Tournament Performance for GoR

[Javaness2]

What is the best way to calculate a Tournament Performance Rating(TPR) for GoRating?

FIDE has a system here https://www.fide.com/component/handbook ... ew=article
Which has a formula TPR = Average Rating of Opposition + dp
Their definition of dp confused me a little, but it seems just to involve taking your percentage score, and lifting the corresponding number
100% : dp = 800 etc

Is it workable with handicap having to be taken into account?
Or is there an alternative algorithm?

[Uberdude]

I've done this by putting in the tournament results to the EGD gor calculator and then changing their input rating until it is the same as the output rating.

[Javaness2]

That's certainly one way to do it, but I suppose you must need a program to iterate all particapants to near zero
FIDE's approach has the advantage of being easier to calculate

[Laman]

Is it workable with handicap having to be taken into account?
Or is there an alternative algorithm?

GoR description says:

The system also allows to include handicap games assuming that the rating difference D is reduced by 100(H-0.5), where H is the number of given handicaps.

you can adjust the opponent's rating (resp. the rating sum) in the same way for the purpose of calculating TPR.

i think i'd prefer Uberdude's way. it just fits the rating algorithm nicely. FIDE way appears quite ad hoc to me. but i could be just missing the deeper reasoning for the dp calculation.

[Uberdude]

Apart from being an iterative method (instead of closed form), the downside of my way is it gives +/- infinity if you have all wins/losses. You can anchor it with a fake loss/win to a super strong/weak player but that's a bodge (and the rating choice of the bodge affects the result). Then again with all wins/losses infinity is kinda the real answer without any extra information.

[Javaness2]

Maybe this infinity is why FIDE have their +-800 limit in place, you do in effect need some sort of compromise.
Maybe using just the third iteration of the GoR update algorithm across all players in a tournament is enough.

[Laman]

Apart from being an iterative method (instead of closed form), the downside of my way is it gives +/- infinity if you have all wins/losses. You can anchor it with a fake loss/win to a super strong/weak player but that's a bodge (and the rating choice of the bodge affects the result). Then again with all wins/losses infinity is kinda the real answer without any extra information.

true that. but then again, straight wins against stronger opposition should be further in infinity.

i've learnt about this flavour of performance rating from Arimaa tournaments (failed to find the source at this moment). they indeed anchor the performance by including a dummy draw for each player. with two variants: UTPR (unseeded), where the draw is against a dummy with some default rating, and STPR (seeded), where the draw is against a dummy with the player's pre-tournament rating