you could make the Winning divisible by the amount of weeks that a game has been out... like the winnings(or losses) for NSMB would be divided by 100 weeks or whatever... it is pretty pointless to pick those kinds of games anyways so this makes it so the max you could get off of a game like WiiFit would be like 80 $ or so... new releases would be a full 1.0 multiplier..

Or you could just set up a preset multiplier X how long it has been out. If you did the 90% thing like Mneth said you have to many decided each week for each game. This way you could use a simple forula

Bet/(weeks out X muliplier you create) = winnings....

I'm sure you guys will figure something out.

I also like Mnementh's idea. Well thought out.

I say we wait and see how it turns out before making more changes. We haven't even through our first predictions yet.

when will the rounds payoff be? a week, 2 days?? after the numbers are up

ioi said: I quite like Mnementh's idea - I am going to look into it and see. This way the average prediction becomes what you are compared to so if you are spot on and everyone else is miles away you get the best payout, if everyone is close then not so. I like it.

Nice :o)

On the other hand, it could be more fair, I think the idea was also his :) :

The total ammount off cash bettet on one item (lets say Wii) goes in a pot, and payed out by accuracy:

Lets take a game called X:

Guy 1 betted 50$ on 100k

Guy 2 betted 100$ on 50k

Guy 3 betted 75$ on 75k.

The pot size is 225$. Lets also say you give a 5% on top "bonus", the pot size is  236.25$

The game turned out to be 70k.

We know calc. The accuracy and pay out by put in potsize:

Guy 1 has a acc off: 1-30/70= 57%*50$ = 28.5 points

Guy 2 has a acc off: 1-20/70= 71%*100$= 71 points

Guy 3 has a acc of 1-5/70 =  93%*75$ = 69.64 points

Thats a total of  169.14 Points or 236.25/169.14 =1.40$/point.

Ergo :

Guy 1 gets  39.9$ (lost 10.6$)

Guy 2 gets 99.4$ (lost 0.6$)

Guy 3 gets  97.50$ (won 22.5$)

So you win more if you bet on something others are inaccurate, but the system itself doesn't generate money in that high way the current system can do. (5%/week of bets).  

Halve the payout each week. So a first week bet on a game is 1:1, 2nd week 1:2, 3rd week 1:4, etc.

Multiple the winnings by the inverse of the percentage of error, i.e., a prediction of 90k for an actual 100k would be a multiplier of .90.

So a person bets $100 on a first week game, predicts 200k, actual number is 210k. Payout is $95. For second week the same amount bet with same number predicted and sold would net $47.50.

With consoles rank bettors and have a staggered payout system based on rank.

I guess you could fiddle around with a lot of different systems.