Saying Hi

[Stone]

Hi all, I found this forum as a bit of a curiosity and have an interest (sometimes obsession) with programming and have recently been looking at lottery numbers though I don't know much about statistics. I'm from the UK and have been playing around with the complete UK lottery history.

Stone

 

[foreverbob]

Welcome!!
Hope you'll find some inspiration and knowledge here.
I sure did. What is it you like to program?

Best regards,

Bob

 

[Stone]

Thanks, programming wise, a lot of software rendered 3d stuff. Not much call for it now but I like the challenge.

Stone

 

[blitzed]

Thanks, programming wise, a lot of software rendered 3d stuff. Not much call for it now but I like the challenge.

Stone

Hi Stone, welcome to the forum Cool that you know how to code 3d stuff...ages ago I used to be into raytracing & fractals...I did not know how to code that stuff, but was into it early on whenya had to enter cryptic commandline parameters or complex scene files to render...also used stuff such as Fractint, Apophysis, ChaosPro, UltraFractal, Fractal Design detailer, and heaps more on the old Comodore Amiga computer.

Ah, you are in the UK...I recently got a true random device which was made there...if you use Linux, you might consider getting one to play with for coding your own lottery stuff to roll your own picks & stuff at home too
http://www.entropykey.co.uk/

I am a programming/lottery nerd, made a buncha stuff like this:
http://crazynuts.hollosite.com/my_lottery_quick_picks/

cheers!
blitzed

 

[Stone]

I've done a bit of raytracing coding everything from scratch and building the scenes in code but not done much with fractals. Interesting device there but I'm on windows and am just relying on whatever the rangom generator the language I use uses though I do a bit more than using it's output to give strings of 6 numbers.

Say I want to generate 15 random lines I seed the rng with the timer and use an array of something like 200 lines*6 numbers.
I then select a random position in the array and check that it's empty, if not select another random position until I find one that is. Choose a random number and check that it's not already in that line, if it is then choose another and then fill that space, repeat all that until the whole array is filled.
At this point I can apply any filtering if I want and remove unwanted lines from the list if of 200 and then randomly select 15 lines from what's left.

I know it's still dependant on the rng and that with the same seed it'll give the same results each time so isn't truly random (although is dependant on filtering and array size). If anyone else has a preferred method of doing this I'd be interested in seeing it.

So I've been doing stuff like that and testing against the UK history with various filtering, sometimes making sure all numbers are covered, sometimes including key numbers or re-using numbers from the previous draw, selecting completely new numbers each draw, sticking to the same numbers or modifying for numbers that have come up and found that all of that makes absolutely no difference to the results either good or bad.

So where to go on from there?

Stone

 

[blitzed]

hiya Stone, fractals can be fun, zooming into near infinity, tweakin palettes...you might try playin with these freebies:
http://www.apophysis.org/
http://chaospro.de/
http://www.povray.org/
http://www.blender.org/

yup, built-in pseudo-random can be kinda funky, but can add more logic and shufflign routines to shake out better results...I hear windows has some similar entropy collection...on Linux there is an add-on program called Haveged which checks changes on hardware interrupts, another called RandomSound actually uses soundcard to record dirty low quality noise.

I used to run some pseudo-random then shuffling logic, and randomly grab RGB color values from coordinates in some pics...especially two I have which use the full 16.7million color spectrum logarithmically, then shuffle that around through another routine...I never ran it against the gauntlet of random number testing programs, but seemed pretty random to me, but I liked the novelty of knowing I had my own routine to shake out numbers...here are two example pics I had code readin values from...pics look strange, do not look like they hold all the colors, but they are in there.
http://crazynuts.hollosite.com/pics/000_latest_uploads/sp_index.php?file=./RGB16M.png
http://crazynuts.hollosite.com/pics/000_latest_uploads/sp_index.php?file=./RGB16M2.png

what size game history sample do you work with? I usually try to isolate some probabilities which are not too far out, and just check practical intervals of game history and try to exploit.

cya,
blitzed

 

[Stone]

I'm not looking into predicting or isolating probable numbers just now, more seeing how different ways of choosing random tickets could affect performance so I use the entire history of 1667 draws to test against.

One of the things I'd been looking at for filtering was the number of odd and even numbers that are drawn and I find this a bit strange.

In draw order

OOOOEO has occured the most at 38 times
OEEEEEO has occured the least at 15 times

In numerical order

OEOEOE has occured the most at 37 times
OOOOEO has occured the least at 12 times

Is such a big difference between the least and most occurences to be expected? And how does this compare to other 649 lotteries?

 

[Stone]

that should have been

In draw order

OOOOEO has occured the most at 38 times
OEEEEO has occured the least at 15 times

In numerical order

OEOEOE has occured the most at 37 times
OEOOOO has occured the least at 12 times

 

[blitzed]

hiya Stone, cool...I have not looked much into that...it is nice to get the raw draws, then you can play the unique probabilities of each ball-slot.

for probabilities, I look shorter term...since over time ballsets & draw machines retire...so I would use somethin like this to play percentages...chop x percentage off each end of the spectrum and play closer to median hits.

Any 6/49 number drawn X times within 76draws | Probability
============================================
0x 0.00%
1x 0.05%
2x 0.27%
3x 0.93%
4x 2.37%
5x 4.77%
6x 7.88%
7x 10.99%
8x 13.23%
9x 13.95%
10x 13.04%
11x 10.92%
12x 8.25%
13x 5.67%
14x 3.56%
15x 2.05%
16x 1.09%
17x 0.54%
18x 0.25%
19x 0.10%
20x 0.04%
21x 0.02%
22x 0.01%
23x 0.00%

later,
blitzed

 

[blitzed]

and rough probability per ball-slot...I would have to do 1of48 and so on...
1of49 481Draws
0x 0.00%
1x 0.05%
2x 0.25%
3x 0.82%
4x 2.05%
5x 4.06%
6x 6.72%
7x 9.50%
8x 11.72%
9x 12.83%
10x 12.62%
11x 11.26%
12x 9.19%
13x 6.90%
14x 4.81%
15x 3.12%
16x 1.89%
17x 1.08%
18x 0.58%
19x 0.29%
20x 0.14%
21x 0.06%
22x 0.03%
23x 0.01%
24x 0.00%

 

[Stone]

Am I getting this right?

The table can be made by taking a sample draw and counting the frequency of each of the 6 balls over the previous 73 draws and incresing the values in the tables based on the 6 frequencies and then repeating for many samples?

Then from that a number that has appeared 8-11 times in the previous 73 draws has a higher probablity of appearing?

I've just done that for the 43 numbers that didn't appear instead of the 6 that did and get almost exactly the same table so that numbers that have appeared within particular frequencies have a high probablity of both being drawn and not being drawn.

 

[blitzed]

Am I getting this right?

The table can be made by taking a sample draw and counting the frequency of each of the 6 balls over the previous 73 draws and incresing the values in the tables based on the 6 frequencies and then repeating for many samples?

Then from that a number that has appeared 8-11 times in the previous 73 draws has a higher probablity of appearing?

I've just done that for the 43 numbers that didn't appear instead of the 6 that did and get almost exactly the same table so that numbers that have appeared within particular frequencies have a high probablity of both being drawn and not being drawn.

Hi Stone, the probability chart was generated via binomial distribution distribution function in a spreadsheet....such data can be crossref'ed with your frequencies...sometimes the bulk of probability will continue to move and you can get in on it

later,
blitzed