Andernut has a new Public Global Ranking score of 2933 - 75 = 2858 from game Invention

Lol... I lost a few games recently but crikey do you ever take a hit on Global Ranking from losing a game!

That'll learn ya.

Lost 100 once. Thanks to the cap, if I can say that :(

Not to mention tournament rankings.

Yeah, I've had more than a few that bad...

Andernut wrote:

Andernut has a new Public Global Ranking score of 2933 - 75 = 2858 from game Invention

 

Lol... I lost a few games recently but crikey do you ever take a hit on Global Ranking from losing a game!

And right here we see the volatility inherent in the current rating system.

The volatility is at the edges, where it belongs. The system has an entropic nature; poor analogy maybe, but that's all I've got before my first cup of coffee in the morning.

btilly wrote:

Andernut wrote:

Andernut has a new Public Global Ranking score of 2933 - 75 = 2858 from game Invention

 

Lol... I lost a few games recently but crikey do you ever take a hit on Global Ranking from losing a game!

And right here we see the volatility inherent in the current rating system.

Help! Help! I'm being oppressed!  

(not sure if you were going for a monty python quote there, but that's what it sounded like)

Ozyman wrote:

btilly wrote:

Andernut wrote:

Andernut has a new Public Global Ranking score of 2933 - 75 = 2858 from game Invention

 

Lol... I lost a few games recently but crikey do you ever take a hit on Global Ranking from losing a game!

And right here we see the volatility inherent in the current rating system.

Help! Help! I'm being oppressed!  

(not sure if you were going for a monty python quote there, but that's what it sounded like)

Bloody peasant!

I thought the same thing :)

Ozyman wrote:

btilly wrote:

Andernut wrote:

Andernut has a new Public Global Ranking score of 2933 - 75 = 2858 from game Invention

 

Lol... I lost a few games recently but crikey do you ever take a hit on Global Ranking from losing a game!

And right here we see the volatility inherent in the current rating system.

Help! Help! I'm being oppressed!  

(not sure if you were going for a monty python quote there, but that's what it sounded like)

I wasn't.  I was referring to the fact that there are rating systems that take into account how active someone is, and have their rating bounce around less.  This causes active people to have ratings that tend to converge near their "true strength".

btilly wrote:

I wasn't.  I was referring to the fact that there are rating systems that take into account how active someone is, and have their rating bounce around less.  This causes active people to have ratings that tend to converge near their "true strength".

Are we talking about two different things here ..general volatility, or volatility near the edges?  I'm not sure that improving one necessarily improves the other.

M57 wrote:

btilly wrote:

I wasn't.  I was referring to the fact that there are rating systems that take into account how active someone is, and have their rating bounce around less.  This causes active people to have ratings that tend to converge near their "true strength".

Are we talking about two different things here ..general volatility, or volatility near the edges?  I'm not sure that improving one necessarily improves the other.

They are connected.  Volatility is more extreme near the edges.  But measures to make ratings converge will help with both.

btilly wrote:

M57 wrote:

btilly wrote:

I wasn't.  I was referring to the fact that there are rating systems that take into account how active someone is, and have their rating bounce around less.  This causes active people to have ratings that tend to converge near their "true strength".

Are we talking about two different things here ..general volatility, or volatility near the edges?  I'm not sure that improving one necessarily improves the other.

They are connected.  Volatility is more extreme near the edges.  But measures to make ratings converge will help with both.

Suggestions for improvement?

Personally, I hate that someone with an h rating of 51% can eventually climb his way to the top by playing a ton of games. Doesn't seem fair to those with h ratings over 60%.

erastus25 wrote:

btilly wrote:

M57 wrote:

btilly wrote:

I wasn't.  I was referring to the fact that there are rating systems that take into account how active someone is, and have their rating bounce around less.  This causes active people to have ratings that tend to converge near their "true strength".

Are we talking about two different things here ..general volatility, or volatility near the edges?  I'm not sure that improving one necessarily improves the other.

They are connected.  Volatility is more extreme near the edges.  But measures to make ratings converge will help with both.

Suggestions for improvement?

Personally, I hate that someone with an h rating of 51% can eventually climb his way to the top by playing a ton of games. Doesn't seem fair to those with h ratings over 60%.

You make a rating a combination of two variables, one for the rating, the other for the uncertainty.  Uncertainty grows over time, and drops every time you play.  When two people play, the one with an uncertain rating sees their rating move more than someone with a certain rating.

People can get any rating, and can get to any rating from any other over time.  But active players won't see their ratings bounce around as much day to day based on the luck of the last handful of games.

erastus25 wrote:

 

Personally, I hate that someone with an h rating of 51% can eventually climb his way to the top by playing a ton of games. Doesn't seem fair to those with h ratings over 60%.

I think it is important to take into account the difference between H ratings for dueling (one vs. one) and H ratings for multiplayer games.

