I wrote about this before, but I am bringing it up again because I feel it is important.

I thnk there should be a cap on how many points you can be awarded.  I think this cap should be 30.  There are two reasons for my thinking:

1. Uncapped points do not make sense.

2. Uncapped points are a STRONG deterrent for good players to start games.

To illustrate both points, if a 2000 rated player loses to a 500 rated player, does it really make sense that the 2000 rate player lose 80 points?  The intent of the system is to award more points to new player so they can reach their equilibirum.   It doesn't reflect the intent of the point system to punish the high rated player so harshly.

I think 30 point max is about right because that is the point where the higher rated player must win 75% of the games versus the other player to break even.  Even though some players have H ratings greater than 75% I think they are the rare exception (in fact there are only 9).  Asking someone to win more than 75% of their games given all the luck variables is unrealistic, so once a players rating is more than 1.5 times their opponent they will not want to play given this point system. 

In this situation high ranked players just wont play given the potential loss of points.  Even when you reach a rating of 1500 you have to begin to worry about this.  You can sit back and wait to join games with higher ranked players but this greatly stifles the number of games that can be started.

This was prompted by a recent games where I lost 54 points. 

How many other players out there have stopped creating games because of this?

I fully agree with you and I already wrote about it. Without any effect...

I also stopped creating games because of that and also stop playing games  when my win % expectation is too low comapred to the high cost.

I am re-posting this from another thread to support what I said above and to correct my 'break even' statement above.  For a ratio of 1.5, the break even is 69%, not 75%

======================================

I wanted to present the reasoning behind why I propose a Point Range (min - max) addition to the scoring system.

I have listed below the ratio of players scores and the resulting points to be awarded using the current formula.  The Break Even column lists what % of games you must win versus that opponent to not lose any points (break even) over the course of many games.  So, for example, if a player has a rating of 2000 and is pitted against a 1000 rated player, he must win 80% of the games versus that player to prevent his rating from dropping.

You can chose any %, but the sake of the post I will assert that given the influence of luck in the game it is extremly unlikely that you can win more than 80% of your games against any opponent regardless of how superior your skills are.  If that is the case then the point system should not allow a scenario where you must win more than 80% of your games to break even - hence a cap is necessary.  If the best players can win 80% of their games at most, a system without a cap can have players performing at a maximum level and still lose points. For example a 3000 rate player can win 80% of their games versus a 1000 rate player and still lose point.  The system forces players to avoid playing any whose rate is less than 1/2 their own.  A point range whould fix that.

I think you can only lose a max of 100 points, so there is a cap 

I think you make a great case though and while normally I would be less for this since those that win more and have a higher ranking also have a higher win percentage and know the board better, but since luck and random initial placement (for most of the boards that players get high rankings on or lead to higher rankings) can play a role I wouldn't be against seeing something like your proposing.  

I'm not positive 30 is that cap, could you expand on why 30 or some math behind what 30 really means (perhaps you've done this already though, I haven't studied everything quite in depth yet)?  Perhaps 40 would be good as it would be double what players would get if both were 1000 ranking.

we are the 99%

it is unfair of you to hoard all the points, they should be redistributed.

Personally: I agree with Candy Cane.  In no other game/sport is the person on the top of the leader-board protected against a loss to a much weaker opponent.  Sure you might lose a chunk in this particular game, but if you aggregate that over a large number of games, your ranking should end up right back where it was, if you're really as good as your previous score indicated.

Also, if you're as good as your ranking suggests you are, you shouldn't have a fear of playing 2000 vs 500 games.  This lost might be the 1 in 30 that you'd lose to this opponent, but that just makes this the outlier, but no less meaningful.

This is a game.  This isn't national rankings that are leading to sponsorships, nor is it earning you any money.  Get your a$$ back in the game and take the loss like a man, then earn it back again.  And those of you who DO avoid games against the lesser players, I feel that that earns you a "Family Feud X" for being pansy, and it makes your current ranking score less meaningful to the rest of us (because you only have it because you've been selective over your opponents).

