I come back to play on voobly every few months and I jump from 1700 -> 1900 in like 5 games because I gain 40 points vs high rated opponents. Lower the starting ELO, and minimize gains and losses like you said -/+10. That's a decent start.

You guys never switched to the Elo algorithm we are still using in AOC, and AOT still uses the ESO algorithm we implemented. The decay limit has always been set to 1699 and was changed with the ladder reset in 2020 and then raised to 1799.

Also the set values, that you lose / win many points are not the default settings and this could be changed any time with only a few clicks. I actually see the problem here, that this was never discussed with the community.

Thank you guys for all your contributions once again. We really appreciate you taking the time to share your thoughts.

In reply to ChronoJJ: you are making some valid points and I like the solutions you proposed. The point system should be more forgiving then it is right now, but it shouldn't get too forgiving either. I have two remarks however:
1) The idea of making new players win/lose a lot of points isn't inherently bad, because they will arrive at their real rating sooner. Mainly in TG this poses a problem however. If the new players lose/win a lot of points so will their teammates. So the teammates are stuck losing 30 points for a game they never had a chance to win.
2) I most certainly agree that rating isn't the only motivation to kick new players. However, people care a lot about there points and you can't blame them for that. Ratings are there for a reason. Though not all the motivation to kick the new players, it remains to play a part.

In reply to Pemera: I agree, I think that would make the system a lot more fair and stable.

In reply to Chris: I will say it's interesting that there is so much disagreement with the current system while it could've easily been changed years back. Seems like it is about time to do what's right.

As someone who hosted thousands of team games from low level to high level through my main and my , I will try to share my experience about the rating.

1) New player experience: I joined voobly on 2020 iirc, I won my first two games on 1V1 idk how although I am sure it would have been vs 1500 players. But when I joined my first tg I was heavily criticized (if not bullied) for my lack of understanding of the game. I used a 4:00 min no hd build even on high hunt. And on water map I wouldn't build a dock. But from the rating alone I thought 1600's players are as bad as me. I took me literally a month to realize that my rr is probably 1500 and started to play more 1v1s until I learned the basics of the game and then started stopped (although not completly) being kicked from rooms.

2) Hosting a room : Many 1600 joins tgs. I ask them if they are new. If they say yes, I automatically tell them to host a noob lobbies and if they don't quit kick them. I don't kick them because I am afraid to lose points idc about them, but there is two reasons: The game is not interesting and waste of time: The team who has more noobs lose. You can play the game of your life and demolish a good opponent but because you have one or two noobs in your team you find them dead and have to resign despite playing better than every player. This happens also in games with more advanced players but at least they know how to hold better.
If the player is not totally new, I judge from the 1v1 rating. Idc much about tg rating if the 1v1 is in 1400s. I personally think 1v1 is a real estimate of the skill.

3) My solutions:
A) Idealistic solution: Noob players join a noob lobbies where they have to win x number of tgs or 1v1 to be able to join the main lobbies. But I don't think the number of players on aom is enough to contain it.
Alternatively, you could make tgs be unlocked after 5 1v1 games so you have a real judge of the player's ability.
B) Alternative solution: In the spirit of the last one, make special rooms where everyone is welcome and tell new players through prompts that in these rooms they won't be kicked.
C) Alternative solution 2: Make new players start at 1350 and tweak the rating system as said before.

Thank you for sharing your opinion NoAtlantis. AOC actually has a new player lobby, but I'm afraid this will be hard to implement for us since our community is a lot smaller. Our beginners will have no players to play against I'm afraid. The idea is good, but probably not possible for us.

The second solution is completely up to the community. I personally host "All Welcome" rooms 90% of the times I'm hosting. I highly encourage others to do the same so new players would feel more welcome.

Third solution, I agree. Spot on.

As the game has a low pool of players often at dead times of the day (not weekends) will have to go far outside of there rating to play a game without having to wait excessive amount of time, however the rated system does tend to punish this way too much such that an 1900 player who plays a 1800 or 1800 playing 1700 exc.... will win 1point yet lose 30+. Even though the player who is 100 point lower lets say has a 30% chance of winning overall favoring the lower ranked player. This removes a lot of incentive for people to play those lower than them unless they just don't care about the points which makes the point system pointless then anyway. Would like to see the formula fixed to deal with this bit better so people don't just pick opponents within 50rr.

