Should the online starting ELO be 500? (Making new player experience better)

k-factor is a multiplier:
if you have a k-factor of 20 you win and lose twice as many points as you would with a k-factor of 10

wikipedia:
The formula for updating that player’s rating is
R A ′ = R A - K ⋅ ( S A − E A ) . {isplaystyle R_{athsf {A}}'=R_{athsf {A}}+Kdot (S_{athsf {A}}-E_{athsf {A}})~.}[1]

K is literally multiplied with (actual result - expected result).

it’s still a 0-sum game, as long as both players have the same k-factor. which would be the case in most aoe2 games, where you mostly play against people with a very similar rating. You might have been a chess player, but evidently you didn’t bother to understand the rating system

but that’s not true.

From https://handbook.fide.com/chapter/B022022

8.2.1 If an unrated player scores zero in their first event this score is disregarded. Otherwise, their rating is calculated using all their results as in 7.1.4.

and

7.1.4 A rating for a player new to the list shall be published when it is based on 5 games against rated opponents. This need not be met in one tournament. Results from other tournaments played within consecutive rating periods of not more than 26 months are pooled to obtain the initial rating. The rating must be at least 1000.

so starting elo has to be at least 1000. but can be hypothetically any value.

Please don’t just make shit up

2 Likes

One more important difference between FIDE and aoe2 rating ist inclusivity: FIDE only rates you if you are good enough (ie currently above 1000), if you fall below that threshhold you become ‘unrated’. Of course this skews the average rating way higher, as all the ‘low elo legends’ get ignored in the average

This would not work for aoe2 since we need to rank all players, to get fair matchmaking even in the lower elos.

1 Like

k-factor means that’s the maximum amount of points you can win/lose from a game.

Rating changes can be calculated manually by using the FIDE ratings change calculator.[16] All top players have a K-factor of 10, which means that the maximum ratings change from a single game is a little less than 10 points.

The foruma you quote isn’t on wikipedia.

So you have a hard limit on how much you can win/lose per game, no matter who you play with.

you start from a hypothetical 1000, and only after the calibration games you ELO is revealed, but you still have elo before that.

Another important note is that FIDE used to lower the starting elo specifically to keep the new players’ elo above the average.

Never said we need to make those under 1000 unranked.

Hera discussed this topic in his latest video and he agreed that the starting ELO should be 500. About 16:00 minutes into the video:

1 Like

Then he doesn’t know how the Elo system works
The Elo system is a metric that measure every single player against all the players
TBH, the 1v1 ladder is the best thing that the game has but I understand that is difficult for new players
Although, after 10 games you will have close to 50/50 games won at your right Elo

6 Likes

Aoe2 is not chess, so why do people keep using chess arguments? It makes zero sense to start noobs off at a ridiculously hard level. It should start easy and get progressively harder as you improve, that’s how all games should work. Maybe have a system similar to Call of Duty where the early ranked games are against bots on standard ranked maps and settings, then once you beat opponents of progressively harder difficulty, it starts matching you with real players seamlessly. This would also deter experienced smurfs with multiple accounts.

We should all care about recruiting new players, because they are essential for the future health of the game. We shouldn’t be scaring them away or dismissing their concerns. A significant number of them are getting put off the game before they’ve even started, which is not a good thing and we shouldn’t just ignore it or blame it on the new players’ lack of skill.

I’m not directing this at you in particular, but I feel the aoe2 community as a whole has too much of a gatekeeping attitude. Not enough thought is given to welcoming new players to ranked. So it’s not surprising that the majority stay stuck playing single player. As the player base of the game gets older, some eventually leave because lifestyles change. Some people settle down, have families, etc. Or just move onto other games. Some may even have health problems or worse. If we want the aoe2 scene to survive and thrive, we shouldn’t make it difficult for new players (especially young players) to get into multiplayer. It seems to be a common theme for a lot of noobs to have a very bad, disheartening first few experiences on the ladder and just quit before they’ve even given it a chance. You all seem to think this is ok, like “Oh well, this game is not for everyone” but should we really be ignoring the complaints of people who join this community and support this game we love?

The learning curve should be gradual and there absolutely must be a way to incorporate that into the ranked system for the sake of the future of the game.

