Proposal - degradation of rating over time

I have a proposal regarding the global rating system.

I am playing FAF for the last 3-4 years with periods of stoppages. The longest one was in 2021 for almost 9 months.

Of course when you are out for so long time, your gameplay level is suffering from that. And if you are 1k+ rating, when you came back, you are pushed to play with the similar number, because noobs is not willing to take you with them, and basically you are frustrating yourself, frustrating your teammates, as you are not playing on the rating level what is supposed from you.

My personal example, that my previous stop was in 1.4k rating player, and stopped for 11 months for personal reason, came back in beginning of 2021, and then i was obliged to create new account.

So the general proposal - that for every 1 inactive month you loose 100 rating or 200, whatever is balanced up to the level of 0, so if you are out, you can restart your gameplay.

Something tells me Blackheart isn't playing like a 1200 when he comes back for his yearly game.

meanwhile I found that I play better after a long break (longer than yours) than I did before, maybe instead your rating should be more volatile after not playing, but only in downward direction (easier to lose it, instead of directly losing it, but not easier to gain it, idk if it already even is so, possibly?), still extremely abusable though, such as for equivalent of smurfing

Functionally that is what a rating degradation implementation would look like in trueskill, it's just about making the uncertainty increase large enough to represent being rusty but small enough to not just allow for inflation by people not playing for 1-3 months while still being just as good.

The problem is the rating what is shown near to my nickname, as if i will try after the break join the game of people of 700-1000 people, just to try myself, remember the builds, and then came back to "my" rating - i simply have no chance, so to come back, im obliged to play the games with 1.2k+ and constantly loose, destroying the game for other people, just because i cannot go to my current "real" rating people.

How many games do you expect you have to play until you are roughly around your old skill level again?

@mach said in Proposal - degradation of rating over time:

meanwhile I found that I play better after a long break (longer than yours) than I did before, maybe instead your rating should be more volatile after not playing, but only in downward direction (easier to lose it, instead of directly losing it, but not easier to gain it, idk if it already even is so, possibly?), still extremely abusable though, such as for equivalent of smurfing

this is facts, i play 5x better when i get back from a long break.

Moved to General Discussion

I have now played four 1v1 games after a break of almost a year and won two of them. In my rating graph you can see that I already had another break of roughly a year and I could continue with my rating.

When you watch some casts or some replays to refresh your memory of how to play the game you don't significantly lose your ability to play even after long breaks.

Agreed BlackYps. But there is certainly a difference in my gameplay after a long break. Sometimes I'm much better or worse. Having a larger uncertainty to reflect an extended break would be nice. But having your actual rating change after several months break seems bad.

what about a color code which is apllied to your ingame name in lobby, so the lobby can decide itself if they want to play with someone or not? Maybe from green to dark red. green would show you played regulary in the last time dark red would show you didnt played for a long time. so very tight playing high rank lobbies can decide to kick a dark red 1700 from setons while you can still join casual mid tier gap as the needed 1100 on mid or so ever.

no rank decrease but some kind of maybe-a-little-rusty-stamp

This is literally going back to the whole gray rating problem that made it impossible for new players to get games

hmm indeed, but your match count wont go away ofc. but you are right every good player who just plays once or twice a week would be punished.

overall i personally have no problem with overrated rusty players in coustom teamgames due to coustom rank has the least explanatory power. in tmm everbody should decrease one subdivision by default every month i think (?, dont know the word i mean you decrease eg from silver II to silver I)

the only ranking where a steady rank decrease over time is debatable is 1v1 imho.

Or everbody regardless if playing or not gets a (small or even smaller) decrease in every rank. this will counter rank inflation and its fair for everybody. it could be just a small downwards tendency which maybe isnt even recognized by the single player but measurable in the overall statistics. you counter the decrease by just playing. Maybe 1 point a day, would be up to debate and checked by people who acutally got a clue about numbers.

There is no agreeable solution for rating degradation. Every solution that has been proposed so far has significant side effects that will make the situation worse than before in many cases. If the rating decrease is small enough it doesn't really matter anymore and you might as well just leave it out completely.
The division system provides a soft reset as you will become unlisted when a new season starts and you also will be placed in a slightly lower division when you play again. We could show the division instead of rating in custom lobbies, but every time this gets proposed it causes a massive outrage by the custom game players.

