AndrisBirkmanis wrote:
z1235 wrote:
Glad you finally agreed that things don't have to be 100% certain to be predictable, i.e. that smaller then 100% predictability does not imply UNpredictability.
For the benefit of the rest of us, what exactly is meant here?
A. One does not have to be 100% certain which way the die will fall in order to predict its value will be from 1 to 6? (perfect prediction of a general outcome without perfect prediction of a specific outcome).
B. One does not have to be 100% certain which way the die will fall in order to predict its value will be from 2 to 6? (imperfect but often true prediction).
C. Something else completely?
C.
To clarify my terms, a fair die (with 1/6 chance of falling at any number 1-6) is both perfectly uncertain and perfectly unpredictable. One does not have to have a perfectly biased/certain ("cheater") die (e.g. one designed to always fall on #4 with 100% certainty) in order to profit from a biased die. The die could only be biased to fall at #4 with 1.02 * 1/6 chance (i.e. 2% more likely than "fair") and you could still come ahead (profit-wise) if you bet #4 on large enough number of throws. In this example, OBF would say that betting on #4 is unpredictable because it is uncertain (i.e. you don't know
with certainty which number may show at any particular throw/bet). I'm saying that, even though it's not certain, betting on #4 is not completely unpredictable and that there exists a statistical edge which can be exploited over a large enough number of throws/bets. The fact that I can not predict with certainty does not mean that I can not predict (i.e. can not profit from predicting) at all.