Leaderboard points computation "ultimate question"

SirGorgoroth

Exemple's context: Leaderboard's scores from rank 1st to 150th: 1st = 1000 -- 50th = 700 -- 100th = 200 -- 125th = 175 -- 150th = 25

Exemple's event: Player S rank and score are 93th with 350pts and he played 4 random matches that day, all against the same opponent, Player O: rank 60 with 600pts.
Player S won the first 2 match and lost the 2 last one.

Final score on the board after each match:
Match1- P.S(+075pts) = 81th with 425pts / P.O(-085pts) = 65th with 515pts;
Match2- P.S(+050pts) = 69th with 475pts / P.O(-150pts) = 93th with 265pts;
Match3- P.S(-130pts) = 87th with 345pts / P.O(+050pts) = 91th with 315pts;
Match4- P.S(-150pts) = 115th with 195pts / P.O(+020pts) = 88th with 335pts

Question: How can one end up to a much lower rank by winning 2 and loosing 2 matches facing the same opponent and same situation for the opponent, knowing that all other scores on the board did not change? How does one get from being the 69th best player to the 115th by facing and loosing twice facing the same opponent, knowing that there is no win/lose global ratio in the computation? Is it not most likely to create situation where a player having with 300 wins ans 100 lost ending up at rank 110th behind some players ranked 80th that only played 5 matches over the month (2 wins and 3 lost).

Is it not a score computation equation that is incoherently oppose to the nature of a ranking system rewarding on performance?, if it is not rewarding players that put the efforts and get good ratio reflecting they're score wich determine also the reward they get, would it not be called discrimintation and anti-sportive spirit?

Personofsecrets

@imPDA may be able to answer why this happens.

Your opponent probably had poor MMR.

imPDA

These values most likely do not depend on any global values. Your and your opponent's MMR are the only two values that matter, and specifically the difference between them.

In general:

- If you have a low MMR and lose to a high MMR opponent, neither you nor your opponent will suffer a big change.
- But if you win against a high MMR opponent while having low MMR, you will gain a huge amount of MMR and your opponent will lose a huge amount.
- If your MMR values are close to each other, the gain and loss will be something average, like 50-70, etc.

But this does not align with your data well, so you either have incorrect data or something changed in the ToT MMR calculations. I personally collected data from high ranks with 1000+ MMR, and it was mostly predictable.

First, I would like to mention my addon: ImpressiveStats. It saves stats for all ToT matches (BGs and duels, but you can turn them off), and it is as precise as if you manually gathered the data (I am planning to improve it shortly). It can save your data for future analysis, and it also shows your opponent's MMR in the match. More about the addon is in the description.

Second, I tried to find out the exact formulae, but I stopped playing ToT actively and never finished it. My assumption was that the MMR you gain and lose depends solely on the MMR difference between you and your opponent before the match (it isn't that simple - there's something else, but even with this assumption the trend is close to linear).

On the chart below, you can see the difference in MMR between you and your opponent (x) and how much MMR they gained/lost (y). E.g., if the difference is close to 0, you either gain ~100 MMR, ~125 MMR, or lose 80 MMR.



I am still interested in this btw, so if you can install the addon and share your data once a week or so, I will process it and try to figure it out.

Basically, anyone can join. All I need to see are the SavedVariables files- you can send them via Discord or any other way you want. I will extract the data and process it. Everything is anonymous; I only need leaderboard place, player and opponent MMR before and after a match to draw a chart like the one above, group the data, find all the variables, and fit the dependency.