SE4 Rating System

[Slynky]

OK, here is the "GeoSly" formula in a real game. Remember, Geo's base formula was intended to lessen the effects of losing while rewarding a person able to pull off a win in a multi-player game.

Game Players in order of finish (with ranking at the time they entered the game):

Alneyan - 1000
Slynky - 1104
Master Belisarius - 1050
Gecko - 990

Round 1:

Gecko out receiving -26 points
Remaing players receiving +8 each
All Ratings adjusted

Round 2:

Master Belisarius out receiving -32 points
Remaining players receiving +16 each
All Ratings adjusted

Round 3:

Slynky out receiving -42 points
Alneyan receiving +42 points
All Ratings adjusted

Net results:

Gecko: -26 points
Master Belisarius: -24 points
Slynky: -18 points
Alneyan: +66 points

Here are the points layed horizontally as we have been seeing them in examples:

+66, -18, -24, -26

At first, this looks a bit unbalanced from what we have been looking at in other examples that used the sample rating of 1000 for all players. In the above, we see that the person with nearly the lowest score won the game! And the person with the highest score (me) came in second. This should explain the (kind of) wild swing in what we have been seeing when explaining the formula in the earlier Posts...the person who should have won it came in 2nd and was beaten by the person with nearly the lowest rating. Also, the person with nearly the lowest score won the game over some higher rated players.

Here is the "horizontal" for 4 players rated at 1000 points each:

+58, - 6, -22, -32

This looks a bit skewed, also so I tried to figure out why and I think the Last computation (when only 2 players remain) should not get the modifier (# of players squared) applied and just be a straight out computation as in a 1 x 1.

This would change the "horizontal" for 4 players rated at 1000 points each to:

+42, +10, -22, -32

So, what do you guys think? Does this look fair? (and your vote doesn't count, Alneyan... ).

[parabolize]

sounds good to me

[Baron Grazic]

If 4 equaled rated players play, 2 should gain points, 2 should lose them.
This seems to do this, and your post didn't have any mathematics, so you didn't confuse me.
Test it out for 6 players, all with 1000 points, and let me know what place I have to get in the Pairs game so I don't loose points.

[ March 29, 2004, 03:43: Message edited by: Baron Grazic ]

[geoschmo]

If the examples you give are typical, it will probably work out fine. My only concern would be that it doesn't appear to be "point neutral", that is there are not an equal number of total points gained as lost. It's not extreme in these examples though. In fact in these examples it appears there are a couple less points earned then lost. As long as it doesn't end different in real games. You don't want to have a bunch of point inflation, for reasons already stated in the thread.

[Slynky]

Originally posted by geoschmo:
If the examples you give are typical, it will probably work out fine. My only concern would be that it doesn't appear to be "point neutral", that is there are not an equal number of total points gained as lost. It's not extreme in these examples though. In fact in these examples it appears there are a couple less points earned then lost. As long as it doesn't end different in real games. You don't want to have a bunch of point inflation, for reasons already stated in the thread.

I didn't want to bore people with all the math. I suspect final results and what the prospects might be if entering a rated multi-player game would be of the most interest.

I think, though I have not bothered to calculate decimals, the slight disparity in points lost and gained (overall) was due to my rounding when I divided the computations by the "gaining" players. Rounding up and down.

More importantly, I was looking for some opinion as to NOT using the "square root of number of players" modifier in the final computation (between 1st and 2nd place)...it seems to put a big gap in there.

[primitive]

Yo da man Slynky. Looks very good for 4 players.
However; Does it work 3 players or for larger games ?

Two suggestions.
A: Try using the Square of Players - 1 instead of Square of players as the factor. That way it will work for all sizes (even 2 players).

B: The same adjustment you did between first and second should be done for every calculation:
3rd player get whatever points he has accumulated - a 3 player loss (square of 3 or square of 2, see A)) and so on.

Using both suggestions would give these results (all 1000 point players):
37 / 5 / -13 / -28

But, as I said before: Its your call Slynky, these are just suggestions

[Joachim]

Fire v Joachim Rated Challenge

Can you kick us off when you get a chance please Asmala.
Cheers.

P.S good luck Fire.

[primitive]

Okie Dokie.
This work for all games. For easy use:
A: Calculate the points for the knocked out player towards the average of the remaining players
B: Multiply with the Square root of the number of remaining players.
C: Distribute the winning points evenly between the remaining players.
D: Repeat until game over.
Nothing to it.

Example 4 (1000 points) players:
Loosing player (4th in example) = 1 loss * Root (3) = - 28 points
Second Last (3rd) gets 1/3 rd of the points from the loosing player + (- ?) 1 loss * Root (2) = -14 points
Third Last (2nd) gets 1/3 rd from the looser, + ½ from 3rd + (-?) 1 loss = 5 points
Winner gets 1/3 rd from the looser, + ½ from 3rd + 1 win = 36 points

Should be an easy formula in here, but its been 20 + years since I’ve done any math so I have no idea how to write it down

[Slynky]

Thanks, "Big P" .

To sum up...if I understand, our formulas are exactly the same except you compute the "multiplier" based on the amount of people left in the game at the time a loser is being computed when I was using the same multiplier (in the case of a 4-player game...the root of 4) for EACH round of computations. Perhaps that was why it was swinging wildly as it got down to the Last computation.

Do I have it right, now?

[Slynky]

[EDIT] Piece of crap board..."sit back, we are taking you back to the forums"

And there it sits. One never knows it has been posted...so this is my way of deleting the double post .

[ March 29, 2004, 15:17: Message edited by: Slynky ]