European Rapid Championship (2017)

I actually meant that players should have no rating after being not active for a while. Let's say 3 years.

<"The RD itself decreases after playing a game, but it will increase slowly over time of inactivity.">

https://en.wikipedia.org/wiki/Glick...

I think another improvement would be a slow decay of inactive player ratings. Might be perhaps better than a fixed cutoff date.

I agree with the sentiment that a rating needs to be defended to be maintained.

I would go even further; to conform to the rating system there should be an award for the player who played more games than another player during period x.

This way things would look more professional. Because extreme difference in performance rate between players over period x could be solved by matter of compensation.

In that case we can speak of time/game frequency of a player. A simple factor to implement.

Also the definition of the 'active player' becomes more transparent. One could apply this formula on recent chess competition statistics in Excel, for example.

To define the correlation between frequency of performance and ELO rating.

With the intention to compensate the more active player and express this as bonus ELO points. And to determine what is the (average) activity factor per player in each ELO segment.

At the end of each year a player has their ELO rating 'corrected.'

Obviously, I have based my observations on other sports. And applied the concept of annual promotion/relegation on ELO rating this way.

But I'm immediately struck by the tension between a rating system accurately reflecting a player's strength, versus their activity.

Clearly an inactive player couldn't beef if their rating slowly decayed away, but a 2800 super-GM who plays 5 games/month vs. one who plays 10 games/month?

Normally the reward for activity, at least for the very best, is the ability to accumulate rating points faster than a less active but equally strong player.

Yes, it's more complicated than I thought; I can't apply compensation on old statistics really, because the players must be aware of a change in system first; to test that system under proper conditions.

It's funny, the rating system has compromises built into it, somewhat related to this issue.

E.g. you want a large RD for young players in order to allow quick rating climbs for new players without a lot of games.

But very active players get decreasing RD's in order to damp down rating changes (the idea being that their ratings are well established and so any given win/loss should count for less).

So I suppose the system's reward is the more you play the less you lose. Trouble is, the more you play the less you win. Rating points, that is.

This suggests maybe building in an RD(-)/RD(+) asymmetry - but that could just be a "bad" idea.

https://www.fide.com/component/cont...