Why ELO doesn't suit to BGA

Suggestions & ideas / Suggestions & idées
Locked
User avatar
sourisdudesert
Administrateur
Posts: 4630
Joined: 23 January 2010, 22:02

Re: Why ELO doesn't suit to BGA

Post by sourisdudesert »

demiurgsage wrote:How will ELO formula work with beginner in play
The idea is to keep the existing ELO formula and keep things simple.

As we introduce a "minimum ELO" (ex: 1400), this is going to produce a global inflation of ELO for everyone.

On the opposite, at each beginning of Season, we will "reset" the ELO to lower values => this produce a global deflation of ELO for everyone.

Of course we'd like to balance these 2 things so that ELO is more or less the same than today, except for weak players (minimum ELO) and very good player (who "race" for an high ELO).
demiurgsage
Posts: 229
Joined: 07 August 2015, 10:08

Re: Why ELO doesn't suit to BGA

Post by demiurgsage »

sourisdudesert wrote:
demiurgsage wrote:How will ELO formula work with beginner in play
The idea is to keep the existing ELO formula and keep things simple.
As we introduce a "minimum ELO" (ex: 1400), this is going to produce a global inflation of ELO for everyone.
I see a potential issue. If you want to get a trophey of the season there is no point to play at start of the season. With inflation every day playing with the players of your level becomes easier. If at start of the season there is around 10% of Experts and Masters, at season's end there will be around 50% of Experts and Masters.
User avatar
sourisdudesert
Administrateur
Posts: 4630
Joined: 23 January 2010, 22:02

Re: Why ELO doesn't suit to BGA

Post by sourisdudesert »

demiurgsage wrote:I see a potential issue. If you want to get a trophey of the season there is no point to play at start of the season.
Yes and no. This true in theory, but practically when there is a new game on BGA a lot of players are rushing to get the #1 position.

By the way I think this is positive : if a very good player is not available at the beginning of a Season, it will be easier for him/her to catch up.
User avatar
glitter and doom
Posts: 78
Joined: 14 January 2013, 15:27

Re: Why ELO doesn't suit to BGA

Post by glitter and doom »

sourisdudesert wrote:We have in mind a system based on "Season points" like this :

_ when you beat an average player : +1 pt
_ when you beat a Good Player : + 10 pts
_ when you beat an Expert : + 100 pts
_ when you beat a Master : + 1000 pts
_ seasons points are reset to zero when each period (ex : year or season) begins.
_ at the end of the period, we attribute trophies based on who get the most points.

Note : only the first match against a player during a given period gives you season points, otherwise you'd just have to play 100+ games against the same player to be champion.
Well, I'm sorry to say this, but it doesn't seem like a good idea. That system has many flaws and it seems highly unfair.

1. The way the points are awarded based on the opponent level is unfair

The groups are too large (200 elo), and the difference between the points is too big (10x).

As a result, you get the same number of points for beating the players with 1800 and 1990 elo, which seems unfair. Even worse, beating the 1800 elo player would get you 10x more points than beating the player with almost the same skill - 1790 elo. It just doesn't seem right.

The following figure depicts the expectation of earned points based on the opponent skill, for a player with 1700 elo, and no-luck game.

2. The system rewards the frequent play rather than skill

It is far better to win 30 out of 100 games, than 20 out of 20. Basicly, it will become a contest who will play more games, not who is the better player.

There are players who specialize in a single game and plays that game over and over. It is impossible to compete with these players if you enjoy playing many different games, like most of us do.

3. It will become more difficult to find the opponents; the total number of the played games may decrease

Some players will become desired targets (players at the bottom of the range, i.e. 1800-1820), while others will be heavily avoided. Who would like to play with a 1790 player, when you have the same chance against 1800 player, and 10x bigger reward!?

Therefore, it would be extremely difficult for players at the top of the range to find the opponents.

In addition, the fact that only the first match against a player is rewarded will make it even more difficult to find the opponents.