If a change in the rating system makes it harder for smurfs to impact the ladder that'd probably be a good thing. Find a way to discourage people from making new accounts and have them play on their mains is the best way to go.

100% agree with this point, i've seen some turkish player which i dont want to say his name create many accs, get them quickly to 1750 and then feed his main account with them and then just camp on the top10 ladder just like that

That is definitely something we'd like to do with a rating system update as smurfing remains to be a problem in our community. We have also been considering e-mail verification to make registering new accounts harder, but this is something we haven't decided on yet.

Please report violations like this in the appropriate place in the future: https://www.voobly.com/forum/view/568. We take rating manipulation very seriously. I'm pretty sure who you are talking about and I will investigate the manner.

Hi there,

I just had a long conversation with Chris, the same Chris that replied in this thread. He has a lot of experience in managing the rating system for AOC that has proved to provide stable and representative ratings for everyone for years. He gave me some insight in how we can reform the AOT rating system in this conversation. Based on this I wrote this proposal.

We start with changing the maximum rating players win/lose per game. We want to change this so the ratings will be more stable and therefore they will be more accurate. Limiting the amount of points won/lost per game will eliminate the luck factor. We want the ratings to actually mean something, so for that this would be a good change. Limiting the points won/lost per game will make players ratings closer to their "real ratings".

Chris told me the values that AOC currently has: 1-8 points for 1v1, 1-5 for 2v2 and 1-3 for 3v3 and up. For easier reference I will refer to this as 8-5-3. I think this would be a huge step to take from the 1-35 we have right now. For example: a 1700RR player may have just dropped to 1600 in four games and changing it to 8-5-3 right now would result in him needing a lot of games to get back to the 1700 where he belongs, especially for TG players.

I believe we should implement an in-between step because of this. I'd propose 12-12-12 initially and change it to the 8-5-3 the month after. This so players that recently gained or lost a lot of rating with luck or unluck have some time to get back closer to their "real rating". We could also decide that 8-5-3 is too small and we go for 10-7-5 or something different instead. Keep in mind that the smaller the maximum is, the more accurate our ratings will be.

We also want to set a multiplier for the first 10 games, so new players win/lose more points so they get to their "real rating" faster. Think of this as the placement matches that League of Legends has. We can make it so they gain/lose 3 times more points in their first 10 games.

I propose we leave it at these two changes for now. Things like the starting rating and decay rating are not off the table, but we believe the above changes may already fix these problems, so change may not be needed. I'd suggest we revisit those options in half a year after the change.

In short I propose the following:
1) We change the maximum points won/lost per game to 12-12-12 and a month after to 8-5-3.
2) We set a new player multiplier of 3 for the first 10 games a new player plays.
3) We revisit the starting rating and decay rating options in 6 months.

I'd like to hear your opinions on this proposal. It is just my proposal and it is not set in stone. Please let me know what you guys think and if you disagree, tell me what you'd do different and why.

All the best,

Bombus

If you use the same formula but lower the points won/lost, mathematically isn't it the same thing, just lower numbers? Or are you just talking switching over to elo?

ELO sucks for team games. There's no reason players should win/lose more rating in a 2x2 than a 3x3. ELO works well enough for 1x1 (what it was designed for, as I recall), but is a little too forgiving when playing uneven games. (also a reason why we originally switched from elo to eso back in the old day). Top players would just play 200 points down, still win 4~ points, and not bother playing vs close opponents.

I do like a new player multiplier. I assume it would only affect the new player (not others in the game), so kind of like self-rate but the way it was meant to be.

I just don't see the need to completely change the rating system again. There are ways to improve what we currently have, but simply changing to a completely new system (or ELO) won't provide a magical cure for all the problems, and will just introduce other problems. AoC likes ELO because AoC always used ELO. AoT never liked ELO because everyone was originally used to ESO, hence most everybody wanted to change to ESO when Voobly grew into the main AoT community (that is why it was changed, it was based on community feedback and conversation, not just a random idea).

