Welcome Guest
Log In | Register )
You last visited January 17, 2017, 12:05 pm
All times shown are
Eastern Time (GMT-5:00)

Lottery Prediction by Recurrence Analysis

Topic closed. 170 replies. Last post 11 years ago by Raven62.

Page 6 of 12
PrintE-mailLink
NemeSys's avatar - Nemesis
Harrogate
United Kingdom
Member #16589
June 1, 2005
107 Posts
Offline
Posted: August 7, 2005, 9:48 am - IP Logged

I'm going away for a week on Monday, so I thought I'd post this before I go.

I was looking at the VRA analysis of the Thunderball results. The current values of dim and del accurately predicted two out of the five main numbers quite regularly. It occurred to me that if the predictions are good for two numbers, perhaps that's how I should look at the results. To explain: for the main 6/49 draw, analyse columns 1 and 2, then 3 and 4, then 5 and 6. It might provide that oh-so-elusive breakthrough. This doesn't mean I've abandoned other approaches, but I want to test this when I get back. I've also decided to look at the results in ascending numerical order rather than drawn order, it appears to give better results.

Further: the main draw numbers have a fractal dimension! D=1.3090 +/- 0.1204

I really must get some graphics of the attractor worked up when I get back too.

    rowdy117's avatar - image php?u=2413&dateline=1135265132
    Tennessee
    United States
    Member #12466
    March 14, 2005
    29 Posts
    Offline
    Posted: August 7, 2005, 4:46 pm - IP Logged

    Nemesys,

    Thank you for all of your input on VRA.  Everyone appreciates your effort.

    I have worked with this program a bit.  Granted it is free, but I find it very difficult and time consuming to work with.

    It would be nice if you could loop the dimension and delay instead of using a slider bar.  It takes a lot of time to manually

    grind out the results. 

      NemeSys's avatar - Nemesis
      Harrogate
      United Kingdom
      Member #16589
      June 1, 2005
      107 Posts
      Offline
      Posted: August 7, 2005, 5:43 pm - IP Logged

      Unfortunately there's no way to automate the search process from within the program. A batch file might work, but my programming skills aren't up to it. For the time being you just have to go through it manually.

        NemeSys's avatar - Nemesis
        Harrogate
        United Kingdom
        Member #16589
        June 1, 2005
        107 Posts
        Offline
        Posted: August 15, 2005, 6:39 pm - IP Logged

        Hello all, I'm back. At home I dug around through my library of software and maths/physics books and brought back a couple of things I forgot I even had, like the random number generator based on linear congruence, also a little program I wrote to display lottery numbers in interesting patterns (I wrote it years ago on a BBC Master in BASIC, but it can easily be translated to other languages). I'll put it all up on my site, I just have to catch up with a few things in the non-math world first.

          Hyperdimension's avatar - latest trace_171.gif

          United States
          Member #9059
          November 26, 2004
          128 Posts
          Offline
          Posted: August 15, 2005, 7:21 pm - IP Logged

          Welcome back Nemesys,

          Do you know how to see a fractal dimension, I mean how to create the fractal (the image)? Which parameters?

          H=2-1.3090=0.691 (Random H=0.5, if H>0.5 less chaos)

          HDelta=0.1204

          HAlpha=? 

          Ex.1.3090+/-0.1204

          Regards

          El pensamiento ordena el caos..

          http://1x2quinielas.blogspot.com

            NemeSys's avatar - Nemesis
            Harrogate
            United Kingdom
            Member #16589
            June 1, 2005
            107 Posts
            Offline
            Posted: August 16, 2005, 5:47 pm - IP Logged

            To calculate the fractal dimension I use Fractan, available at:

            http://www.tsm-soft.com/category/education/edupro03.html

            After unzipping and running, load in an ASCII file of results, then click on the 'Process' button to the right. Next, click on the Process menu, then 'Hurst exponent'. Save the file, and the values of Hurst exponent and fractal dimension will appear at the bottom of the window.

            To see the attractor I use AVTS, available at:

            http://www.cg.tuwien.ac.at/research/vis/dynsys/avts/

            This one runs in a full-screen DOS window. Again, load data as an ASCII file (you will have to save it with the .asc extension first, use Notepad for this), then choose which views you want from the 'Windows/Visualisation' menu.

