Welcome Guest
Log In | Register )
You last visited December 6, 2016, 12:39 am
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 7 of 12
PrintE-mailLink
NemeSys's avatar - Nemesis
Harrogate
United Kingdom
Member #16589
June 1, 2005
107 Posts
Offline
Posted: August 22, 2005, 2:27 pm - IP Logged

You'd be surprised just who does lottery research! Here's some work from the University of Southampton:

http://www.hpcc.ecs.soton.ac.uk/lottery_f.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
    New Member
    London
    United Kingdom
    Member #9904
    December 27, 2004
    12 Posts
    Offline
    Posted: August 22, 2005, 2:31 pm - IP Logged

    I meant to add Volcano and Lyapunov colour schemes give a pretty regular colour of the parameters I gave.

      NemeSys's avatar - Nemesis
      Harrogate
      United Kingdom
      Member #16589
      June 1, 2005
      107 Posts
      Offline
      Posted: August 22, 2005, 5:18 pm - IP Logged

      Hello Zulu, I tried the values you mentioned with the Lyapunov colouring scheme. I think the Lyapunov colouring shows up details that the others tend to swamp. Phase space portraits are often filled with more structure than the eye can see at first glance. Fractals coloured with the sepia tones of Lyapunov rather than by a more typical 'rainbow' spectrum often look clearer and more striking.

      "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 22, 2005, 9:27 pm - IP Logged

        Site update: placed three links at the top of the main page, added four more PDFs to the research papers list, added a link to the University of Southampton lottery work, and a new sub-page about visualizing lottery numbers:

        http://zarnia.250free.com/visual.htm

        Now I can back to analysing actual results.

        "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

          Hosaka's avatar - glider
          New Member
          Seattle, WA
          United States
          Member #20689
          August 23, 2005
          2 Posts
          Offline
          Posted: August 23, 2005, 3:50 pm - IP Logged

          NemeSys:

          This work is intriguing. I have gone over your webpage, but never came away with a clear sense of how to read a recurrence plot. Which is better - the smaller magnitude, red end of the spectrum or the larger blue? When a checkerboard structure appears, why is the number of squares always an integer number within the time series? Are there ever parameters which only show partial structure at the largest - or smallest - values in our time series, thus displaying a partial square? Is the checkerboard the most elucidating structure we should seek? Should I read along the diagonal or the axis in reference to determining high correlations in time? Et cetera.

          This is certainly not an attack, just a few questions generated by my observations. And in searching for clarification, I found the following .pdf file which is easy to read and highly informative:

          http://www.jstatsoft.org/v07/i09/JSS_055.pdf

          I have been following the thread and thought I would offer that bit of reading. I am currently attempting to add it to my schedule of reading and look forward to continuing, and hopefully contributing, to the results here.

          -H

          Edit: Ha! I just saw you have the above paper on your website. However, I cannot load it. Some of the linked files have improper formatting.

            NemeSys's avatar - Nemesis
            Harrogate
            United Kingdom
            Member #16589
            June 1, 2005
            107 Posts
            Offline
            Posted: August 23, 2005, 6:20 pm - IP Logged

            The coloured plots are there only to give a broad indication of the time series structure (or indeed if there is structure at all). The graph is really what you need to look at for actual predictions. But in answer to your queries (I'll assume you're using VRA):

            1) The colour spectrum seen at top right corresponds to actual numbers. For a 6/49 lottery the white end corresponds to 0, the black end to 48 (it goes from 0 to 48 rather than 1 to 49 because that's just the way VRA was programmed, but it's a very minor niggle).

            2) Why is the number of squares always an integer? The data is processed in such a way that it forms squares. VRA turns numbers into vectors and plots these at right angles to the main diagonal; numbers get represented along both the vertical and horizontal axes; result: squares, every time.

            3) Can you ever get a partial square? No, at least not with lottery results (though other sets of data do give asymetric patterns, but even they often appear bounded by squares).

            4) Is the chessboard pattern the best? Well, it's the only one that ever comes up with lottery results (again, other types of data give different patterns). I'd go so far to say that the chessboard pattern is actually characteristic of lottery results.

            All the PDFs should load with no problem if you have Acrobat installed on your PC. If you really get stuck with one you want I can always e-mail it to you (I might store them as Zip files in future to avoid the download problem). Thank you for your offer of help, too.