4. The system is unfavorable to the Masters

In some games there are only a one or two Masters. For those Masters the system is unfair. For example, take a look at Tokaido:

There is only one Master (Journeyman), and 50+ experts. Journeyman is clearly the best player, however, he has no opportunity to play against other masters. He may play with 10 experts, win 9 of those 10 games and get 900 points. However, the player who beats him will instantly get 1000 points and get on the top of the ranking. It seems unfair, but is seems you already know that:
sourisdudesert wrote:So being less than a Master is a kind of advantage too (as your opponent will gain less point than you in case of a victory).
Yes, being a Master is a disadvantage. Don't you think it's unfair?

5. The new system would not solve any elo related problems, it would mearely introduce new ones

Since the new system uses elo as a reference, it will necessarily inherit most of the elo related problems (luck-based games, multiplayers...).

And since the titles (master, expert...) will continite to be based on elo, people will still care about elo. In fact, I believe that many players care more about the title than the rank. For example, take a look at the Race for the Galaxy rankings. Do you think someone with elo 1810 is more happy about the fact that he is TOP 200 player, or that he is an expert?
User avatar
glitter and doom
Posts: 78
Joined: 14 January 2013, 15:27

Re: Why ELO doesn't suit to BGA

Post by glitter and doom »

sourisdudesert wrote: [*] At the beginning of each Seasons, decrease the ELOs of everyone to a starting point (ex : to 1900 for those above 1900, to 1700 for those between 1700 and 1900, ...).
OMG. :shock:

I take back everything I said about the previous proposal. It was excellent. It was perfect. Please go with the first proposal, just don't decrease the elo.

To decrease someone's ranking from 2050 to 1900 is huge. Do you know how difficult is to get it back to 2000+? The same goes for 1850 -> 1700. I wonder how many people will be happy when they lose their elo and their titles.
User avatar
glitter and doom
Posts: 78
Joined: 14 January 2013, 15:27

Re: Why ELO doesn't suit to BGA

Post by glitter and doom »

By the way, I fully understand your objectives, and I think they're good. I just don't think it is a good idea to reduce people's elo. For many players elo & titles are achievements per se, and they are much more important than prestige points. I'd rather lose all my prestige points than Master titles.
sourisdudesert wrote:
  • We want to introduce regular "Seasons", to give you permanent achievements when you reach the top, in replacement of the current "volatile" trophy that you lose as soon as you lose your ranking.
  • We want to give "Seasons trophies" to players who really deserve it, which means you won't be able to win a trophy just by beating a beginner every month.
  • We want a system than is less discouraging for weak players. At now they starts at 1500, and then decrease to 1200-1300 after few games which is quite discouraging.
User avatar
sourisdudesert
Administrateur
Posts: 4630
Joined: 23 January 2010, 22:02

Re: Why ELO doesn't suit to BGA

Post by sourisdudesert »

glitter and doom wrote:I take back everything I said about the previous proposal
You should remove this post as we do not consider this solution anymore.
glitter and doom wrote:I wonder how many people will be happy when they lose their elo and their titles.
You will keep your titles, forever, as "Seasons champion" and a trophy for having reached "Master" level during one Season. This is much better for you than today, because if you stop playing a game you lose everything.

Yes, this is difficult to climb ELO from 1900 to 2000+, but a competition is something difficult, and the rules of the race will be the same for everyone.

We think about it a lot, but keeping ELO for ranking + having Season champion = decrease the ELO at the beginning of Season. Otherwise it's like a race where you give someone a 10km advance.

To say the same thing differently : if we give you every 3 months the "BGA Season trophy" - the most prestigious BGA awards - for Spyrium/Battle Sheep/Neutreeko, while you only played these games 3-4 times during each period, this is not fair not serious.

