Gradually declining ranking / Reassessment

[Bismarx]

Hello,

i have a concern with the ranking system. Until 1.5 years ago, i was still a student and had lots of time to play FAF. In that time i build up a ranking based mostly on playing Setons which was something 1300-1400ish. Now, this was ok at the time. But it isnt now.

I have recently come back (as usual) to playing a couple of games and will continue to do so in the future. I am still ranked 1300ish or something. But i suck. First of all, clearly i have forgotten some of the things that used to make me an OK-player on Setons. Secondly i think the general level of play has further improved over the last two years. Now i could just play more, but for some changes in private life etc. and work-related i have much less time to spend on FAF.

This would not be a problem per se, but i continue to lose team-games for my team for not being able to play up to the lvl that FAF attributes to me ranking-wise. And with maybe three games a week and a few hundred games on record this will remain this way for a long time until my rating seriously goes down. And especially Setons-games can be long endeavours, the balance a delicate thing and it is all the more annoying to 7 other players if the game is slowly but predictably lost just 1 player playing out of his league.

So the easy fix would be to just create a new account and play as a smurf. But oviously this is unfair to anyone else and also spoils a ton of games for the rest until the rating-sytem reflects the level of play somewhat. Why not make the rating decline over time with inactivity? Or maybe make a reassessment based on the old ranking and maybe 5 games every half year? The latter thing is done by e.g. Starcraft 2 every season and helps to put players where they belong.

I have not found this on the general forums, so i will just put this here. Kudos if this has already been discussed elsewhere.

Bismarx

[Mr-Smith]

If you dont play ladder for a while you get removed from the leaderboard anyway, to get back in you ahve to play a few games than, this already corrects your rating kinda, the rating system is fine, if you think you are overrated, host a few 1v1,i guess after 10-20 games your rating should be accurate again.

[Bismarx]

host a few 1v1,i guess after 10-20 games your rating should be accurate again

Yes, i might actually do that. However a 1on1 is a totally different thing from a Setons 4on4. So to determine, "if my rating is accurate", is going to be entirely up to me then, as i will predictably lose most 1on1-games. Would it not be better to just make the rating decline by a percentage/week after 4 weeks of inactivity maybe? And then use this rating, counting as maybe 20 games and readjust the rating based on the games you then play?

So the rating could adjust somewhat flexibly without the need to play other game-modes.

[Mr-Smith]

Well losing games on purpose is against the rules

You will not lose on purpose to lower your rating, your rating will be reset in addition to any other sanctions that may apply.

[Bismarx]

I did not say i would lose on purpose. I said it would be likely. A 1300-rating based on Setons 4on4 is as strong an indicator for winning an 1on1-game as playing Diamond-league 4on4 is an indicator of being good at Starcraft 2 in 1on1. So none.

I expect to get my nose bloodied by anybody playing over 600-700.

[Luminareo]

If you dont play ladder for a while you get removed from the leaderboard anyway, to get back in you ahve to play a few games than, this already corrects your rating kinda, the rating system is fine, if you think you are overrated, host a few 1v1,i guess after 10-20 games your rating should be accurate again.

For people who care more about accurate matchmaking than leaderboard ranking, this system could improve. These "rating dump" games are part of the problem. The points have to go somewhere, so someone ends up being overrated. Some dude dumped like 100 pts on me in a custom Theta game that will take a dozen team games to burn off. Meanwhile I'm letting my team down.

There's a bona fide plague of people who come back after some years and ask for special balancing consideration. If nothing else, rating decay would make this transition easier.

[Axle]

I can throw in a point or two here:

- FAF's trueskill rating are not optimal [ viewtopic.php?f=45&t=11698 ]. The optimal parameters are much more reactive to changes in player skill
- I have played with the idea of putting rating decay into trueskill and it maybe does help predictive power a little. I haven't uploaded results anywhere yet but it look like players might lose on average about 1 rating point per 24hrs of inactivity.