Re: what is the best/state-of-art lottery prediction algorithm in statistics?

mensanator_at_aol.compost
Date: 01/04/05 

Date: 3 Jan 2005 19:10:31 -0800

pie wrote:
> hey
> can u teach me how did u do this ? plz . i am intrested in this ?
> and what major did u choose for this field?

Who are you replying to? Surely you can't be asking to
be taught how to lose money so fast it makes your head spin?

Did you notice from kiki's post that a box of 7 numbers
cost 42 units and if you match 3 you win 36 units?

Of course, you could always increase your box to, say,
11 numbers. That would give you 28 winning cards for a
payout of 168 units. But then, to box 11 numbers you have
to play 462 cards at a cost of 924 units. Increasing the
box size increases your chances at losing a bigger amount
of money.

If I extend my 7 number box to 11 numbers and compared them
to the 30 events given, there would be

7 match 3's for 1,176 units
15 match 4's for 12,810 units

for a total win of 13,986 units.

At a cost of 27,720 you would "only" be down 13,734.

Actually, I counted the two match 5's as match 4.
So wouldn't the "millions" you get from the match 5
make up for all your accumulated losses?

There are exactly 26,334 ways to select 22 items 5 at
a time. There is no way the match 5 payout can be
millions. It would only take 52,668 units to box all
22 numbers.

Finally, keep in mind that I am _not_ predicting the
future, I am predicting the past. If the numbers are
truly random, all the correlations of the past occurences
are meaningless.

To answer your last question, I whipped up that analysis
in about an hour using MS-Access (you could probably use
any SQL database). If you really want to know how to box
numbers, I could post my database design.