Positional MM is still in place for a majority of the ladder despite the removal of positional ques. That being said, its just my guess this impacts LP as I don't really know what this does because i don't know how Riot generates the We variable or what ever its equivalent is to their system. My assumption is that you now have hidden MMR for each role with a soft cap ( all values are tethered to each other to not let them exceed to many Tiers) and that then is applied to the context of your current Division and Que in relation to MMR value and changes your Gains.
Unless Riot gives up their adjustment formula and what the variables represent and why I really am just taking a stab in the dark.
I think its related to We or riots equivalent which is the gap between probability expectations which then modifies the # of points you can gain off a win and how much you lose off a loss.
Rn=Ro+K(W-We). Elos adjustment formula, in english this reads : Your new raiting (RN) after a game is built by takeing your Old raiting (Ro) prior to the match. Looking at the result (W), which is a Win or Loss in the context of league for the match after calculating system probability expectations (We) that's then multiplied by system uncertainty(K).
Also that's an aberrated formula, each variable has a lot more math behind it.
I happen to think its gonna be fky until all of last seasons accounts have played the 100 or so games needed to adjust for new tiers. Pretty shure they just dropped the Value of Tiers for lower ladder. B5 to challenger was 27 Titles that represented all players. b5-b4-b3 etc etc to challenger. This year I4 is the bottom and.. thers… 27 titles that represent all players. i4-i3-i2 etc etc to challenger. Looks like this. 5+5+5+5+5+1+1 Old To 4+4+4+4+4+4+1+1+1 New.
If they overlap the same with old and i4 is the same range as b5 then you end up with the titles becoming same value as old at plat but see GM eat into Diamond.
Assuming Iron 4 = b5 and the MMR range each tier represented did not have its value changed or moved then Iron (4-1) is b5-b4-b3-b2. B(4-1) is b1-s5-s4-s3. S(4-1) is s2-s1-g5-g4. G(4-1) is g-3-g2-g1-p5. P 4-1) breaks even with last year and is p4-p3-p2-p1. Diamond drops down a division near the top and is d-5-d4-d3-d2. Gm is D1. M is M C is C.
This probably was riots solution to stuck players not trying to play competitively. Make the lower ladder where folks get stuck fast and give up and become toxic for allies because they just wont try for a win as much as they should put in a bit more effort because their "Rank" Got worse. Gives the toxic stuck Diamond Players that you see Diamond boards users complaining about more incentive to make it to the top with a Fancy new title. that feels closers because its "4" divisions not 5.
Frankly... I'm not as nice? Rather then change player perception Id just tell people to grow up and try or gtfo ranked ques because half assed attempts hurt teammates who want to be competitive. Then if they don't? Id run a system that aggressively removes players who don't try.
Of course they could also shift the values as well and widen MMR ranges to better reflect what they would like to see as set skill expectations. Could easily downshift the ladder a bit, widen the MMR value for what when accurate gives you your title and see the player % Population breakdown shift to what every they desire. Its their system after all.
This is just my guesswork, I showed my logic in my usual way. The fucking why is there always a wall when this dude posts WTF KAI style.
I really wish riot just took the hit for player participation in ranked and made their Titles be locked behind system expectations for accuracy. People would bitch about it being grindy but they expect 0 to 50 games to accurately represent some ones skill? Stupid assumption if you ask me but player do this shit daily.Then again... there is a lot of folks who might just treat the system like a provisional system the entire time and just... add more noise and hurt the quality of the ladder.
People suck sometimes and ruin a good bit of math because they just wont behave right. Any fool proof method underestimates the ingenuity of fools as Douglas Adams once said.
Like... A Silver account with only 3 games is really not the best tool to gauge skill? Players also improve so some one who fixes their gameplay issues that dropped them into bronze Can become a better player. A one trick who can only play 1 thing well whos main gets reworked can become deadweight to their teammates overnight and deserve to sink.
There's no magic ball to predict the skill of some one with low data... that's why MMR is self correcting. Folks forget that skill is an unknown variable that has to be tracked by proven impact, this is way harder then you think. You can sort Coins by size pretty easy by making a them roll down a slope with holes cut into it that correspond with their size. Ok.. that's easy. Smallest to largest coins. BOOM done. sorted by size. but if we want to sort them not by size but value thou. Know the values and you can add a few extra steps then sort them by value. It just takes a bit longer, might need to have a few extra slopes to sort them out to the right order. But... Try to do this with coins from a country you don't know the exact size and you just have a vauge sheet that tells you some random info like Coin A is smaller then coin B but valued higher then Coin E which is larger then B and D, also C is the highest value... eh... You get the point. You can use logic with some basic data.
The basic data used for Elo and MMR is probability theory. Anything that does not match the theory is moved to a location where the results consistently match the theory. if % are not correct it changes till they are. And that % is simple. 50%50% when your playing some one with the same Rating, Different expectations off different probability models for every other impact of matchup off the ratings in relation to the distribution used.
It works eventually, its impossible not to. Its never 100%. Its made for a range not an absolute value. MMR and ELO is just your mean average performance vs every player you vs and then reflects probability expectations from the result nudging you to the range your results match expectations and then the system is accurate in its grasp of your skill... till it changes.