            "God is subtle but He is not malicious." - Albert Einstein

            "In my experience there's no such thing as luck." - Alec Guinness, Star Wars Episode IV

              Hyperdimension's avatar - latest trace_171.gif

              United States
              Member #9059
              November 26, 2004
              128 Posts
              Offline
              Posted: August 16, 2005, 5:55 pm - IP Logged

              Hi nemesys,

               

              I downloaded fractan and avts from your site, I have next question, in the avts program in order to see the fractal you need all data of loto and avts calculate fractal dimension or only knowing the fractal dimension 2-H +/- D is enough to see it?

               

              Regards 

              El pensamiento ordena el caos..

              http://1x2quinielas.blogspot.com

                NemeSys's avatar - Nemesis
                Harrogate
                United Kingdom
                Member #16589
                June 1, 2005
                107 Posts
                Offline
                Posted: August 16, 2005, 6:15 pm - IP Logged

                You don't need any extra figures for AVTS, just the data file. The UK main lottery results give a shape like a tetrahedron. This is only part of the attractor since it's only one trajectory, but I'm guessing whole thing has the same basic shape, like the Sierpinski Tetrahedron, some beautiful models of which you can see here:

                http://www.public.asu.edu/~starlite/sierpinskitetrahedron.html

                "God is subtle but He is not malicious." - Albert Einstein

                "In my experience there's no such thing as luck." - Alec Guinness, Star Wars Episode IV

                  Avatar

                  United Kingdom
                  Member #9759
                  December 20, 2004
                  20 Posts
                  Offline
                  Posted: August 17, 2005, 3:35 am - IP Logged

                  hi nemeys, this is a very basic question, ive been playing around with vra, what would you say was the best approach to input the data, one single column with all 5 numbers in one cell, or all the numbers in 1st postion in one column, and to check them a column at a time, and which del dim is recommended, im not having much luck, but i think thats down to my ignorance.....

                  so ie whats the best format for inputing

                   

                  1 15 24 26 34

                  2 16 25 27 33 as one column or

                  1

                  2  as one column....many thanks

                    NemeSys's avatar - Nemesis
                    Harrogate
                    United Kingdom
                    Member #16589
                    June 1, 2005
                    107 Posts
                    Offline
                    Posted: August 17, 2005, 11:17 am - IP Logged

                    The arrangements look different on a word processor, but VRA treats them the same, namely as a continuous string of numbers. However, if you want to analyse each column separately then they do have to be loaded as individual files. You can extract single columns of numbers from the Excel file on my site (which I still haven't had time to update recently, I've a job interview Thursday, so early to bed tonight...).

                    As to which values of dimension and delay are the best...I honestly don't know for sure. It appears that in fact there are no single 'best' values, but a small range of values which give reliable results (and by reliable I mean two or three accurately-predicted numbers on a fairly regular basis). All you can do is go back over previous results, select a starting point, then work forwards from there to find values of dim and del which give consisent results. As other people here have discovered it's a time consuming process, but there's no other way to do it (yet).

                    What I generally do is start a hundred results back from the current draw and number-crunch upwards from there. The EuroMillions draw only has about 80 results so far, so it will be interesting to see if recurrence analysis will work reliably on a small sample.

                    "God is subtle but He is not malicious." - Albert Einstein

                    "In my experience there's no such thing as luck." - Alec Guinness, Star Wars Episode IV

                      NemeSys's avatar - Nemesis
                      Harrogate
                      United Kingdom
                      Member #16589
                      June 1, 2005
                      107 Posts
                      Offline
                      Posted: August 20, 2005, 11:37 am - IP Logged

                      I'm having problems with my ntl connection at the moment (I'm sending this from the local library). I don't know when they'll sort it out, only a few days I hope.

