Why ELO doesn't suit to BGA

[senatorhung]

current top20 in lifetime ELO for each game get the first 'seasonal' trophies for prestige (replacing the current lifetime ELO-based trophies and prestige).

each new season lasts 3 months, with 'seasonal ELO' set at 1500 to start for everyone (lifetime ELO still maintained and no longer tied to prestige).

at the end of each season, you award new trophies based on the 'seasonal ELO', even to those who already have a seasonal trophy for that game. (prestige points can be tweaked)

in place of the current lifetime ELO trophies, you would award 'top' player trophies for those with the most overall prestige in a game (which includes tournament prestige as well as seasonal trophy prestige). these can maintain the current 50k, 25k, 10k prestige values.

so, in order to maintain a prestige ranking, players will have to compete for seasonal trophies, or win tournament trophies. if someone doesn't play, they will eventually get passed by those who do play.

[demiurgsage]

Some though about tournaments.
For a game like Seasons, 11.800 players played recently. If we want to organize a tournament with the "top 1%", which is quite exclusive, we must organize a tournament with 118 players. It means that we must organize a 8 round Swiss system tournament, 45 minutes between rounds. This is too long for a single evening, so let's say 2 consecutive evenings (Monday+Tuesday) at 21:00 CEST, 4 rounds each.
Needless to say that this is impossible to organize a turn based tournament with so much players : if two players are from very different timezone, they are going to play 2 moves / day and you cannot have your 8 games during a 3 months season.

There are many solutions to solve this.
For examle, you can create real-time qualifiers. Each player can choose a qualifier in his time zone. One or several players from each qualifier will advance. In final round only 8-10 players should remain. And they will play round robin turn-based tournament.


Also I have a proposal for not tournament players. I agree you shouldn't give any seasonal points for just being in top of the ranking. Give points for every win when you are in the top against the player of the same level or 1 level lower than you. How it can be realized.

• When you win against a player of the same level (e.g. Master vs Master) or higher you get (21-R) points, where R is your rank in top20 before the game.
• When you win against a player one level lower you get half of the points.
• No matter what level has your opponent if he also is in top20 you get the points as if he would be the same level as you.
• Only 1st game against every opponent in a week is counted for seasonal leaderboard. So players can't help each other to advance in the leaderboard.

It isn't final solution, so everyone is free to improve it.

[chrisa]

[*] 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, ...).

Another reason why periodically resetting Elo is a really bad idea, is that it disadvantages a lot players who are close to the reset levels, at least in the first part of the season (and if the season is short, 3 months is short, they may be disadvantaged for the whole time).

Let's consider 3 players and their Elo score in the current system: A (1899), B (1700) and C (1699).
At the beginning of the season, A and B get 1700 and C gets 1500.
A plays against B. The current Elo scores give him a probablity of 76% to win. If he does, the outcome will be +20 for A and -20 for B in the new system, while it would have been +10/-10 in the current system. Then A will be able to climb up quite fast, while B will lose more than he should have.
Now B can also play with C, and win with a probablity of 50% according to the current system. But, with the new system, a win will increase his Elo by 10 only, against 20 in the current system. And if he loses, his Elo will be decreased by 30 instead of 20 in the current system.
Conclusion: B should avoid playing with A and C, and should rather select players who were around 1500 in the current system to have a chance to improve slowly but surely.

Well, B could also manage to voluntarily lose a few games just before the beginning of a new season to have his Elo reset to a lower level (1500) and then to have a chance to improve faster. This is very similar to what people do when they create new accounts to look as beginners while they have, in reality, a Good+ level.

[Zetes]

This thread is getting really intense as the changings BGA wants to make are very important.

Sourisdudesert, as you are so interactive with other players, could I also have an answer about real time and turn based?
As you are making major changes to the ranking system don't you think it's time to have separate rankings for the two playing styles?

However you want to solve the other stuff, I suggest this change because it really doesn't make sense as it is now.

Example: at the top of the overall standings, if I am not wrong, only player Deszyck plays real time. He will never cross the path of the other top guys who play turn based. If you want to know who is the best you should separate the rankings because as I said in my other post, those two playing methods require different skills and even if you do not agree on this you cannot deny that real time players never gonna catch up with only turn based players…so who will ever know who is really the best? There can't be a mixed overall nr.1 , there only can be a nr.1 for real time and a nr.1 for turn based.

I repeat, the 2 playing styles require different human skills, if somebody likes to argue I am all ears.

Ok, let's also say something positive for turn based players….they play sometimes 10 different games at a time, that also require a particular human skill which is not needed in real time play. 2 different styles….

Also, splitting the 2 rankings seems to not create particular problems , players who play both styles appear in both rankings and that's it.

So, is this off topic here? Should I start a new thread with this argument?