Hi, after reading all of the replies and the original thread.
I think there are some great ideas and suggestions in this thread - Discussing the original ones:

*Change the starting rating for new players.
*Set a maximum points won/lost per game for 1v1, 2v2, 3v3, 4v4, etc.
*Make all teammates in team games win/lose the same amount of points.
*Set a multiplier for rooms containing a new player so less points are won/lost for all players.

1. New players getting a 1300 rating with handsome multipliers (win multipliers but not lose multipliers to encourage them to play more)
2. If we want to create a bigger span on the rating system say: 1500 to 3000 and not just 1800 1900, it would be better to go for a decent number of change in ELO due to win loss. I would say 20-15-13 or something like that. With 1v1s holding a little more importance than others.
3. The current system is great when it comes to the calculation of individual win/loss ELOs during a Team Match. For example a 1700 player having 2 1400 players in team losing to three 1600 players will not lose a lot of ELO. And that is great.

4. Agreed with the 4th point. Also, new accounts should have UNRANKED ratings displayed on their account until they play at least 3 to 5 1v1s in order to gauge their initial skill level. And then mimic the same rating for other game modes on that new account. Say I make a new account and I start with all of my ratings as: UNRATED and once I play a few games as 1v1 I get assigned an actual rating which will be same for TG as well so people can have an idea about me.

Another possibility is that when people are making a new account for AOM - a question should be prompted to them asking:
Are you new to AOT online/voobly or are you an experienced player returning to play the game?

For the ones selecting they are new players can get 13 or 1400 starting ELO
Whereas experienced players creating a new account should get UNRATED ranking until they play 3 to 5 1v1s or TGs until their skill is gauged.
This would completely solve the issue regarding how to identify which is a new player and which is a new account.

In reply to ChronoJJ. Our thought was to keep the ESO system, but to just change the values. There is no need to completely adopt ELO. We don't necessarily need to differentiate between TG and 1v1 like AOC does. We can also set 5-5-5 instead of 8-5-3. The new player multiplier only applies to the new player himself, Chris said.

There is something we need to point out though. We do not have the same players as we had in 2014. Many of our players started playing after the days of ESO and some don't even know what it was. With new players we have different opinions and wishes. So whatever was a good choice back then, may not be a good choice anymore today. For reference, Chris found this thread: https://www.voobly.com/forum/thread/182710

Based on the relatively big amount of replies to this thread compared to other AOT threads, it is clear that our community of today isn't completely happy with the current system. This is what counts for me. The rating system is not about what I want, what Chris wants, what the staff wants, but it is about what the community wants. Since the community isn't satisfied with the rating system right now, maybe it would be a good idea to give something new a try.

By decreasing the amount of points won/lost per game the ratings will be more stable. Players will find their rating linked to their actual skill easier (real rating) and therefore they will be able to find equally skilled opponents easier. This will enhance the experience for new players and noobs that will no longer be bashed by medium players and it will help medium players find each other instead of just playing in noob rooms, not being able to find something better. Understand that these categories make up about 85% of our community and so these are the players we need to serve the most.

========================================

In reply to thesavagehitman. Thank you for contributing as well. I like the idea of the win multiplier, so new players are encouraged to play more. This is however not possible. The multiplier amplifies both wins and losses.

20-15-13 would be a step down from what we have right now, but it will not solve anything. Anything above 10 will result in unstable ratings, so the problem we want to resolve wouldn't be solved.

The question if someone has experience with the game or not is a great idea in itself. You can count on it however that many smurfs will not answer fairly and so abuse this system.

========================================

Question for the community: do we want to differentiate between 1v1 and TG (8-5-3) or would we prefer the point system will work the same for both (5-5-5)?

I guess I still maybe misunderstand the workings behind it, but if it's the same formula but fewer points, then the proportion of rating would still the same. So it'd really be no difference at all, just 1/4 the points available. 1650 to 1700 would be the same as 1650 to 1850 currently. Unless there is something else going on with the formula, but so far it seems a bit confusing exactly what is being proposed.

Most of the discussion for the previous rating change was on RTS Sanctuary I believe.