                      "God is subtle but He is not malicious." - Albert Einstein

                      "In my experience there's no such thing as luck." - Alec Guinness, Star Wars Episode IV

                        Avatar
                        New Member
                        London
                        United Kingdom
                        Member #18841
                        July 21, 2005
                        8 Posts
                        Offline
                        Posted: August 21, 2005, 11:41 am - IP Logged

                        Hi Nemesys,

                         

                         

                        I have several queries. Firstly, with regards to the results that you have received with Thunderball, although the ‘best’ delay to use seems to be mostly 4 or 5, with regards to dimension, the numbers you use range from between 10 to 20 with no one number appearing more dominant than any other. Therefore, this can be interpreted either as there being a pattern, more specific with ‘delay’ but less so with ‘dimension’, likewise one can see your results as specifically selected for matching some numbers post-draw. The question therefore is whether there is really a pattern or whether it is just wishful thinking (isn't the human brain supposed to have an inclination to 'seeing' patterns in everything)?

                         

                         

                        However, what I have noticed is that often when the predicted and actual results do not correlate, the actual result is often fairly near to the predicted result. It might therefore be possible to use a wheeling system whereby you employ, say 5 numbers on either side of (and including) the predicted number and wheel those to produce lines that may be more likely to hit (as the number is likely to be a few points off)?

                        Keep up the good work - its very interesting.

                          Avatar
                          New Member
                          London
                          United Kingdom
                          Member #9904
                          December 27, 2004
                          12 Posts
                          Offline
                          Posted: August 21, 2005, 2:04 pm - IP Logged

                          The usefulness of RP is it tells you what numbers will not come up column by column. If you divide the 6/49 lottery into patterns then it tells you which patterns will not be forthcoming in the next draw or the next few draws. In other words it is very good at anti-predicting. For example randomly delete 10 numbers. Then use RP to anti-predict 10 numbers. I tend to delete 2 of the winning numbers while RP anti-predicts more successfully than me. Nemesys has to be congratulated in presenting this way at looking at the lottery. 99.9% of what is written on lottery boards is complete rubbish and his contribution is a real nugget.

                            NemeSys's avatar - Nemesis
                            Harrogate
                            United Kingdom
                            Member #16589
                            June 1, 2005
                            107 Posts
                            Offline
                            Posted: August 22, 2005, 12:15 pm - IP Logged

                            Infinitysymbol: you're right about dimension. Of the two parameters it's always the harder one to pin down (there's tons of theory about finding the best value for a given data set). Regarding 'post-draw'predicting, it's the only way to do it. By building up a table of values for dim and del you can then begin to see (hopefully) which ones to use for real predictions. As for patterns, yes, the human mind is especially good at finding them where they don't necessarily exist. But VRA is just a set of equations and algorithms; it has no bias, no prejudices, no desire to be right. The patterns it discovers are purely the result of mathematics, and so are more likely to be genuine than anything we see.

                            Also, as you discovered, VRA often comes close to the real (drawn) numbers, so wheeling is probably a good way to offset the near-misses.

                            Zulu:  I never thought of looking at 'anti-predictions'! A novel way of using RP, to say the least. Thank you for the praise, I'm flattered!

                            From now on I think I'll concentrate on two ways of looking at draw results: by column, and by full draw results (i.e. all five or six numbers at a time). Also, I will look at results in ascending numerical order. All the evidence so far indicates it gives the most consistently reliable predictions. Enough people out there now have VRA and know how to use it to test their own approaches. Time to update my site too. So much to do, so little time...

                            "God is subtle but He is not malicious." - Albert Einstein

                            "In my experience there's no such thing as luck." - Alec Guinness, Star Wars Episode IV

                              Avatar
                              New Member
                              London
                              United Kingdom
                              Member #9904
                              December 27, 2004
                              12 Posts
                              Offline
                              Posted: August 22, 2005, 2:09 pm - IP Logged

                              Try this to test RP. Generate a 6/10 lottery of say a thousand combinations. Apply RP to predict. It struggles even with this simplified lottery(+- 220 possible combinations). Test column by column and then on an all six combination basis. I am pretty sure the Maths wizards at Oxford, Cambridge or wherever have had a go with their Cray supercomputers. Or perhaps they don't research lotteries as it is too vulgar for those chaps and they could be kicked out of academe. As a matter of interest Dimension 59 and delay 25 were pretty close with the 5 numbers of last Saturday's UK draw. I used all 6 columns simultaneously for the draw which included every draw in the lottery.

                                 
                                Page 6 of 12