Stuart wrote:
>misscraps@... wrote:
>> Played some single line DDB at Palms tonight and sure enough,
>> lots of quads. It reconfirmed my feelings that the NCL Epic
>> VP didn't just have bad pay tables, but was gaffed.
>
>Playing single line, and accurately recording the number of hands vs number of quads, how long would you have to play in order to have even a 75% confidence that your results reflect the expectation of the machine, rather than random short-term fluctuation? There must be math experts here who know exactly. Personally, I'd guess it'd take much more than one day.
>
>If not recording the quads and hands played exactly, but relying on feelings, I'm not sure that anyone could ever get a reliable conclusion re whether a machine is a little gaffed.
>
>Stuart
>http://stuart-randomthoughts.blogspot.com/
If it's possible, how long it would take depends on how much it's
gaffed. But I'm not sure there's any way to tell that a machine is
gaffed using trials. How far results deviate from expected, as per a
particular theory, can be measured, but that's not enough. If one's
results deviate from expected, based on the theory that all cards have
an equal chance of occurring, by 8 standard deviations, is that an
indication that the machine is gaffed? How about 3 standard
deviations? 2? Where is the line at which the probability that the
machine is gaffed exceeds 50%? It's still a guess. If being "off" by
8 standard deviations is an indication that the machine is gaffed,
isn't the result of every hand some indication that the machine is
gaffed?
