Assume you have the following information:
-> Player's 1v1 winrate
-> Amount of 1v1 games played
-> MDL ELO
-> 1v1 Ladder ELO
-> Amount of Idle Achievements

Is it possible to create a function of these parameters which would return a number (for example (amount of games * winrate) + MDL ELO + 1v1 ELO - Amount of Idle Achievements*100), which could be used to estimate player's skill (the bigger number, the better the player would be at Warzone)?

I have come to an (obvious) conclusion, that the more games a player played, the better winrate, the better ELO, the fewer idle Achievements has, the better in Warzone player is, but I am not sure how to put the weights properly/ if I haven't missed an important, easy to find indicator of player's skill.

https://www.warzone.com/Profile?p=6834905062 this guy must be a champ

https://www.warzone.com/Profile?p=15142797657 this guy must be a champ

oh wait, he just stalled on a joke ladder. nvm

no shit

@καλλιστηι

Is it effective to use idle achievements to assign a skill level? I think that it's kind of subjective whether you like and play idle or not

it's simple.

ask berdan

MDL Elo is the only thing there aside from 1v1 ladder rating that I'd put "any" emphasis on, but even the 1v1 rating is very subjective. You can't stall for rank 1 on MDL, so it's a lot more meaningful, but it's also not template-specific, so it won't show how good someone is at a specific thing so much as how well rounded they are overall.

Is it effective to use idle achievements to assign a skill level?
Yep. The more you play idle, the worse you are in classic. On average.

Berdan who is the champ

arrow is

I would like to see a ranking system that ranks the last 100, 500 or 1000 games, by map. Much of winning is map specific.

Especially I would like to see ranking systems by map.

@Fizzer: Dr. K would be happy to design and test such a system, if I had access to the game database. Or a subset, even.

MTL only will give you a more precise skill estimation than if you add any additional parameters

I think aura guardian is jealous cause he couldn't poach silvers

The following are pretty irreverent:

• Player's 1v1 winrate - doesnt break down by opponent strength
  
• Amount of 1v1 games played - could be lottery games, not analysing games afterwards so no improvement, isn't template specific
  
• Amount of Idle Achievements - idle is a different game format altogether
  

Quickmatch rating can be misleading. Its activity based, not ELO based. Someone who's actually good could have lower rating than someone else who plays lots of games but is only 'average'.
Real time ladder rating - very few players took part and was common for there to be no competition to get first. The ladder was ladder removed because of that.

MDL is best indicator. Has many templates, covering a variety of settings and maps/bonus systems. Most popular in the strategic community. However there's probably more players in the 1v1 ladder than MDL so there isnt as much of a skill gap.

1v1 ladder is the next best option. Doesn't use third part site and is an unlockable feature with gets advertised once you reach the right level. 1500 rating is what is call average (not knowing/not checking move order, not making correct amount of picks, not getting good map coverage, not checking game history - to predict were the opponent is or to see which bonuses are wastelanded before expanding into them). 1700 is better than average. Top 50 are significantly better than average. Top 10 are best.

next best indicators are seasonals and awp (and cl performances). 1v1 ladder means next to nothing

Edited 4/18/2022 17:34:37

Player's 1v1 winrate - doesnt break down by opponent strength
Amount of 1v1 games played - could be lottery games, not analysing games afterwards so no improvement, isn't template specific
Amount of Idle Achievements - idle is a different game format altogether

No data give you 100% certainty. Disregarding information because it can be false leaves you with little to no information to work with.

It depends who you are comparing. MDL is great indicator for the fraction who plays it, 1v1 is next best but not too many play it either. If you have data from the ones above, weights for win rate and games played should be tiny. Problems come when you have to compare two players where one is playing MDL but not doing well and the other has 70% win rate but never played MDL (I would give it to the latter in this case, but its not straightforward). If you only have win rate and games played, I would give more weight on win rate.

Number of wins vs Rufus is best indicator

It only takes one run on ROR to skew anybody's record.