Thought i was clear that 3.88% represents a ridiculously unrealistic assumption. That all Diamond players are still active, online, qued up, on a gold smurf and there are not players with multiple accounts in diamond. There would be 0 people in diamond que or on a diamond account to make this value happen.
If you used a Binomial formula I believe it looks like this when filled out for the values of 5 games and 4% odds. (5÷1)×0.04^(1)×(1-0.04)^(5-1)=0.169869312 . Its possible I Errored in filling out the variables properly, feel free to check the math and get back to me if I derped the variables.
Honestly theres a lot of extra steps and assumptions and the above generated values are made with Bullshit unrealistic values. The numbers I checked when rounded and aberrated are result in **roughly 3% odds a match Is decided because the enemy team had a smurf. **so 3 per 100 matches.... but.... that's a really misleading number and I want to be clear on it not remotely representing an accurate expectation for the game.
Remember we are looking for the possibility for the smurfs to impact the 1 consistent player. Its not total # of matches impacted. Its the odds of Player A's games being **negatively **impacted.
Also. Lets be honest that the math I used is bad btw. I did not account for the Adjustment of % values for population size. I don't know a site that tracks population and rank distribution on a daily biases and I don't have access to riots numbers or their API personaly. I do hope you wont hold this agent me because ultimately the fundamental assumptions that smurfs would be so common they make an impact exclusively on 1 MMR range is pretty fucking stupid.
I was clear with what slanted my values, I don't mind that its a bit Biased to support my point because there's no accurate values generated from inaccurate assumptions.
The argument to claim there are enough smurfs in gold to demote a player to silver is just trash. Especially when you consider that MMR is whats used by MM not the Title. If the population size and skill gap grew to a point that Gold now is represented by the skill of what used to be diamond then inherently the MMR system would redistribute lower skill players down, but... that's its job? Sort the ranked ladder by Probability to win Vs players. All ladder movement is done to reflect the closest representation of your average to reflect when your in a range your at or near 50% odds.
If the population becomes a majority where your now a minority and getting Stomped… then its working properly to move you way from that range. This realistically can happen in situations where some one loses their ability to play in a class interval. Like if a OTP gets their main reworked and sucks at the rework, cant play anything else as well as their old main? That dude can and should sink. If some one buys a boosted account? Can and should happen.
Next up.
The problem is that the K factor slows down at all. It scews ranked progress one way or the other, keeping people too high or too low. It doesnt need to slow down
Lets look at the OG Arpad Elo Formula. Rn=Ro+K(W-We). This is 1v1 chess adjustment. Behold! K factor. Why is K a variable if not to have Variance?
What K does is track system uncertainty. A player at 100% win rate out of 1 game is not as impressive as a player at 99% out of 100 games.
There is nothing in K or MMR that screws progress in any direction. The results are from match results. People seem to think some how the systems looking to lock you into 50%. when realistically its goal is just to give you a raiting that generates 50% results. When your not near that? your raiting changes to reflect the gap in expectation V reality. Climbing is just expectations not matching reality so you gain points until the two are more in line. Same for sinking.
What high K does is help lower calibration time. Early trends when impacted by luck are covered by large numbers in the end, so you just see the extreme cases where a high K account that has impact get moved more rapidly into a range you find their probability and results be more in line.
Its needed for 1v1, its needed for 5v5 with RNG teammates even more, because you do not want clutter. It helps keep folks out of a range off luck by having slowdowns, reduces the impact high uncertainty accounts make on low uncertainty accounts by allowing it to be a variable.
The main Error you see people make is the assumption that you can treat a Teams K variable as an average or the same value when its RNG teammates. K is how you protect account raitings when there is uncertainty gaps and thats why you track it by player.
It absolutely needs to slow down, you can also have triggers on a MMR system where it speeds up. Like a soft reset.
I have NO REASON to keep playing because even if i maintain 70% winrate I will get demoted.
No you wouldn't. Don't be dramatic. Your MMR would increase, maintaining 70% would adjust probiblity expectations and your account would shift to the proper range. It takes a massive MMR gap to demote from divisions 4 and last time I checked Riot literally sends players a warning alerting them that they are at risk of dropping down a tier. So as your MMR is always a net gain at 70% this is in effect impossible to achieve.
Your rating is built on a probability model. The formulas run the same for everybody, despite the # of variables the rules are universally the same.
with your current win % your gonna be climbing pretty fast.