I know it's difficult to hear because you are very strong with the current system, but in fact you'll be strong too with the new one. You may have to focus each season on a specific game to get the trophy if you don't want to spend too much time, but at the end you'll have this trophy forever, without the need to "defend it" every month :)
User avatar
glitter and doom
Posts: 78
Joined: 14 January 2013, 15:27

Re: Why ELO doesn't suit to BGA

Post by glitter and doom »

sourisdudesert wrote: You should remove this post as we do not consider this solution anymore....
Well, this is too bad, because I actually think that your first proposal could be improved in a way that it overcomes all the shortcomings, and still meet the objectives you've mentioned previously. I hope you'll take a time to read it and think about it:

The following is the same as your initial proposal:
- The elo shall remain unchanged, as well as the master/expert/good player titles
- The season points are reset at the beginning of the season
- Everyone starts from the same position at the beginning of the season, no one has an advantage
- The season points can only increase during the season, i.e. it is not possible to get negative points
- At the end of the season, trophies are rewarded based on the most season points

Here are the changes:
- Points awarded based on opponents elo should be smoothed. There should be a lot more groups with smaller elo range. The difference between awarded points should not be too large. For example:
  • o < 1400: 1 point
    o 1400 – 1420: 2 pts
    o 1420 – 1440: 3 pts
    o 1440 – 1460: 4 pts
    ...
    o 2000 – 2020: 31 pts
    o 2020 – 2040: 32 pts
    o 2040 – 2060: 33 pts
    ... and so on
- Maximum of the ten best results is taken into account
- Maximum of two games with the same opponent are taken into account (not necessarily the first two games)

It is important to set the maximum number of games for several reasons:
- everyone has a chance, not only the players who play hundreds of games each month
- it is impossible to get at the top by winning a huge number of games against weak players
- the system is not discouraging to players who started badly; they have a plenty of time to recover
- the contest will be uncertain and exciting until the end of the season
User avatar
sourisdudesert
Administrateur
Posts: 4630
Joined: 23 January 2010, 22:02

Re: Why ELO doesn't suit to BGA

Post by sourisdudesert »

Thanks, this is a very interesting system.

However there is no "perfect" system. Each system encourage a certain type of competition :
  • The current system encourage to be very strong at 5-6 games, and then to play few games to keep its position at the top.
  • The Seasons point system encourage to play a lot with a maximum of different players (possibly strong ones), to focus on 1-2 games.
  • The system you are proposing encourage to play few games against very strong players, and discourage to play again with the same players (no more point to win).
  • The "ELO reset" system encourage to play quite a lot, it doesn't matter if you play weak/strong players (it gives you the number of point you deserve).
So the question for us is more about the objectives of the competition than on the system itself. What we'd like is to build a competition where you are NEVER discouraged to play against someone :)
latexrhombus
Posts: 65
Joined: 14 November 2011, 22:40

Re: Why ELO doesn't suit to BGA

Post by latexrhombus »

I am nowhere near a top-ranked player like Glitter and Doom, but the idea of resetting ELOs periodically would not make me want to play more. Working up to those high levels takes a lot of work, and constantly resetting it seems really unfair. I also do not believe the proposed system would take away the natural discouragement higher ranked players have from player lower ranked ones. In fact, the benefit for taking down a master is so great a master would be best served not to play them...it is so extreme, in fact, it may encourage untraceable cheating as it would only require one play to gain massive "seasons" points.

I think players should be encouraged to play people of a similar ELO ranking and I believe the existing system does that pretty well (except for 3+ player games, which I have stated my 2 cents about). The proposed system seems to basically penalize the top players, so I do not think it would be advantageous.

Also, I think an important aspect of BGA is to encourage people to play/try lots of different games. I imagine it would be hard to diversify one's efforts in seasons if one wants to do well in any of them.

For the top 20-50 players in each game by ELO, why not just have some type of unranked (no ELO) leaderboard system where players can challenge players 1 space above them to turn based games. If a player cannot finish in 30 days, they lose their place in the leaderboard and the corresponding rewards. They could start again from the bottom later.

This encourages the players at the top to stay active. Because as it stands, lower-level players really have no reason not to play because they do not lose much ELO when they lose and gain a lot if they beat a higher level player, so I think the system should actually encourage the top guys to play each other or for lower level players to work their way into that top group.
Locked

Return to “Suggestions”