[sourisdudesert]

don't you think it's time to have separate rankings for the two playing styles?

You can make the same argument for the different players configuration (ex : Legendary Inventors is a complete different game with 2 players). And you can make the same argument for the different options. At the end, we could have many many rankings for each game, and the result will also be unsatisfactory. Among all the unsatisfactory solutions, we chose the simplest one.

[Zetes]

Obviously 2p games and multiplayer games are different, not only in the game you mentioned, most games are pretty different depending on how many players play. If I play Tobago with 3 in no time I'll be out of the top, lots of bad players who involuntary and randomly help other players. So what?

That is a thing which is the nature of games, my point is that turn based and real time are different sports, different animals.

What you write seems more an answer to close this argument because you have more important things to do rather than really thinking about the matter.

So let's say this: how can I become nr.1 on BGA if I am not able to get elo points from this top5 players called "we all play turn based" ?
Should I start playing turn based? Pretty wrong. To overtake top positions (and get enough elo) you need to play against those top guys to take their elo.
It can also be done by winning against just experts…but is that a competition? The best should play against the best. At least periodically.

Should I check personally all the top rated guys to see who is the first ranked real time player and then have him as reference for best real time BGA player?

Anyway, the rankings are false this way, I say Karupisu and other strong real time players could maybe beat those top turn based players frequently…but we'll never know…the only thing I know is that the ranking doesn't display the true value of players this way.

Anyway, am I the only one stating this? Could someone who feels the same stand up and let MR. BGA know their opinion?

Reality is that those BGaming sites all tend to attract the most possible players, that's why they enable to play with the 2 methods and not discuss the flaws of it, in a perfect world there is yucata for turn based guys and BGA for live games.

[demiurgsage]

That is a thing which is the nature of games, my point is that turn based and real time are different sports, different animals.

I agree that these play styles are very different. But I also see the point in the sourisdudesert's words who want to keep the things simple. Maybe a partial solution is possible. I suppose that some games should be played "ELO on" only realtime by its nature like "Race for the Galaxy", "Oh Hell", "Belote" and many others. But I'm not sure that for a game like Madeira would be a good thing to separate realtime and turn-based ratings.

[sourisdudesert]

But I also see the point in the sourisdudesert's words who want to keep the things simple.

That's it.

[Zetes]

I don't see how having 2 rankings would be complicated….the other process that's going on (elo/non-elo), that is complicated.

If we all agree that the 2 playing styles are so different then it seems a natural process to get the 2 separate rankings.

If you are nr.1 in turn based ranking, you just don't care if there is another nr.1 with more/less elo or trophies in the real time category…he's playing another sport.

Again, if someone wants to argue on the fact the 2 styles requires different skills…I am all ears.

Five tribes (not on BGA) is a good example of a game which in real time creates possibly analyzis paralyzis, there are so many options every turn that it is impossible math wise to get the best possible move in a few minutes. As AP is a stress factor and a way to start fights on a table, nobody dares to think too long and makes their decision based on the best move they see in a limited time…that is what this game is about.

I think we could even go to the point where we say that games for turnbased should made just for them. Almost all BG's requires you to think fast and that is a skill. Sometimes I ask myself if there are some of those turn based players who calculate all the possibilities and then make their move. Do you guys have an underground lab where you elaborate all day the best moves with stolen NASA software

Now, I would dare to state that BG's are not for turn based players, play live or do something else.

But I will go with some more moderation and say that modern society took away time from "we the kids" and why should those busy software progammers or whatever not have a chance to play BG's?

So, welcome turn based, welcome to another sport.

Life is not turn based, when you speak to a person you have to think and react, take this post for example..and all what's letters, mails…

When you write you have all the time to think, and you explore, and you correct and on and on untill a message is delivered.
The result will be more detailed and the message should be clearer than someone who speaks live.
But you will not get a real time feedback, you will not feel the emotions of the persons you are delivering the message to and so on..

What do you prefer? Answer will be 50/50

Sometimes it may be good to write, sometimes it's better to talk.

But you cannot deny that you need 2 different skills to be good at writing and at speaking.

In turn based BG's the detail in writing words translates in detail of math calculations, so if you really only care about being nr.1 I would state that you really need to play turn based. No need to talk.

[glitter and doom]

Again, if someone wants to argue on the fact the 2 styles requires different skills…I am all ears.

I agree there is some difference in required skills for turn-based and real-time mode. I just don't think it is such a big deal. I have played real-time games on BGA for quite a long time. Then I switched to turn-based mode. I haven't noticed any significant change in my Elo, rankings, or my win-loss ratio.

In my opinion, the difference between 2er and multiplayer is much more significant. Tactics and strategy are completely different.

However, I don't think there should be separate rankings. There are a lot of players who play both styles: 2er and multiplayer. The same goes for real-time and turn-based, many people play in both modes.