Something I meant to ask about before. Now I understand the ELO system and as I'm at the top of the league table my win/lose for most games is 5/14. That is against one player. That's essentially saying I am 5/14 favourite to win. I'm basically gambling 14 points in order to gain 5.
However when I play against 3 other players it stays the same. Now I'm no maths expert by any means but I have fair understanding of odds and it shouldn't be the same. If I'm playing against 5 players, or 10 players would it be the same? No of course it shouldn't , so it shouldn't be for 3.
If my ELO rating is 5/14 with each of the three opponents then that is 5/14 x 3 = 15/14. My odds are just above evens. I should gain 15 points if I win and 14 if I lose.
At the moment playing a 4 player game risking 14 points for just 5 is ridiculous odds.
Sorry to hijack your thread xeno, I also have a points related query. Can one's overall pts go negative? I've played around 14 games and lost all (yay). Just wondering if I'm still at 0 or -3 million or something
@Xeno, at the moment it works like this for 3 or 4 player games: It calculates the average league points of the winners and the losers, and then calculates the difference between the two numbers. It then does the following calculation (if D is the difference):-
If D is negative the winners get 5 points. Otherwise the winners get "20 x (0.5 + (D/500))".
(And the reverse happens for the losers). Don't ask me how I came up with this calculation! ;) I don't even remembering writing it; I think it's the result of lots of tweaking after implementing the ELO algorythm, several years ago now. Feel free to request a change though.
Right, most of that has gone over my head Steve, all I know is that I'm currently in a non-practise 4 player game and tbe odds of 5/14 for me to win are unrealistically very short and not worth the bet whatsoever.
After thinking about it, I think the flaw is that it doesn't take into account whether the game is 2v2, 1v1v1 or 2v1. I'll do some calculations and see the best way to rectify that.
Okay Steve, but are the odds that someone being the top is 5/14 when taking on three other players right? I can understand the ELO system against one, but surely it must change playing against three. It can not stay the same. If a horse (I come from Gloucester; Cheltenham festival) is 4/15 against one runner, if you enter three others then my odds must change. In this game it doesn't.
Sorry that is 5/14, almost 1/3. Fair enough if you are top form, but why does it not change between one, and then three other "runners"? I find this in every four player game I play if it isn't practise?
That's right; the calculations currently don't take into account playing against more than one other opponent, which it should since the game is presumbly harder against 2 other opponents.
"That's right; the calculations currently don't take into account playing against more than one other opponent, which it should since the game is presumbly harder against 2 other opponents."
Any plans therefore Steve to make the calculations do take that into account and make it a fair representation of the "odds", the point win/loss ratio?
Yes, I'm just working on it. I've got a spreadsheet loaded as I write this. :) The good thing about the points system in SF is that it calculates it all retrospectively, so any change will affect the points for all past games as well as future games.
Cracking stuff. Hope it doesn't take you too much effort or time Steve, but I think it is an important adjustment. Will be very interesting if it does it retrospectively. Won't affect my points too much as I've been aware of the 3/4 player point anomaly, let us say, for a long time now and have tended to only play and create practise games as a workaround to it.
Just to let you know, I haven't forgotten about this! I've come up with some calculation, and I'll do an update tommorrow evening (so I can tweak it if need be).