How does the ELO system work?

[Clagger18]

 

My question is: If I win a bunch of unranked duels, will I face tougher opponents once I go back to ranked?

 

 

[razimove]

3 minutes ago, Clagger18 said:

If I win a bunch of unranked duels, will I face tougher opponents once I go back to ranked?

 

you answered your own question. Either way, the current elo system isn't exactly good, since we don't have a big enough player sample(number of players), to make it viable, part of the reason why people with <50% win rate, and 50~52% win rate end up in top20s.

Edited August 29, 2018 by razimove

[OrangeManiac]

13 hours ago, razimove said:

you answered your own question. Either way, the current elo system isn't exactly good, since we don't have a big enough player sample(number of players), to make it viable, part of the reason why people with <50% win rate, and 50~52% win rate end up in top20s.

Low winrate isn't inherently the problem with ELO system, if you happen to win 52% of the time against the top 20 chess players in the world then you must be pretty damn good and "deserve" a high ELO despite a seemingly mediocre winrate. The problem here is that ELO isn't the greatest system for a game that is largely about luck - as a player you can only do so much. Winning a top 20 player doesn't inherently make you a good player and in Pokemon winrate against everyone has arguably more value than winrate against only "the best".

[razimove]

40 minutes ago, OrangeManiac said:

if you happen to win 52% of the time against the top 20 chess players in the world then you must be pretty damn good and "deserve" a high ELO despite a seemingly mediocre winrate.

Not arguing against that, I agree. But in this small player sample, having 52% doens't equal good imo, one can just play 1000 games, win 520 of those, and end up abusing the elo system in place, in a larger sample, like in pokemonshowdown, I'd agree totally with it. Right now, the only way to measure skill, in my opinion, is with tournaments, as ranked almost irrelevant. 

[Ghartark]

I dont know the inner workings of the elo in this game but being with 50%-55% winrate at the high elo only means that matchmaking tries to push your winrate to that range to make a fair experience for all players, its similar for games as hearthstone and league, where despite you having winrates in that range, you still push the ladder thru grinding and winning a slight more games than you lose.

[razimove]

21 minutes ago, Ghartark said:

I dont know the inner workings of the elo in this game but being with 50%-55% winrate at the high elo only means that matchmaking tries to push your winrate to that range to make a fair experience for all players, its similar for games as hearthstone and league, where despite you having winrates in that range, you still push the ladder thru grinding and winning a slight more games than you lose.

ok, but the difference, is the playerbase in those games is really high, so you actually have to be really good in order to achieve something.

In pokemmo, although the playerbase, and the competitive scene grew a lot recently, I still think there's not enough incentives for people to actually play the ladder, and therefor it's easily abused by those who even have <50% win rates, there's some on the ladder like 

 

Although top84 isn't that high, previous to this, there were people even in top40 with similiar or worst win rates.

[Clagger18]

On 8/29/2018 at 11:23 PM, razimove said:

you answered your own question. Either way, the current elo system isn't exactly good, since we don't have a big enough player sample(number of players), to make it viable, part of the reason why people with <50% win rate, and 50~52% win rate end up in top20s.

If you are right, then I can also go to normals and lose a bunch of games, then return to rank and face weaker opponents to increase my winrate more easily? This doesn't make sense

[razimove]

3 minutes ago, Clagger18 said:

If you are right, then I can also go to normals and lose a bunch of games, then return to rank and face weaker opponents to increase my winrate more easily? This doesn't make sense

No, doenst matter what you do in normals, ranked system has a elo system behind, but that system only represents your place on the ladder, you might face though opponents or not, a good player doens't need to be necessarily on the ladder, in fact, a lot of the best players, dont even play ranked actively.

[Ghartark]

1 hour ago, razimove said:

ok, but the difference, is the playerbase in those games is really high, so you actually have to be really good in order to achieve something.

In pokemmo, although the playerbase, and the competitive scene grew a lot recently, I still think there's not enough incentives for people to actually play the ladder, and therefor it's easily abused by those who even have <50% win rates, there's some on the ladder like 

 

Although top84 isn't that high, previous to this, there were people even in top40 with similiar or worst win rates.

But then the problem won't be that the system fails at ranking properly a player but instead, the competitive playerbase is so low that the race for the top is not that much of a contest. Anyway, regardless of the true cause for this inconsistencies, your solution seems like an attractive one, a decent incentive for top 50 players each month (like pvp only cosmetics) would make for a more competitive experience and a more fun pvp.