On many boards, a duel is extremely dependent on luck.  This causes luck to be much more important than strategy (the outcome has a high sensitivity to luck).  As a result, an H rating of 65% should indicate a strong player.  This is because, if we assume luck determines the outcome, then a player would have had bad luck on 50% of the games, yet their skill in a large portion of those games overcame the poor dice.

For multi-player boards, strategy tends to be more of a factor than luck (the outcome has a lower sensitivity to luck).  So, an H rating of 70-75% could indicate a strong player.

These are simplifications, of course, to underscore my point that low H ratings do not necessarily indicate  non-deserving players - you need to look at the mix of games played.

You can see on several boards that I have done well on, my H rating is lower than those ranked below me.   This is because I built up my rating by dueling and then switched over to multi-player games to increase my ranking further.  I need to do this because dueling is so luck dependent it is difficult to get a rating much over 1500 on many boards.  (You need to win more than 65% of your games to increase your rating against players of 1000 rank)

SquintGnome wrote:

erastus25 wrote:

 

Personally, I hate that someone with an h rating of 51% can eventually climb his way to the top by playing a ton of games. Doesn't seem fair to those with h ratings over 60%.

I think it is important to take into account the difference between H ratings for dueling (one vs. one) and H ratings for multiplayer games.

On many boards, a duel is extremely dependent on luck.  This causes luck to be much more important than strategy (the outcome has a high sensitivity to luck).  As a result, an H rating of 65% should indicate a strong player.  This is because, if we assume luck determines the outcome, then a player would have had bad luck on 50% of the games, yet their skill in a large portion of those games overcame the poor dice.

For multi-player boards, strategy tends to be more of a factor than luck (the outcome has a lower sensitivity to luck).  So, an H rating of 70-75% could indicate a strong player.

These are simplifications, of course, to underscore my point that low H ratings do not necessarily indicate  non-deserving players - you need to look at the mix of games played.

You can see on several boards that I have done well on, my H rating is lower than those ranked below me.   This is because I built up my rating by dueling and then switched over to multi-player games to increase my ranking further.  I need to do this because dueling is so luck dependent it is difficult to get a rating much over 1500 on many boards.  (You need to win more than 65% of your games to increase your rating against players of 1000 rank)

 

 

 

Agreed to both.

@ Erastus25: While the lower H-ranked player can beat the higher H-ranked player in rating by playing more games, his equilibrium rating will be reached much earlier. So on the long term the better player will always have the higher ranking.

@Squintgnome: I totally agree on your point here. There is a big difference in playing heads-up and multiplayer games. In 1 vs 1 games your actions directly affect the other players position. Luck is therefore a more important aspect of the game. Often the starting seat is an aspect of 'luck' as well, while this is less important in multiplayer.

As to the original poster: High winds blow on high hills. You need to maintain a H-ranking of approx. 75% (against 1K average opponent) to stay above 3K rating. Good luck!

Luieuil wrote:

@ Erastus25: While the lower H-ranked player can beat the higher H-ranked player in rating by playing more games, his equilibrium rating will be reached much earlier. So on the long term the better player will always have the higher ranking.

This is true, but the long term can be very long. In games where you hit an 80% H-rating, it seems to take 50-100 games. But when it's more like 90%, it's a lot more games.

I have experienced what Erastus is talking about A LOT. I play 30 games or so before wanting to move on to another game and then suddenly there's this person passing me (with a noticably lower H-rating) who's played over 150 games.

My experiences with the kind of system btilly describes is that equilibrium is hit very quickly in comparison.

Agreed.That system would be fair towards the 'better' player. However this site not only rewards skill, but also active play. Without activity there no games and no games is no fun. So I do not mind giving up a few boards to the more active players who have lower H-ratings.

The global rankings is basically a measure of H-ranking as equilibrium rating will be achieved much earlier then at the board rankings. The board rankings on the other hand also reward active players. You may not be the best at each board, but by playing alot you can grab lots of points as well.

You also have to keep in mind that some players (like me) will choose who they are playing against. In my case, avoiding players with a very low GR will help me to keep a high GR but will limit my H-ranking. On the opposite, people who play a lot agaisnt weak players will get lots of easy victories thus a high H%. But when they will lose, because of bad luck for example, their GR will drop sharply. 

Toto wrote:

You also have to keep in mind that some players (like me) will choose who they are playing against. In my case, avoiding players with a very low GR will help me to keep a high GR but will limit my H-ranking. On the opposite, people who play a lot agaisnt weak players will get lots of easy victories thus a high H%. But when they will lose, because of bad luck for example, their GR will drop sharply. 

Good point Toto, opponent selection is another important factor.  I do not have the patience for that, so I will almost always start a game instead of joining a game (you can only select opponents by joining).  So that leaves me in your second category of players.

Yeah I don't select opponents, I just find it odd that you can lose more points in a 1v1 match than you might gain from winning in a 4 player game.  I am guilty of jumping into those 1v1 luckfest speed games.  Generally I need to win 7 of those for every 1 I lose to not plummet in rankings.