@blackyps said in Proposal - degradation of rating over time:

The division system provides a soft reset as you will become unlisted when a new season starts and you also will be placed in a slightly lower division when you play again.

How is this more than a cosmetic change? Is someone's rating changed when they become unlisted, or do you balance unlisted players as though their rating is lower, or what?

Also, relatively small changes can still be worth doing. I think the pros outweigh the cons for slowly increasing sigma over time. It could be a slow increase (ie 0.2 sigma/day) with a low max (ie: it won't raise sigma above 150). Here's a relevant thread: https://forum.faforever.com/topic/2767/rating-sigma-should-increase-over-time?_=1677797840656

It is just a cosmetic change but it seems that for some people the cosmetics are the issue. I.e. they feel pressured because they see their old rating and think they must perform on this level.
Small changes don't matter because the range you can get matched with in 1v1 is about 200 rating. If I as a 1k queue up it depends on who else is in the queue. If a 1150 joins I will get a hard game, if a 850 joins I will get an easy game. It makes no difference if some rating degradation places me 50 rating lower on paper.

A sigma increase is definitely bad because now your first games after a break have an unusually high impact on your rating. Let's say you do lose your first four games in a row. You now lost even more rating than normally and you now need more games to climb back even though your skill has returned rather quickly

@blackyps said in Proposal - degradation of rating over time:

you also will be placed in a slightly lower division when you play again

It is just a cosmetic change

How can both of these be true? Can two people with the same rating be in different divisions?

Small changes don't matter because the range you can get matched with in 1v1 is about 200 rating. If I as a 1k queue up it depends on who else is in the queue. If a 1150 joins I will get a hard game, if a 850 joins I will get an easy game. It makes no difference if some rating degradation places me 50 rating lower on paper.

Small changes do matter. If a person returns after not playing for a long time and is better or worse than they used to be, their rating can adjust to match their current skill level more quickly with an initially higher sigma. So, if a player had gotten worse at the game after not playing for a long time, this would help them to start with a lower displayed rating iniitally and get to a more accurate rating more quickly without having to go through the frustrating process of losing numerous games in a row. In the case of the matchmaker, a rating difference can still have an impact. I'm not saying it always would, but it's not like it never would either. If, for example, a potential opponent would be in the upper end of the range that a player would match with, then a reduction in the player's rating could prevent that match.

A sigma increase is definitely bad because now your first games after a break have an unusually high impact on your rating. Let's say you do lose your first four games in a row. You now lost even more rating than normally and you now need more games to climb back even though your skill has returned rather quickly

Let's not use false certainty. You personally think a sigma increase would be bad. That does not make it definitely bad. I didn't claim it was perfect, but there are clearly both pros and cons, and I believe the pros outweigh the cons. If a person's skill has decreased to the point that they lose 4 games in a row upon returning, I think their rating dropping a bit as a result seems quite reasonable, and it's not like their sigma would just insantly drop to 0 after that. If you think sigma decreases too quickly at like 150 sigma to be accurate, then perhaps some settings should be changed regarding that. However, sigma is basically just rating uncertainty divided by 3. After a person doesn't play for a long time, there is more uncertainty regarding their skill level than when they were playing regularly. Having a sigma increase to match that actual uncertainty makes sense to me.

Let's not force the people who do get a lot worse after not playing for a while to lose so many games in a row to get their ratings to be more accurate again. That is a hindrance for players returning to FAF after not playing for a while, and it results in additional players not returning.

A sigma increase over time would also help with some of the people who manage to get overrated and then don't play many/any rated games afterwards because they don't want to lose that high rating, since the sigma increase would result in their displayed rating decreasing over time as a result of inactivity. This would apply to both custom and matchmaker players who reach a peak rating and just hold it indefinitely without having as much incentive to keep playing as they would if sigma increased over time (a lot of players care about rating rather than or in addition to their listed vs unlisted division status).

Two people of the same rating can be in different divisions, that’s the point.

the idea is not very