But for the mathematics, if you're going to cap anything, you need to calculate the cap on the losing end, not winning.  If you cap winnings, then there has to be some ridiculously complicated algorithm to determine how the 30 points awarded to winner are pulled from the losers (based on their respective rankings).  Set the cap on the losing end, if you must, just to make Tom's life easier (and easier for a layperson to calculate and confirm that they were awarded/decremented the correct points).  Either way, I still disagree with the idea.

wow.  actually i agree with a point loss cap as it was stated at the beginning. i just felt like stirring the pot a little to see if it would boil.

i just lost a 1400-1000 game.  it cost me 28 points.  we had a rematch and i won.  i got back 14.

i understand and agree with this reasoning.  but again, many games here are one on one, and to loose 100 points just because of crappy starting position (possible for even the best player) I think that may be a little much.

If there was going to be a change, it shouldn't be capping point transfers, it should be calculating them logarithmically.

But Boris has the right of this.. capping (or otherwise limiting) point losses only serves to inflate rankings.

Candy Cane wrote:
i just lost a 1400-1000 game.  it cost me 28 points.  we had a rematch and i won.  i got back 14.

Right, and this is how it should be.  You lose 28, you win back 14.  And then you play that opponent 8 more times, and you should still win significantly more than half the games (assuming you're really a 1400 rank player, and he is a 1000 rank).  By the end of those 10 games, your point should have gone up, and his down, just like expected.  This ranking is an aggregate, and I don't think that we should be putting boundaries on the wins/losses to protect anyone.

Exactly. Like BD implies, a cap that "bites" will only inflate rankings among the higher ranked players (while holding down gains for lower ranks), so it's not obvious that a capped system would be more fair.

In reality, how often does anyone hit the 100 point cap? I just looked at my 315 games, the most I've lost was 77 once, 64 once, and a few in the 50s.

But SquintGnome makes another point, about incentives. 

From my perspective, it seems like there's some incentive to shy away from smaller games (<5, say) -- the risk isn't worth the reward -- except, of course, several of the top GR players are dueling specialists, so maybe not. I don't avoid low-ranked players, and I haven't noticed other high GR players doing so (except Toto says he does, of course). You can't get up there unless you start a fair number of games, and whoever joins, joins. Conversely, there is a particular incentive to play other high ranked players, which is a good thing. 

You gotta play... to win!

Please keep in mind that my suggestion is not about avoiding opponents, games, or point loss.  If you look at my stats you will see that I have played many games against opponents of all levels.  But I want to be fair to all players.

For substantiation of my claim, focus on my second post where I mention that a player can perform at maximum level and still lose points with this system (this is when a player needs to win 80% of games to break even).  In this case you cant make up points you've lost by playing more games.  So please address this point in a rebuttal.

In my opinion as your rank increases you get to a point where you don't want to play anymore.  To me this definitely indicates a flaw in the system.  Maybe my solution isn't the right one, but there needs to be some change I think.

There are those who mentioned that this is only a game so the ranking should not be important.  This is true and I laud those who embrace that philosophy fully, but I would not play if there was not a ranking system so to me it is very important and I guess there are many other players that share that opinon.

My 2c:

The system as it currently stands is designed so that your ranking should naturally level out at a score reflecting your skill level.

Capping will lead to huge runaway scores because for example if you are able to win 80% of your single player games on a particular board (certainly possible on some boards) you could in theory have a score that continuously increases because your downside exposure is limited.

Huge runaway scores are a disincentive to new players - they see these huge totals where they will have to spend a significant portion of their life to overcome.

The incentive to keep playing is generated by a) lots of different boards to get ranking scores on and b) other players can take your championship points by overtaking you in the leaderboard.

That said I am not totally averse to some kind of cap... just that it would be closer to 100 than to 30.

I am not sure about run away scores.  In my case I have a 3.4 ratio vs a 1000 rated player.  Which means if I play him 10 times and win 8 times I win about 40 points and lose 60 points in the capped system.  I would have to win about 85.7% of my games vs him with a cap of 30 in order to maintain 3400 vs a 1000 rated player.  In the current system I have to maintain a 93.1% win record vs the same player.  

The other thing to remember according to squint's suggestion is the points won by a player with a higher rating does not have a floor cap.  This also prohibits someone with a runaway score because as the gap spreads the win % needed still increases, just not as fast without a cap at either end.

Toto is not the only one who has stopped creating games for this reason.  

I don't really focus on my ranking & scores, so my perspective is limited.