3 Likes

A 0-sum game is defined by the total “loss” always being equal to the total “gain”, resulting in the net total always being the same. As I recall, in AoE2, the Elo gain is always equal to the ELO lost by your opponent(s). If you lose 100 points of ELO, then your opponent gains 100 points of ELO. If you win 5 points of ELO, then your opponent lost 5 points of ELO. Either way, the total number of ELO is the same after the game as it was before the game. Therefore, ELO in AoE2 is 0-sum. The only exception to this is if a player leaves the game, but if you include those players, the average ELO would (or rather, should, since bugs may exist) remain at 1000.

I don’t know how ELO works in Chess (and it isn’t inherent to Chess anyways, so different Chess apps may use different systems). I do recall having to select a skill-level estimate (as in “beginner”, “intermediate”, “advanced”, etc. - I don’t remember the exact categories) when I started playing Chess, which implied to me that the starting ELO wasn’t a single constant. But if you ignore the ELO (which once again, isn’t inherent to Chess) and focus on number of wins/losses/draws (with wins being worth 1 point, losses -1 point, and draws 0 points), Chess is 0-sum. The number of losses will always equal the number of wins, and draws have no effect on the sum. It might be possible for an associated ELO system to not be 0-sum, but that would be Chess-app specific.

In AoE2 ranked, the number of players who win is always identical to the number of players who lose. If there was ranked FFA, then this wouldn’t be the case, but the game could remain 0-sum by weighting wins and losses. For example, if the winner won 70 points and all 7 losers lost 10 points each, then the ELO would be 0-sum (since +70 == 7 * -10). If the game was weighted differently such that the winner gained 50 points, the 2nd and 3rd place losers gain/lose 0 points, and the other 5 players lose 10 points, then the game would remain 0-sum (assuming no draws).If in the last situation, it was possible for 2 players to tie for third (with both gaining/losing nothing), the ELO would not be 0-sum because there would be a way for the total number of ELO points to change (as in this example, 1 player gains 50 points, 3 gain 0 points, and 4 lose 10 points, for a net change of +10 points)

Long-term, I think this would only be helpful if newcomers actually got to chose their starting ELO (or at least an estimated skill level like “new to RTS”, “intermediate”, or “advanced RTS player”). Otherwise, the average will start to shift towards 500 as new players come in. And as the average ELO shifts lower, players will find themselves with lower ELOs. Allowing players to give a rough skill estimate to determine their starting ELO would be more friendly to new players as they would be less prone to losing lots of games when they first start playing (if they give a reasonably good skill estimate). It would detach the average ELO from 1000, but that wouldn’t really be a problem for the game (just would make statistics harder to calculate)

2 Likes

I respect Hera’s opinion on the game and his skill but his take about ELO is off the mark.

Dropping the ELO to 500 does not fix it anything. It just makes the new average 500. 1000 ELO is literally as close to average. If you float between 900-1100 ELO you are an average player. If you go a bit lower you’re below average a bit higher above average.

The advice I give to all new players is try to beat the extreme AI without cheesing. If you can do that your theoretical ELO is much closer to 1000. Than it would be otherwise.

Low skill players are always going go on long losing streaks in competitive games ESPECIALLY solo ladders. Even if you remove ELO from the picture players will lose a good number of games.

You can’t compare ELO system to League of Legends for example where it’s teams competing. You can get carried. You’re loss streak will be softened by a mixture of RNG (Good team comp, somebody getting fed, etc) and getting just carried in the odd game because there is an equally as bad player on the other team.

If this discussion was around team ELO for AoE2 some points would stand. We’re talking the 1 vs 1 ladder. Loss is part of the experience. You need to get over it.

If somebody jumps straight onto a ladder especially of an older game with established meta and skills and loses 10-20 games in a row it’s mostly their fault for not learning the basics. It’s trying to run in a race without the ability to sprint or the endurance to do a marathon. You need to drill some basics and build up a foundation. That way if you do lose you can shift some of the blame to “I need to try harder and learn more”.

I took a break from the game, I did not jump straight back onto the ladder I went and beat up the extreme AI a few times to get my mechanics straightened out. I’d know I’d otherwise lose due to rust and poor timings. I still lost a bunch but I did not plummet down in rank.

Rant over.

5 Likes

Admittedly AoE2 ranked is one of the toughest ladders even for beginners, which is the major reason why new players come and go so quickly. Not many people can handle consistently losing streak. I assume most of people who are already here have played this game or RTS in general no less than 10 yrs, but we can’t just have the community full only of old-schools, we need new players.
Also, playing single-players won’t play a decisive role in this game’s longevity, not in this decade. That’s why I said in other threads I DON’T need more civs or campaigns. Players nowadays - they don’t care about single-players, they want to learn fast and rush into competitive matches, and games like LoL gives 'em that. As for Microsoft, I don’t know if they will want to do something about this.

