Pedro, chessmec, Walter: "I am strictly against any resetting of the BKRs.", "ACK", "I'm with you on this. Why reset them?"

I'm surprised by such a black and white sentiments. Can you not see the usefulness in any of the instances given? Surely the Kingpin1 (at #21) (because molotov now resolving itself) and 515484 cases have some merit? And what about that cheat that grenv pointed out? You want all these guys to remain where they are? They will be stuck at the top of those lists forever.

Pedro: From the formula that you've given it's quite apparent that the rating is indeed the average from all matches played (although the formula page uses the term tournaments while BK actually counts matches). Do the maths with a win against 2100 and three losses against 2300 - you say it should be around 2500. The formula, as I calculate it, produces 1975.

ProvisionalRating = (RatingSoFar x NumGamesPlayedSoFar + OpponentsRating +/- 400) / NewNumberOfGamesPlayed.

The calculation is done after each individual match:
(0000 x 0 + 2100 + 400) / 1 = 2500
(2500 x 1 + 2200 - 400) / 2 = 4300 / 2 = 2150
(2150 x 2 + 2200 - 400) / 3 = 6100 / 3 = 2033
(2033 x 3 + 2200 - 400) / 4 = 7900 / 4 = 1975

The final result is the same as adding them up and taking the average at the end:
(0 + 2100 + 400 + (2200 - 400) x 3) / 4 = 1975

Those three losses matter. It's common sense, too. What use is a formula where the first match result is a major factor and the next 24 are trivial?

grenv: According to the formula the first 25 games are averaged in this way.

Rex Nihilo: molotov for playing. I think I'll invite her again. Lol, you've certainly got your work cut out (at move 4), but as you say, early days. ;-) Good luck to you both.

Reza: That's a tough break. As you say, you've got to take your lumps and learn the lesson until there's that other condition for protection. ;-) Feature request Perhaps the No Unrated players could have an accompanying No provisional players.


Edited to add
Here's the calculation for molotov's backgammon rating:

Date ended ..... Opponent....... Them --> Contribution to molotov
28th Sep 04 .... bobbobby62 .... 2045 --> 2445
29th Sep 04 .... HEATHER ....... 1914 --> 2314
15th Oct 04 .... sko ............... 2364 --> 2764
15th Oct 04 .... taikoki .......... 2208 --> 2608
17th Oct 04 .... Rex Nihilo ...... 2479 --> 2879
.2nd Feb 05 .... Rex Nihilo ...... 2592 --> 2992

Total = 16002
Average = 16002 / 6 = 2667

The opponent's ratings were taken from their rating graph for the day that the game ended (which is presumably the value used) but as it's not possible to pinpoint actual games one or more of my figures may be wrong. And indeed:

Molotov's actual rating is 2696, so the calculation's in the right area.