New Player ELO of 1000 is Too High

The ELO of above 1000 given to a First time new player is too high in my opinion. As a result the ~1000 Elo matchups are messed up. New players are unable to match with the pace and game (especially TG) suffers.

I am 900 Elo in TG now after playing 400 games. I started with 1000, went down to 500 and back to 900 after playing 400 games. Almost half of my TG games suffer only because there is atleast one new player in the game who is 1000 elo just because he is starting up but actually plays like a 400 Elo.

Devs are requested to REDUCE THE STARTING ELO TO 400. If players are good, they can climb up quickly. It is better to put a new player in lower tier than a bad player in higher tier. This will reduce the mis matchups.

DO YOU GUYS THINK THIS NEEDS TO BE CORRECTED?

1 Like

please read up on how the elo system works before posting stuff like that. It’s just uninformed

oversimpliefied, the elo system works like this: the average rating is 1000. so if you have a rating of 1000 you have a 50/50 chance of winning against a random opponent.
if you start lower the average rating to eg 900 by giving lower rating to new players, it would mean that people at a 1000 rating would now be above average according to the rating. but since their skill hasnt changed they would start losing more games until they are at the new average rating.

if all new players were to start with significantly lower elo that would lead to overall elo deflation (ie everyone’s elo would drop)

a large influx of new players (like there might be with the xbox release) might lead to some change in ratings though…we will have to see how this plays out

5 Likes

I am sorry I dont get this.

You are saying that every new player starts at the average rating !!
This is by far lopsided as it is. Most new players playing multiplayer are below average. So based on the elo logic if below average players start above their ratings, it will inflate everones ratings. This is not ideal either.

But the whole point of inflated or deflated ratings is not as important as getting good games and this system is not doing it, specially in team games. One bad player in your 4v4 team will mean you loose, and stop u from advancing further in rating. And at 1000 elo you get one almost every game.

I am a chess.com player as well and they follow the elo system too. There the new player has to choose in the beginning whether he is a beginner (600 elo), Intermediate (1200 elo) or expert (2000 elo). I think this would make much more sense than to just give Average rating to all New Players.

2 Likes

Yes I agree. Let a new player choose whether they are beginner (600), inetermediate (800) or pro (1200)

2 Likes

the average rating of new players needs to be the average rating of existing players, otherwise it leads to inflation/deflation over time

deflated or inflated ratings lead to bad games if players take a break. currently the matchmaking system is amazing, especially in 1v1. i don’t play teamgames much

this is very different since most players play chess elsewhere before
in aoe this would lead to huge smurfing issues

2 Likes

Because of how elo is zero sum matchmaking, if player 1 earns X points for a win, player 2 will lose X points for that loss. Consequently, no points are ever changed in the overall community average, which is 1000. If you move the starting elo lower, then the average will just recenter around the new starting point, that’s how the system is meant to work.

3 Likes

Elo is NOT a zero sum game. Because new players are being added all the time. Also if as per ELO formulae, if there is a big mismatch in opponent ratings, the points gained by winner and points lost by looser are not the same.

Ideally the average rating of new player needs to be as close as possible to his actual skill relative to the pool of players. 90% of the times the actual skill level is much lower than the average elo rating. This leads to inflation.

Smurfing is an issue, but I think having bad games due to new players is a much more common and recurring problem faced in Team Games right now which needs to be addressed.

The total average of all players will always be 1000. There is never a gain or loss of 1v1 elo outside of that. Whenever someone gains elo, the player who lost will lose the same amount. Always. Adding a new person at 1000 elo doesn’t change that.

2 Likes

And one bad player in enemy’s team will mean you win, so you should advance further in rating. Do you think you’re the only one getting those fresh players in your team? How do you know the other player is a fresh player? There are many reasons why some players are bad and their ratings are as they are.

5 Likes

Thats incorrect. Inflation is not caused by players coming into the system, but by players filtering out: If someone starts playing, looses 5 times, and then deletes his account, he gave about 75 points into the system (or a bit more, i think sensitivity is increased for new accounts). Note that if he stays active it does not cause inflation, because he will have slightly increased the average elo of all other players, but will himself drag the average down. Only when he stops playing the average goes up above 1k.