100% agreed with you.
The game should be welcoming to new players (even to old players that want to try MP).
Ask for a couple games against the AI before try MP is a good idea.
I’m just stating that changing the Elo to 500 won’t do anything for new players in the long-term.
The game is hard enough already:
Hidden bonus don’t help either.
The player base is not anything close to chess, SC or LoL.
All things considered are a good arguments to change the perspective of how the game approaches new players.
But 1v1 Elo system is fine in the game as it is right now. No matter your Elo, the system grants you 50/50 chances of win your next game against someone at the same level as you. That’s fairness and it’s what you want in a game like this.

1 Like

You don’t have to beat the extreme AI to start playing MP
There are players at 300 Elo that I don’t think can’t beat the standard AI in some cases and have more games played than some average players (around 1000 Elo)
MP is about to play games against someone as skill as you, that’s why we need a system, in this particular case the Elo system which I think is great

I’m starting to think that maybe the focus should be on marketing the multiplayer game and showing that at the highest level, it’s arguably the hardest, most challenging game in the world, but that the competitive scene is active all the way down to beginners’ level. I think T90’s community games and low ELO legends etc. is definitely good for noobs.

I also liked the idea someone suggested of selecting your skill level when you join the ranked ladder. E.g. Complete beginner (never played) all the way up to Experienced (understand build orders/macro, unit counters, map control, strategy and micro). Maybe a ‘number of games played’ slider as well.

1 Like

Maybe a stupid idea but maybe players should be given a starting ELO depending on what AI they can beat.
You get matched against AI starting at hard a few times (on different maps and random AI civ) and then get an ELO value according to that.
You only get 1000 ELO if you win all the matches.

5 Likes

did you actually use this calculator?? enter two random ratings, try k-factors of 40, 20 and 10. and you will see that the rating changes will be at a ratio of 4:2:1. it’s called k-factor for ###### ##### factor is a part of a product, a multiplier!!

you are accusing me of lying here?? are you blind:
image

so far I thought you were just misinformed, but now you are actively lying. screenshot from wikipedia

I provided a source from FIDE, where they state how they determine initial elo. you are not providing a source and are just making shit up. this is arguing in bad faith at best, but more likely lying because you are too petty to admit that you are wrong.
In the same document they also state that if you drop below 1000 you are again treated like an unrated player.

2 Likes

I think it is worth considering how long this will take. For sake of argument let’s assume that there are
40 000 currently active players
current starting elo is 1000
new starting elo is 500

then it would take 40 000 new players to shift the average to 750 (assuming that all the old players stay in the game)
or it would take 20 000 new players to replace 20 000 old players to shift the average to 750

I don’t think this would be too dramatic. if we get 40 000 new players, i think the game is thriving, and some shifting of elo benchmarks is acceptable

I like the idea of having some kind of assessment against AI: play ranked ladder against AIs of increasing difficulty and then estimate a rating (maybe even have some ‘undercover AIs’ on the ladder to know the AI’s rating)

one more thing that would be an option: the average has already shifted away from 1000 due to people leaving, and different k-factor in the initial games. As long as the average is above 1000, we could start new players with 900 or so until the average gets corrected back down to 1000

2 Likes

I also think laming should be banned under 1000 ELO. And treaty option checkbox available if both players agree they don’t want early rushes. Should make the new player experience a lot more enjoyable.

That’s a brilliant idea actually.

3 Likes

I agree with the OP that AOE could be much smarter about how to evaluate a player’s skill level than the elo system. Considering that the elo system basically doesn’t make sense at all for team games, and that an increasing number of players are primarily or even team-game only, I think it makes a lot more sense to start thinking of rating systems other than elo.

Many players don’t so much care about a ranking, so much as just being matched against players of similar skill levels to enjoy a fun, fair game, and maybe slowly improve over time. And we have much better ways of figuring that out today than they did in the 1960s. We can use complex statistics of how someone actually plays to model a player’s skill level which would be a much better reflection of their abilities.

I made a post on this a while back that’s relevant to this discussion. Check it out.

4 Likes

Well, it’s time for this topic again.
Here we go: PSA: You cannot lower the starting Elo. Don't recommend it, don't try it

3 Likes

I think it is worth considering how long this takes:

we currently have around 40 000 players. if we were to change the starting elo to 500, it would require 10 000 new players to change the average to 900. I think this would be an acceptable trade-off

1 Like