If playing a diverse group of players would over time would cause your score to drift lower, maybe it is because your score is artificially high?

I don't know if everyone knows this, but tom was the Top Ranked Player on ToS before he created this site.  If I trust anyone to judge what is fair in a ranking system it is him.  Heck, he used to change is screen name in games because N00bs would blitz him just because he was who he was.  I know he wants to avoid this sort of thing.  Also, he stopped playing many maps that he actually enjoyed playing because he could only lose points so I trust he doesn't want that to happen either.

Anyways, if you don't start games because you don't want to lose points, that is your strategy for keeping your ranking.  Another player may want to keep playing because he/she has confidence in their skill to win the game or, crazy as this may sound, may just want to play.

I play Magic: The Gathering on XBox live and aside from that and Skyrim I haven't played much else.  Oh, Nights of the Old Republic, but that's an old school Xbox game.  Anyways, I don't have that many points and so when trying to play Live games of MTG, people don't team with me and always join games against me because they figure I don't know how to play because I don't have points.  They quickly learn that I'm a very good Magic player (even with the XBox limitations).

Anyways, that last part was just me venting but the moral of it was supposed to be that Rankings are cool but they don't tell the whole story.  The best way to prove you are better than somebody is to join a game and place your armies...other than that I can't help you.

I'm RiskyBack and I approve this posting

If playing a diverse group of players would over time would cause your score to drift lower, maybe it is because your score is artificially high?

Correct

Anyways, if you don't start games because you don't want to lose points, that is your strategy for keeping your ranking.

Definitely.

Also, if you don't want to risk your points, play private games?

There have been a lot of good comments, so I want to thank everyone for contributing.

I disagree with a couple of comments so I would like to focus on those aspects this post.

First, I want to set as a premise for my argument that an 80% win rate is the realistic maximum a player will have.  Out of 540 players with 100 or more games only 2 have an H rating of more than 80% -  that is .3%.  There are exceptions, of course, and indvidual boards are different but I think it is reasonable to say that 80% win rate is a maximum level of play.

Runaway scores - At 80% win rate you will break even when your rate compared to your opponents is a ratio of 2 to 1.  This is what prevents run away scores.  If an ideal player (at 80% win rate) plays all newbies, their rate will never exceed 2000.  To increase your score you will need to play people with higher ratings on average.  Also, as Falker mentioned, the points you win will decrease as your rate rises.

There was a comment that if playing those with lower scores makes you drift lower then your score was artificially high.  I assert that this in incorrect.  Lets assume that we know a player is a 2000 rate player playing at 80% win rate.  If that player continues to win at 80% win rate but plays people rated at 500 their score will drop well below 2000.  So their rate was not artificially high but it dropped because there was not a point loss cap. 

I think we might need some mathematician input on this ().

On an individual board a win rate of 90% is not impossible. For example, look at BlackDog's win ratios: http://www.wargear.net/players/info/BlackDog/Player%20Stats

SquintGnome I do get your point about discouraging repeat play once you reach a certain score... but there are only two possibilities here as a result of decreasing the points cap:

1. It does not result in runaway scores. In which case it will just mean that the 'equilibrium point' at which your ranking score does not increase further has just been shifted higher. We have not really fixed anything other than extending the time it takes to reach this point.

2. It does result in runaway scores. In which case we should not implement it.

For (1), this may still be a worthwhile thing to do if it increases the amount of time top players spend playing a particular board.

I agree that win rates of more than 80% occur.  I offer these counterpoints.

* Win rates higher than 80% are more likely when less than 10 games have been played on a board.  Just as likely are players with win rates lower than 20% in this situation. 

*Win rates higher than 80% on boards with more than 10 games played are likely uncommon (outliers).  We can check this.

* If we set 40 as the point loss cap, a player starting at a 1000 rate on a board playing all 1000 rate players would need to win more than 76 games in a row (without a loss) for the point cap to kick in.  This is when they would get to a rate of 2000.  So, to take advantage of (abuse) a 40 point loss cap you would need to keep an 80% or higher win rate after more than 76 games on a board.  We can check how often this occurs.  I have simplified things a bit for the illustration since you will have opponents with rates other than 1000, but I think it makes the point.  After 76 games you should have developed enough competent players on a particular board to prevent any individual to go higher than 80% on that board.