Since people (including me) are complaining about how loosing one fight due to RNG (or some other anything) to an easy oponent massively lowers you result on that day - which is true, as you loose at least half the maximum points I’ve come with an idea to make things less oppresive.
The new idea would be is proposed in two somwhat similar variants:
Loosing fights brings you a maximum reward points penalty for following battles. Each fight reduces rewards from following fights by a set amount (50,100,200,400). With no losses this variant works same as current system.
EXAMPLE 1: You loose first battle (damn RNG!) which means all your following battles pay 50 less. You win the next 4 battles though.
So you would get 0 [loss] + 50 [win] + 150 [win] + 350 [win] + 700 [win] = 1250 instead of current 750.
EXAMPLE 2: You loose 4th battle (awesome defense) so your last battle pay 400 less. You win all other battles though.
In that situation you would get 50 [win] + 100 [win] + 200[win] + 0[loss] + 350[win] = 700 instead of current 750.
Same as above with two chnages: you do not advance to next tier without winning previous one and you have unlimited fights per day (up to five wins of course). With up exactly one loss this variant scores same as variant 1, you have to win the revenge fight though. With no losses this variant works same as current system.
EXAMPLE 3: You loose third battle twice (tough one!) so bouth your following battles pay 2*200=400 less. As in other examples you win all other battles.
This variant would get you get 50 [win] + 100 [win] + 0[win] + 0[win] + 350[win] = 500 instead of current 350.
Both of the systems give points much more based on quality of your wins, also severly penalizing for loosing against high ranks. This change adresses most of popular issues with current GW scores, such as: no way to recover after random loss with easy oponent, no way to fight paragon when you don’t have a top tier army, to name a few, while keeps current model of great uncertainity and need for use of proper teams.