Inner workings of the game engine

I don't think the 'Vexer's repair' is fair. Why should a very high ranked player lose less than a moderately high ranked player? From the list above: Winner Loser Won/lost 1000 2500 28 1000 6000 17 (rather then 34) I understand the 6000 rated player does not like losing 34 , neither does the 2500 rated player like to lose 28. So why is mr 6000's problem fixed but not mr 2500? Furthermore, 28 on 2500 has more impact than 34 on 6000 (1.1% vs 0.56%). So the 'injustice' of the higher ranked player losing too much is fixed for the wrong player with the current repair. I'm more or less in line with Blagoje_Jovovic suggestion: 1 point won 1 point loss for 1v1 games. But I would make the win/loss amount depending on the rating difference between the 2 players. Equal rating => +-20, the higher the rating difference, the lower the win/loss amount for both. Should fix anomaly of old formula (fixed 20 points) and disproportionality of current formula.

Q 1: Why should players with a higher rating lose more points vs a lower rated player than vs a same-rated player? Because they are more skilled, and therefore are expected to win more games to break even on their rating. Q 2: Why not for 2 player games? Because the outcome of 2 player games depend a lot less on skill and a lot more on luck (drop, dice, neutral-positions, etc). Does that help?

what i do notice is tha if yo are in the 300 range...you get dinged the most on a 2 player game why isnt it equal....

Well, there are a few things that isn't equal.....but I will just focus on higher-ranked players playing lower-ranked players. The goal is to get higher ranked players to play lower ranked players. So this site chooses to do two of the following: 1. A modified 1v1 scoring system 2. A 50 point cap for losses. Both of these methods allow the higher-ranked player to keep/maintain their precious score....and yes....some players have figured out to exploit these rules and farm points as a way to enjoy this site more (instead of trying to find games that challenge them) but to each their own. I am disappointed that players do this and refuse my private game invitations for a good challenging game but honestly if it is that important to them then so be it. I do not dictate how they like to spend their fun time. I came from another site (which is not functioning currently) along with a few others and quite frankly we are quite surprised on how sites differ in terms of the community and setup. We have been here for about 6 months. There are many pros and cons and I hope to make suggestions here to improve the site here. I agree the top players should be well rounded in their game play and are able to win against all levels of players from other top players to noobs. This farming for points or playing a narrow range of games to create a high score is fine. Hey, if some players enjoy the same repetitive game, then go for it. However, for a true master, they do need to participate in tournaments but currently there is no incentive to play tournaments. (Sorry tokens are not an incentive especially for paying members...they are kinda worthless). So, at my old digs, our tournaments were tied to rating points. Suppose you have 36 players in a tournament and each player donated 20 rating points. The winner for the tournament will win 36 x 20 = 720 points! It doesn't have to be 20...it could be anywhere from 5 for short tournaments to 20 points for longer ones....up to the tournament organizers. This achieves two objectives: 1. It encourages all ranks to play with each other. 2. A true master at the game will come out on top. At my old digs, the top 3 players for the last 4 years were masters...consistently winning tournaments and having a score of 5000+ Those 3 top players were well respected and really super nice. Two of them took me in and trained me when I was a noob. (But that is for another post) Now some of you think this is not possible under the current setup...yes it is...if the admin can manual adjust points now, then no further coding is necessary. Once the tournament ends, only 1 admin goes in and takes 20 points from each contestant and transfers it to the winner....viola! Complete! That's how we did tournaments at my old digs. Works really well and everyone wins. Players that farm points can continue to do that but eventually the true masters will come out on top. It's ready to implement if you want it.

I believe you posted this exact series of points somewhere else. I personally really like the idea of submitting rating points into a prize pool for tournaments, to be given to the winner, or top 3, or whatever the tournament organiser chooses. There are a couple of reasons why this would currently be difficult, however. One such reason is that all staff here are busy, volunteers. Manually redistributing rating from each participant could take a really long time, especially for the larger tournaments. The other reason I can think of is that rating is a touchy subject for some here (^^), and relying on a manual system leaves room for human error, which would disrupt the purity of the algorithm (whereby all ratings average to 1000 (I think)), and would seriously hurt people's feelings and make a mess. It could be programmed in to make it an automated process, but that would be a big thing for the programmers to do (make the code, design an interface, debug, and optimise). Our programmers are very busy ^^^^ Having said all that, if it could become an automated process, I would definitely love to have extra rating points wagered into tournaments as periwinkle suggested :) If there's a better thread in which to hold this discussion, please redirect me haha

what the current scoring method us guys who have scores in the 3000 range, are reduced to disallowing players of lower scores simply because we cant take the "DING" by playng new players... especially since one vs one games are less about skill and more about drop, dice, and going first. i would like to also add, even though this may not be the place, that i believe that if a game starts with each player getting more than 3 , the whole game should be played at that number. it does not make the game fair if one player can get 4 and in their first turn reduce the other players number before they ever have their first turn. just saying.... going first should not be this much of an advantage.

Yes...I reposted here because where I posted before really didn't belong in that thread.

