Go back
Rating Formula Unveiled!

Rating Formula Unveiled!

Help

RM
Adm¡nistrator

Rio Grande Valley

Joined
16 May 04
Moves
13043
Clock
03 Sep 04
1 edit
Vote Up
Vote Down

Got your attention? hehe, good. ;-)

I was playing with the ratings calculator (i dont' remember how i got there) and I figured out the output table (instead of the actual formula)

I'm pleased as punch at myself, and i thought I'd share my joy and insightes with all (unsolicited!)

The following the MATRIX. It needs a little explaining.

SCENARIO: Player 1 has a rating of 1200 , Player 2 has a rating of 1300

(DR) = The difference in ratings. ex: 1200 - 1300 = |-100| = 100
(LR) = If the underdog wins (in this case, player 1), they get this many points.
(HR) = If the higher rated wins (in this case, player 2), they get this many points.

DR |LR |HR
15 -16-15
25 -17-14
50 -18-13
75 -19-12
100-20-11
125-21-10
150-22-9
175-23-8
200-24-7
250-25-6
275-26-5
300-27-4
350-28-3
400-29-2
500-30-1
600-31-0

More than 600 doesn't change the awarded points.
Less than 15 doesn't change the awareded points.

Round to the nearest DR and that's what you'll get awarded.

NOTE: Your opponent will lose what you win, plus one. (if u win 20 points, they lose 21)

Ragnorak
For RHP addons...

tinyurl.com/yssp6g

Joined
16 Mar 04
Moves
15013
Clock
11 Sep 04
Vote Up
Vote Down

Is that right? If so, its fairly clever alright. Too tired and hungover now to check it.

piderman

Zeist, Holland

Joined
11 Sep 03
Moves
19384
Clock
12 Sep 04
2 edits
Vote Up
Vote Down

I think somewhere a flaw entered your table. If the players have equal rating, the winner gets exactly 16 points, and the loser loses exactly 16 points. There will be no difference. Furthermore, at a difference of 600 points, the players will lose/win 31 points if the lowest rated wins, else 1; at a difference of 1000, 32 else 0. I think you should recheck your table, or maybe the calculator you used isn't correct.

Note: this only applies to players both below 2100 points.

M

Joined
12 Mar 03
Moves
44411
Clock
12 Sep 04
Vote Up
Vote Down

Originally posted by piderman
I think somewhere a flaw entered your table. If the players have equal rating, the winner gets [b]exactly 16 points, and the loser loses exactly 16 points. There will be no difference. Furthermore, at a difference of 600 points, the players will lose/win 31 points if the lowest rated wins, else 1; at a difference of 1000, 32 else 0. I think you should r ...[text shortened]... alculator you used isn't correct.

Note: this only applies to players both below 2100 points.[/b]
I think you are right. The net result of a game is zero.

f
Quack Quack Quack !

Chesstralia

Joined
18 Aug 03
Moves
54533
Clock
12 Sep 04
1 edit
Vote Up
Vote Down

Originally posted by Mephisto2
I think you are right. The net result of a game is zero.
this is true only if the same formula is used for each player (with a negative applied to one of them)

when one is over 2100 and one is under 2100 this is not the case.

- but i am sure you see that every day! 😉

o
Paralysed analyst

On a ship of fools

Joined
26 May 04
Moves
25780
Clock
12 Sep 04
2 edits
Vote Up
Vote Down

Originally posted by piderman
I think somewhere a flaw entered your table. If the players have equal rating, the winner gets [b]exactly 16 points, and the loser loses exactly 16 points. There will be no difference. Furthermore, at a difference of 600 points, the pl ...[text shortened]... ect.

Note: this only applies to players both below 2100 points.[/b]
Are you sure you are reading the table correctly? It's not saying that in the SAME GAME one player goes up 16 and one goes down 15. It's really saying the points change depending on whether the result is the expected (Higher ranked wins) or there is an upset (Lower ranked wins).

So, the first column is saying: 'upset' = winner goes up 16 and loser goes down 16. Expected result = winner goes up 15 and loser goes down 15.

Apparently that means that with equally ranked players the result is +/- 16, so it fits in the 'upset' column.

With the 1000 points difference, maybe he/she didn't calculate quite far enough.

EDIT: Okay, I see you are disagreeing with the 600 figure as well. Maybe you WERE reading it right partially. But as I said, the table is NOT suggesting there will be a difference in how much each player's rating is affected by a single result.

piderman

Zeist, Holland

Joined
11 Sep 03
Moves
19384
Clock
12 Sep 04
Vote Up
Vote Down

Originally posted by Red Matrix
NOTE: Your opponent will lose what you win, plus one. (if u win 20 points, they lose 21)
orfeo, I think this clearly states that what one player wins isn't the same as the other loses.

o
Paralysed analyst

On a ship of fools

Joined
26 May 04
Moves
25780
Clock
13 Sep 04
Vote Up
Vote Down

Originally posted by piderman
orfeo, I think this clearly states that what one player wins isn't the same as the other loses.
Hmmm, you're right!

I didn't actually see that note. I was only looking at the table itself. Usually I pride myself on reading the fine print.

Well, in that case, the table needs a little more work. Very good basic idea though.

t
Retired knight

Chess Castle

Joined
09 Jun 02
Moves
34809
Clock
18 Sep 04
2 edits
Vote Up
Vote Down

Originally posted by Red Matrix
Got your attention? hehe, good. ;-)

I was playing with the ratings calculator (i dont' remember how i got there) and I figured out the output table (instead of the actual formula)

I'm pleased as punch at myself, and i thought I ...[text shortened]... ll lose what you win, plus one. (if u win 20 points, they lose 21)
id like to learn the formula

s

Brazil - PR

Joined
29 Jun 04
Moves
11276
Clock
15 Oct 04
1 edit
Vote Up
Vote Down

id like to learn the [b]formula[/b]
The formula is in another post, thanks to Ragnorak.
The link is http://www.redhotpawn.com/board/showthread.php?threadid=15287 .

Cookies help us deliver our Services. By using our Services or clicking I agree, you agree to our use of cookies. Learn More.