It was not obvious when first introduced, but the Elo decay system has created a brand-new problem.
Especially, it’s noticeable with the recent explosion of low-tier community tournaments.
Before explaining, I want to point out that the Elo decay system doesn’t belong to the AoE 2 game or even the whole genre. It should be similar to chess rather than online games with leagues/divisions or similar. It makes perfect sense for 1v1. At most, it should only happen to the top 100 or 2K+ players (just for the sake of players bragging about their rank).
The ideal solution would be to just flag inactive players or don’t do anything at all.
Now, about why decay is problematic. Not all players constantly play ranked games, a lot of them are playing mostly in non-ranked games of different kinds. But to maintain their Elo displayed, players are forced to periodically play ranked. While this is not a problem in itself and players don’t give much thought about it, the system has been weaponized - it allowed competitive players to reduce their Elo without losing. There are players with a skill of 1800+ but their rating is 1600. This allows such dishonest players to get into lower brackets in tournaments. The more dishonest players appear, the more incentive there is for others to do the same to stay among their actual skill level. This creates a mess of players occupying the same Elo range, and nobody knows if they are going to face a smurf, sandbagger, or an honest player.
The bigger issue is that smurfs can be found, but Elo manipulators are basically invisible.
And also, even honest but inactive players can accidentally fall into this if they are not trying to maintain their maximum Elo after inactivity. Good luck enforcing rules about this in any low/mid tournament.
P.S. I think I can statistically prove the existence of the problem in the game. I may look into this - it should be possible to compare the expected and actual Elo bell curve to prove the point.
I dont really think this is one of the biggest flaws in the system. Elo decay will happen slowly. So you really have to not play for a while on a specific ladder. And even then the decay isnt really that much.
I would much rather focus on improving the rating system in general instead on focussing on something minor like this. If it is for tournaments: Then can base the allowed players not only on their current Elo, but also the max Elo. That will already solve some of the lower elo tournament issues.
Just to be clear: I do see how this issue can exist. But smurfing and Elo manipulation are a bigger threat to the game then this.
But smurfing and Elo manipulation are a bigger threat to the game then this.
Smurfing can be found in the DB or analyzing recs, but abusing Elo decay is much less obvious. And it’s a form of Elo manipulation, so it’s a threat.
Then can base the allowed players not only on their current Elo, but also the max Elo.
Yes, most of the times it’s average with 0.5 or 0.75 weight towards the high elo. But the manipulation still puts player in better positions. In this case and, I would strongly advocate for this, tournaments should use only the highest achieved rating.
I would much rather focus on improving the rating system in general instead on focussing on something minor like this
How? Rating is the Elo system. And it should be like in chess.
I am not saying that the decay isnt a threat. I only state that decay isnt the biggest threat. Smurfing and Elo manipulation are much more easy to detect, but they are mostly left unpunished at the moment. That should be the main focus at this moment. That impact of the decay system is pretty minor compared to smurfing and Elo manipulation.
The Elo system is a rating system. It is indeed the same as used in chess. It is pretty oldskool and outdated. Elo is meant to designed for 1v1s only. For chess this is fine, but for AoE2 it is not, since you also have team games. You also have different type of maps. Some people are great at open maps and suck at closed maps. Some are average over all maps. Elo can’t really capture this at all. Thing like these can be solved by never rating systems, like Glicko or TrueSkill. These are better suited to capture team games as well and are also be able to deal with different kind of maps too.
Glicko or Glicko2 wouldn’t solve anything, because if you can’t reduce a player’s strength to a single number I don’t understand why Glicko, which is Elo + seasonal variation, should solve it.
Like TrueSkill, or even better TrueSkill2, which are systems designed to calculate the strength of a single individual in team games. There are people who are strong 1v1, and absolutely incapable of communicating in TG, and vice versa. Two separate values are objectively the best solution.
That part of the reply was more based on a question, than really on what is the topic of this thread. There is much wrong in my opinion with the current Elo system, so i want improvements in general. That part is not focused on the decay, smurfing or Elo manipulation. Such a thing just need more regulation from the dev team.
There is clearly a correlation between the two. It is not like that the best 1v1 player will be somewhere in the lowest 5% of team game players. Ofc there is no 1 to 1 match between the two, but they are pretty correlated.
The third graph shows the correlation. There is clearly an upwards trends.
If you are worried about elo manipulation, Glicko will make things worse, by a lot. Glicko tries to keep track of the variability of a person’s Elo, to make it simple, the less you play the more elo you lose/gain from a game (obviously the opposite applies). Because the less you play, the less accurate your Elo is, so you have to make more drastic changes.
The problem with Age of Empires 2 is that there is no control. It’s not so much the rating system.
Of course there is a correlation, but it is not a 1:1 ratio, there will be someone who will obviously be penalised.
Decay is part of the rating system, so if you want to do something about that part, then you have to change the rating system. That is why i bring up the rating system. If the devs look into this, then i hope they dont just focus on a single issue, but just revamp the full system to something more modern. Microsoft owns TrueSkill and TrueSkill2. So i would suggest using one of them. Glicko was just another example.
I dont ask for the same rating for 1v1 and TGs. But there rating can be somehow correlated as well. Currently TG elo suchs. If the system used for 1v1s and TGs is correlated, then it already should be better. TG Elo currently is pretty meaningless. I see something like that as a bigger issue then the rating decay issue. That is way i ask for a full rework of the rating system, and not just looking at the decay only. In the end no system is perfect, but Elo is clearly not the best system.
Also I think decay starts at 15 or 30 points per 14 days at 1600ELO and scale when you go up - so its really slow at lower elos or even mid elos (I had -50/14 days at 2k2 level)
TG rating is used for TG ranked games. Yeah, that has a different distribution, that is what you see. Your TG rating pretty much not only depends on your own skill, but also on all your allies, adding much more noise. Then you have the team play between the players as an added thing. So yeah, there are difference, you wont get a perfect fit. But there is clearly a correlation.
In my experiences with playing at randoms in TGs, their 1v1 skill is a much better indicator for their skill level compared to their TG rating, which is used in match making. That just shows that TG rating is terrible at the moment. And that isnt due to smurfing, Elo manipulation or decay. I see this as a bigger issue compared to your decay issue. I want a better rating system in general, which solves multiple issues.