@Matty [i]Q 1: Why should players with a higher rating lose more points vs a lower rated player than vs a same-rated player? Because they are more skilled, and therefore are expected to win more games to break even on their rating. [/i] Only one sport comes to my mind where higher rated players get penalised: horseracing. They get extra weight to carry. And that's just to keep the betting interesting. Lewis Hamilton doesn't get a lower score because he's won I don't know how many times, DeChambeau doesn't need to leave his driver home just because he hits further than the rest. I don't understand why it is considered a given that players with higher rating lose more points. If lower rating players absolutely want to play them, fine, they will lose some more points. So? [i]Q 2: Why not for 2 player games? Because the outcome of 2 player games depend a lot less on skill and a lot more on luck (drop, dice, neutral-positions, etc). [/i] I agree that the outcome of 1v1 depends a lot on drop, dice, and first position, much more than in multiplayer games. However, in multiplayer games with increasing cards - which are the majority - it all depends on getting the set at the right moment and killing all the rest. Not much skill involved either. And if your point was true that 1v1 is mostly about luck and less about skill, then definitively higher rated players should not lose more than lower, since their 'skill' doesn't impact.

[quote=KaiserKnud]@Matty I agree that the outcome of 1v1 depends a lot on drop, dice, and first position, much more than in multiplayer games. However, in multiplayer games with increasing cards - which are the majority - it all depends on getting the set at the right moment and killing all the rest. Not much skill involved either. .[/quote] I disagree with this. Obviously the set is helpful, but if that was the name of the game, then you wouldn't have players with 40+% winning rates vs players with 15% winning rates in those games.

The set is as helpful as the drop, the dice and the starting position in 1v1. One could even argue that 1v1 requires more skill, where multiplayer is more about being very patient and letting the other kill each other. And you'll get players with 40+% winning rates vs players with 15% winning rates also in 1v1. But that's not really the point of the discussion I think. What I still don't understand is this list: Winner Loser Won/lost 5000 500 3 5000 1000 6 2000 1000 13 1000 1000 20 1000 1500 24 1000 2000 26 1000 2500 28 1000 3000 30 1000 3001 29 (rather then 30) 1000 4000 24 (rather then 32) 1000 5000 20 (rather then 33) 1000 6000 17 (rather then 34) 500 5000 10 (rather then 36) Why is there the artificial change of the curve after 3000? What's the logic, apart from giving the very high ranked an easy way of maintaining their rating? Someone can explain?

Let's look at the case of a 500 vs 5000. If we went with the original points by your calculation, the 5000 would win 3 rating for each win, and the 500 36. Now let's work under the assumption that battle rating in this system is at least somewhat accurate. That means that if this 500 and 5000 would play a lot of games, they would end up with a similar rating at the end. However, for the 5000 to maintain their rating, they would have to win 12 games for each game the 500 wins. Given the randomness of 1v1 games, even if the 5000 player plays almost perfectly, I think that is really hard to keep up, The altered system attempts to fix these extremes, meaning that the 5000 only has to win 3 games for every loss, which is probably more inline with expectations on 1v1 games. Important note: I did not implement this system, and I am not trying to speak in favour or against it, I am merely answering the question KaiserKnud raises as to why this was chosen.

[quote=Cireon]However, for the 5000 to maintain their rating, they would have to win 12 games for each game the 500 wins. Given the randomness of 1v1 games, even if the 5000 player plays almost perfectly, I think that is really hard to keep up[/quote] This is the very reason I do not do 1v1, certain game types (Capital), and honestly, tournaments that do not have a minimum entry requirement. I have found out the hard way playing tournaments that I have to win say 10 of the 30 games to break even...which is almost impossible when there are 32 competitors.

Hi Cireon, I understand - and fully agree - that the 'unaltered' 1v1 rating system is not good. However I find the altered system not good either and quite unfair and arbitrary. Why should a 5000 rated player lose less than a 3000 rated player when losing against a 1000 rated player? I don't think this makes sense. And why put the cut-off at 3000? Why not 6000? Or 2000? The case you describe with 500 vs 5000 is absolutely correct. But a single case does not prove the rule is correct. It should be fair in all cases. I don't think it is. The side effect of this is that very few high rated players still play 1v1 with 2000-3000 rated players, as they have much less to lose when playing lower ranked players. Yes, they will win less if they win, but statistically it's the better choice for them. KK

If you have a suggestion for a better formula, I'm sure people are open to hearing about it.

I think we all agree that part of the outcome of 1v1 is determined by the drop/dice/starting position, and the other part by the player skills. Assuming this part is 50%, half of the 20 points to be split should be fixed (luck) and the other half variable (skills), which would lead to the following formula and results list rating[p] = -(10 +floor((p.rating/avgRating) * 10) Winner Loser Won/Lost 6000 1000 12 5000 1000 13 4000 1000 14 3000 1000 15 2000 1000 16 1500 1000 18 1000 1000 20 1000 1500 22 1000 2000 23 1000 2500 24 1000 3000 25 1000 4000 26 1000 5000 26 1000 6000 27 So basically we are narrowing the outcome range from 5-34 to 12-27. It does this for all players, not just for 3000+ players. The formula reflects what is assumed to be luck/skill repartition, rather than using the 'normal' formula with some arbitrary fix for 3000+ players. Of course the 50% is debatable, but once agreed it's just a matter of adjusting the two 10's in the formula. Seems logic?