I posted this over on reddit, but the topic seems to be out of the valorant subreddit's depth. So I'm posting it here to hopefully get some coherent thoughts. I'll just paste what I wrote on reddit below, but basically TL;DR:
I think the RR gain/loss bonus/penalty from having your MMR not converged with your visible rank is not effective enough. Instead of just giving RR bonus/penalty based on the difference between your MMR/ELO/position on ladder, it should also give RR bonus/penalty based on the quantity of matches played while your MMR/ELO was maintained within a reasonable deviation without dropping or rising.
Original post:
I'm seeing a ton of posts where people are showing unlucky losing streaks that the RR gain bonus from Rank being lower than MMR is not able to outrun. If matchmaking is working properly, the game pits you against even match opponents, making the outcome of any given game have a high degree of luck.
Basic fact: The closer your opponents are to your skill level, the larger your win and loss swings will be, since more matches will come down to luck instead of skill.
People are playing 50-60-70 games and losing 70-80% of them. But their MMR is not dropping significantly since each time it does drop, they immediately beat those opponents. This kicks their MMR back up to where it should be which prolongs the potential losing streak.
You shouldn't have to play 200 matches just to get your rank and your MMR to converge while other players are going on lucky streaks and getting their appropriate rank after 40 matches. IMHO a big improvement could be made to where if your MMR stays within a certain range above your visible rank over a specified number of games, your RR bonus from MMR being higher than rank increases.
For example (and the RR numbers are examples used for illustration), let's say your visible rank is Gold, and your MMR is matching you against Diamond opponents. As it stands now, you play 60 matches, winning 20 and losing 40. At 10RR for a loss and 20RR for a win, you do not move at all, you lose exactly what you gain.
It would be better, if after the first 20 matches, if your MMR is still diamond and has not dropped to platinum for more than 2 out of those 20 games (because you won those 2), you now gain an additional 5RR for a win and lose 5 less for a loss. If after another 20 matches, your MMR has still not dropped to platinum for any significant period of time, you gain an another +5RR for a win and lose 5 less for a loss.
This would accelerate convergence of visible rank and MMR ladder position for these players who are basically victims of variance. What do ya'll think?
Someone posted a good faith clarification question in response to an example post I linked where a guy had played 40 some matches with a piss poor winrate but the losses were majority 11 to 13, 12 to 14, etc.:
The person has increased gains because they have remnants of their previous act MMR (Immortal) after the reset and the huge MMR difference relative to their rank. This is only usually present at the start of the Episode and takes a while to diminish down if you are playing poorly (the system gives you a period of time to attempt to "prove" yourself to return to your rank). The images in that post are also not in chronological order / don't match up with chronological RR gains/losses which is sort of relevant to this discussion.
Is your original post referring to the steady state ranked system for all players (like in the middle of an Episode where no one has their previous Episode's MMR as a buffer), or are you criticizing that MMR does not drop fast enough for players doing poorly after a soft reset?And here is my reply:
Well I would make a distinction between doing poorly and losing to players lower on the MMR ladder.
You can do poorly win/loss wise but still play with identical skill. Aassuming even match opponents, your victory for any given match will be 50% in the long run. But since a game of valorant has more factors determining the outcome than the flip of a coin, the chances of encountering a streak of wins or losses due to luck is much higher than a coin flip distribution. If for example you were able to play a 1v1 match against yourself, the only thing determining the victor assuming a fair map is 100% luck. So the better the matchmaking, the greater the influence luck has in determining the victor and skill is minimized.
If you lose a game due to luck several times in a row, your MMR will drop beneath your skill level, increasing the impact player skill has on the outcome of the subsequent matches since your opponents become weaker.
Since MMR is raised when defeating opponents higher on the ladder, and lowered only when losing to opponents lower on the ladder, there's no incentive or even a reason for MMR to drop because you lost to someone with a higher MMR than you by a single round. You have to lose to the people beneath you. Assuming your skill remains the same, as soon as you match against those individuals, your MMR goes right back where it was before because your victory is far more likely since the influence of luck is lowered.
So what I'm getting at, is people that are being matched with even opponents after a soft reset, are potentially getting long strings of unlucky matches against equal skill opponents, punctuated by victories against weaker opponents as soon as their MMR drops. This keeps their MMR from dropping to their visible rank, but since their actual win loss rate is still far below 50% even after 40-50 matches they never receive enough RR to raise rank to match that higher MMR.
There was another thread posted I couldn't find where the guy lost 41 and won 18 or something like that, but still continues to match against opponents in their original MMR, and I think the above is a plausible explanation as to why this is happening. If you lose 41 out of 59 matches, your opponents should either get easier because you aren't as good as you were, but if they are not, how else do we explain a 1:2 win loss ratio other than the above?appreciate any who take the time to read and respond
Discord