = 1282
player5 5th = 1332

*yawn*

 
Sudz

I found an interesting quote:

"What ... showed is how the formula incorporates an extension of the gaussian curve that this site uses as the formula:

32*(1-(1/(1+10^((rx-ry)/400))))"

How is it that you understood this 5 years ago, but it is beyond your comprehension now?

 
rist6

Damn suds.... You're a little harsh on the mathematician. From what i can see, he is at least of above average intelligence. If he weren't, he wouldnt be able to formulate the question. See my formula below:

General population-99%= people capable of comprehending the above conversation.

 
Sudz

Here's the problem with what you are not seeing, rist: he's just smart enough to not know that he isn't as smart as he thinks.

That quote that I found was him actually typing that same formula in 2010.

Now, in 2015, the same formula (that he typed in 2010) is something he claims is not possible.

He cannot have it both ways; either his understanding of math was wrong in 2010 or he doesn't understand math any more.




P.S.- For the record, in modern society, reading a few library books and scribbling some stuff down doesn't make one a mathematician. ;)

 
KRUSHUBOT

Sudz is one jealous sordid person. The formula is sensible and applicable to this game. It's a shame he chooses to feel slighted by the contribution.

 
toom

Dom threat's "fixed point" comment is confusing, but at least it's not degrading as those emanating from sudz.

There would not under this proposed method be a rigid set number of points won or lost because the player ratings versus the average player ratings in the game would influence the points won/lost. There is a standard from which to work from but unless everyone playing all had the same rating, the points won or lost would deviate to a certain degree those standard figures.

  Reply to this discussion

Copernica is a software for e-mail marketing, profile enrichment, websites and short text messages campaigns.