This is why elo inflation can happen, but its also why its pretty slow. You need a LOT of new player loosing a lot of games and then deleting their accounts for it to matter, exspecially if you factor in there are also some deflationary tendencies (imagine hera stops playing, thats 1800 points deleted from the system; or if a smurf account is made, stomps 30 games until he reaches his true elo and then stops playing).

And since inflation is so glacial, it doesnt matter that much. I can stop playing for a few months, and after 3-5 matches be at my actual elo once again.
However, if we reduce the starting elo to lets say 500, and influx of new player would instantly drag down the average elo by a few hundered points. And players who took a break will be screwed. If you are 1800 and take a break, and during that break elo deflates, the new 1800 is the old 2000 - meaning you play vs players you can hardly beat for the first 10-20 matches.

1 Like

I almost have a separate issue. I wish, regardless of what ELO you start out at, that your first 5-10 games did not count against your win/loss record as you settle into your ELO. I started multi-player with about 8 losses, before I settled into the expected 50/50 win rate. Took me 2 years, 750 games and almost a 200 ELO increase to get my Win/Loss record to 50%. Frustrating trying to recover from those first “placement games”.

1 Like

What will happen if we drop the starting elo to 600? The average rating will slowly move towards 600 and and average player will be 600. A new player will then face the same issues as suggested in this thread. Lowering the starting is just not an option.

If it is just your first 5-10 games, then it really will be fixed within 100 games. I am pretty sure everyone (except the bottom and top) will converge to a 50% win rate pretty quickly. Most of the players will be between a 40-60% winrate. If you are more close to 1000 elo, it might even be within 45-55% winrate after 100 games. Even including the first 5-10 games.

Those 5-10 games doesnt really make any difference after playing 200 games. You even have played 750 games. Then these starting 5-10 really dont matter anymore to your winrate. =

But it does matter. If you lose 5-10 games, and then you start winning at the expected 50% rate, your overall win rate with show 49% forever, unless you improve significantly, This is because unless you start consistently winning above 50%, you never recover from the first placement games. Not saying you won’t win 50% of your games, saying your overall win rate is hard to get back to 50%. If you lose your first 5 games, and then play 100,000 games at a 50% win rate, your overall win rate will show 49%. Again, it took me 2 years and a 200 point ELO increase over 750 games for my overall win rate to finally show 50%.

What would be the harm in saying your first 5-10 games, count toward ELO, but not toward wins and losses?

What would be the gain? That just fakes statistics for the very few people who start with a true elo below 1k AND care about that percentage.

Just write down your wins and calculate the number for your more recent games yourself if that number means a lot too you.

Based on the concept of ELO, my guess is 50% start below 1000, not “very few”,

In the same vein if your ELO settles at 1200 after winning your first 5-10 placement games, then it would be very difficult to ever drop below 50% unless you start purposely tanking. (You have a 5-10 game buffer)

My guess is you started above 1000 and so seeing 50% didn’t matter to you. Be the guy stuck at 49% forever, might change your opinion…a little.

Just suggesting the first handful of games be treated like the preseason in sports. Then your overall win rate would be a little better reflection of your wins and losses against opponents of similar ELO.

It’s nothing major, and finally hitting 50% felt pretty good.

Read again - I wrote about players who are below 1k in skill AND CARE about their winrate. I think youd be hard pressed to find many. Im not saying it would somehow be a bad thing, just that a change that most players dont care about and those who do care can just calculate on their own seems pretty needless.

And having 50% to start with would have robbed you of that pleasure :slight_smile:

i think a meaningful piece of data would be ‘winrate in last X games’, where X = 10, 50 and 100?

2 Likes

LOL, that’s true.

I guess that’s better than having a better than 50/50 win rate, that you know wasn’t earned against similar ELO,

I dont see any issu with a winrate of 49%. Not everyone will be exactly at 50%. I consider every between 45-55% as about 50%. If you have played less then 100 games it might be 40-60%. Numbers in that range will be fine.