Currently Online 
| kirby007 said:
Your Winnings Winnings are determined by your prediction accuracy. Prediction accuracy for a each prediction is calculated as the difference between your prediction and actual figure / actual figure. The prediction accuracy for each prediction is divided by the average accuracy across all users for that prediction and then this ratio is multiplied by your bet on that prediction. This is totalled up over the 15 predictions. An example is the best way to illustrate. Example: You predict a game will sell 175,000 copies for the week and the game actually sells 200,000. So, we take the difference between your prediction and the actual sales and divide it by the actual sales 25,000 / 200,000. The result, 0.125, represents the percentage your prediction was off from the actual (12.5% off). Subtracting this number from 1 (one represents being wholly accurate) tells us how close or accurate your prediction is to the actual: 1 - 0.125 = 0.875 or 87.5% accurate. Note that if your prediction is off by more than the actual amount (i.e., off by more than 100%), your accuracy will be 0% representing the fact that you were simply not accurate in your prediction (e.g., you predict 100,000 and the actual is 30,000). Your accuracy is then divided by the average accuracy for that prediction, in this case say 50% (100,000 as the average prediction) to give (87.5/50) then multiplied by your original bet. So if you bet vg$10 on this prediction, you would win 10 x (87.5/50) = vg$17.50 (a net profit of vg$7.50 against the original bet). This is repeated across all your bets and totalled up. |
Hmm, if I understand this right, we all (well most) were saved by fedfed, whose bets messed up the average, so that pretty much each other prediction was better than this average. So big thanks to fedfed, who took the fall for all of us. 
Hmm, with this live example I understand much better how it all works. So I should put in a high bet, if I think the average is WAAAAYYY off.
A thing is though, you should consider using the median instead of the average. With the median single outliers don't affect the result much. On the other hand, this would make the prediction league very hard, so maybe keep the easy mode with averages.
