Rating Decay
-
There's been a lot of back and forth over the years about whether FAF should implement a rating decay. The core issues are A. that quite a few of the top rated players haven't played in a long time, and B.
players who return after years away are often stuck stuck playing in lobbies beyond their skill level until their rating drops back down. It's a semi-frequent request for the moderation team to reset rating or allow returning players to make a new account (we don't do either).On the other hand, rating decay could act as a kind of soft smurfing where someone doesn't play for a while and then gets to punish lower rated players. It could also make high rated players feel forced to keep playing in order to maintain their rating.
So I propose an idea I'm adapting from another game I play, Prismata. They have something called a Vacation Penalty (VP), which makes your rating temporarily lower depending on how much you've played recently. Their VP uses the formula
- Rating - 15 • (15 - # of games played in the last 15 days)
This is just to explain the idea, obviously we'd need to adjust the numbers for FAF and make it work over a longer period of time. But this way you get to play against lower rated people on your return, but your rating will also return over time more rapidly than a true decay.
-
On the other hand, rating decay could act as a kind of soft smurfing where someone doesn't play for a while and then gets to punish lower rated players.
This contradicts the very reason anyone would advocate for rating decay: returning players lost skill themselves. They're not going to 'stomp' players below their rating anymore, so I don't think this is a good argument against rating decay.
I'm not sure how the vacation penalty would pan out, but the 15-game/15-day limit and the name itself imply a different concept than rating decay, and changing the time limit to several years would functionally be the same as implementing regular rating decay with increasing variance over time.
In any case, I'm for any form of rating decay, and I propose the following reference points:
- No decay for atleast a few months of inactivity. You don't quickly forget how to play FAF. I was gone for a few months and only needed a couple games to get back in form, once at 1300 and once at 2400.
- After 2-3 years, rating decay of 200-300 sounds reasonable
- After 10 years, a decay of ~800 sounds reasonable
-
-
What i personally like is the way they do it on some chess sites.
I don't know the formula, but the way it works is that for every x months you don't play you gain/lose x more rating per game played, and it depreciates.
For example, I haven't played any puzzles for almost 2 years:

And this happens when i play some puzzles now:



-
@KnownSniper that stilk has the issue of you coming back as way higher rated than what you actually are now.
You still have to "grind through some loses" to get to your actual rating.
Compared to if it actually decayed you already start there -
Yes but thats only if your skill decayed as well. Id say this is a more balanced solution. The only downside i can think of however is that u win a few games after not having played for a year and u gain a bazillion rating.
Not sure if there is a difference between the gain and loss multipliers. Ill give an update in 2 years when my puzzle rating can be used for testing again kappa
-
I mean, multipliers had to be worked out and same goes for a floor level, may it be 60%, 70% of the rating when last logged in or whatever. But it's a good idea in general, I think.
Hello! It looks like you're interested in this conversation, but you don't have an account yet.
Getting fed up of having to scroll through the same posts each visit? When you register for an account, you'll always come back to exactly where you were before, and choose to be notified of new replies (either via email, or push notification). You'll also be able to save bookmarks and upvote posts to show your appreciation to other community members.
With your input, this post could be even better 💗
Register Login