By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - Website Topics - Who won the Feb 4 Japan Chart prediction league?

Error: 90000-49750 = 40250 % error: 40250 / 49750 x 100 = 80.9% % accuracy: 100% - 80.9% = 19.1%

Around the Network

ok, that pretty much means that if you over-predict by 2x you get 0% accuracy. I guess according to the calculations that is equal to saying the sells are at 0 when under-predicting. Though it is alot more ridiculous to say something is gonna sell 0 copies :) Guess I will adapt and lower the numbers on everything a bit for next week :) Edit: And it does mean that unfortunately my cooking mama prediction will be 0% :p

Well you have the actual value plus or minus itself and you have to fall within that range to score points. You get as penalised if you are 10 over as if you are 10 below - this is really the only fair way of doing it i think. We could do a proportional difference but then that's not fair. To look at it the other way round - if you predict 30k and someone else predicts 50k and the actual is 40k you'd expect to have the same accuracy, no? You've both got the value to within 10k. I think the system does favour underestimating of course, but in reality it will favour those who are consistently within say 10-15% of the actual value. A hint (although of course only based on our own estimates) is that the order is generally our proposed order so you probably want to have predictions that are roughly descending as you go down the list - this should generally give a guide to the ballpark although you definitely don't have to follow what we think - last weeks data shows we were off on a couple of titles....

what a shame only 46 percent lol.