Nightbird, on 15 June 2019 - 03:15 PM, said:
Bad example because regions are not intentionally fixed like divisions are. Some people in region A plays against those in B against those in C against those in A. Another reason it's a bad example is the skill seen in comp play should not be viewed as a systematic difference in skills like you are seeing. The main difference is in the population, the top 12 players in a region with 10,000 players will average a lot higher than a region with 1,000, simply because with a normal distribution of skills, the top will be more standard deviations away from the mean.
Elo is only proven to do what you say in a controlled environment with fixed teams or individuals, fixed match schedules, with fixed opponents. In a random team environment, there is no proof for these claims. If you'd to make some proof, I look forward to reading it,
My suggestion is still to not bother starting with Elo, the assumptions that lead to it working simply don't exist in MWO, better to start from scratch.
How about that Elo is the core of TrueSkill, for a reason. Elo is proven to work in more matchmakers than you can easily count. Like, for example, Overwatch uses an Elo based system for their matchmaking and so does pretty much every competitive online FPS. They may have their own take on a K factor or have additional tweaks based on their unique setting but it's the core of almost every matchmaking system out there. Even stuff like Glicko-2 is based on Elo but has an additional factor to try and measure how reliable someones score is.
At their core though it's all an Elo system. For a reason.
Also there really isn't a 'start from scratch'. All Elo does is exactly what you have to do; it creates a value to identify a players value, then it modifies that values based on wins or losses and it attempts to make that adustment bigger or smaller based on the relative value of who they played against. This is just as relevant in a pug queue as a group queue, for an individual or a team. Like any and all matchmakers it assumes that the value it has currently is incorrect and that the match is NOT perfectly balanced - which is why your score adjusts by varying degrees after the match.
There isn't some new version of Elo you're going to create that does something different, because there isn't something different to do. Every serious matchmaker out there has the Elo equation in some form at its core for the same reason the mean = sum/n. Because if you want to get the mean, you're going to divide sum by n. You want a matchmaker to try and create balanced matches it needs to value each player an arbitrary value then adjust the value up/down based on win/loss and the relative value of who they played. Solo or team or whatever the goal is the same and it starts with exactly what the Elo equation does.
Find a better way to create a matchmaking process and sell it to Microsoft to replace TrueSkill and Overwatch and, well, everyone and you'll be rich and famous.
Until then start with what is already proven to work (as well as anything we know of is going to) and that's Elo. Tweak the result, have variables for how much you trust this or that factor, but the core of it is still exactly the same.