            "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

              Hosaka's avatar - glider
              New Member
              Seattle, WA
              United States
              Member #20689
              August 23, 2005
              2 Posts
              Offline
              Posted: August 23, 2005, 7:24 pm - IP Logged

              The coloured plots are there only to give a broad indication of the time series structure (or indeed if there is structure at all). The graph is really what you need to look at for actual predictions. But in answer to your queries (I'll assume you're using VRA):

              1) The colour spectrum seen at top right corresponds to actual numbers. For a 6/49 lottery the white end corresponds to 0, the black end to 48 (it goes from 0 to 48 rather than 1 to 49 because that's just the way VRA was programmed, but it's a very minor niggle).

              2) Why is the number of squares always an integer? The data is processed in such a way that it forms squares. VRA turns numbers into vectors and plots these at right angles to the main diagonal; numbers get represented along both the vertical and horizontal axes; result: squares, every time.

              3) Can you ever get a partial square? No, at least not with lottery results (though other sets of data do give asymetric patterns, but even they often appear bounded by squares).

              4) Is the chessboard pattern the best? Well, it's the only one that ever comes up with lottery results (again, other types of data give different patterns). I'd go so far to say that the chessboard pattern is actually characteristic of lottery results.

              All the PDFs should load with no problem if you have Acrobat installed on your PC. If you really get stuck with one you want I can always e-mail it to you (I might store them as Zip files in future to avoid the download problem). Thank you for your offer of help, too.

              Hi NemeSys,

               

              Thanks for responding. Yes, I am using VRA.

              1) The spectrum boundaries do correspond to the limits of the actual dataset entered when using values embedded dimension=1 and time delay=1, but what then do they represent when those are changed? The upper bound seems to shift drastically with changing dimension and delay. I understand it is the Euclidean distance away from the actual value, but how can that be interpreted?

              3 & 4) I don't understand why the fact that lottery results require that a full square is produced in the plot when structure is seen. Are we seeing this structure simply because of the symmetries on each axis? Our dataset is plotted on both the X and Y axes of the program. Is there perhaps a way to enter a different dataset on an axis - say the range of the dataset one is analyzing? Does that even make sense in this kind of analysis?

              My issue with the PDF files is not the absence of Acrobat, rather it is that I receive an error from Adobe Reader:

               "File does not begin with '%PDF-'."

              This does not occur with all of your files, nor does it occur anywhere else in my local files or online that I have found. Have you double checked the download yourself?

               

              -H 

                NemeSys's avatar - Nemesis
                Harrogate
                United Kingdom
                Member #16589
                June 1, 2005
                107 Posts
                Offline
                Posted: August 23, 2005, 9:30 pm - IP Logged

                Changing the dimension and delay alters how the recurrence model looks at the data. The plot is a two dimensional view of what's going on in a higher dimensional phase space, which of course we can't see if it's higher than three anyway. The 2-d plot essentially tells you how closely related the vectors are. As dim and del change so do the numerical values connected with the colours (the lengths of the vectors are changing).

                There's no way to change the axial lengths - whatever appears on the x-axis is mirrored on the y-axis perfectly (so you always get a square). Think of it as taking a right-angled triangle with adjacent and opposite sides of equal length, then mirroring it along the hypotenuse - you get a square.

                I've tested the PDFs, and apart from differences in loading time because of size they all download perfectly. I'm using the current version of Acrobat and I have a fast broadband connection, so I really couldn't say what the problem is, other than a file getting corrupted in transit somehow. All I can suggest is try loading it again (or ask me to e-mail it).

                "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

                  paurths's avatar - underground
                  Switching between Fairfax, VA and Belgium
                  Belgium
                  Member #19287
                  July 29, 2005
                  2254 Posts
                  Offline
                  Posted: August 24, 2005, 4:14 pm - IP Logged

                  Well, i"ve been testing VRA for several weeks now and it pretty much got me to the nuthouse.
                  Dim & Del work for about 2 draws, then it shifts, the dims & dels become pretty much useless, and i'm in pursuit of the "new best dimensions and delays". 2 draws later it shifts again.

                  Also it is very time consuming. One way or the other, it would be just magnificent if the calculations (for ALL dimension-delay combinations) could be automated. Then the results could simply be written to a DB-structure where crosschecking with actual draws would be an automated process. (the latter part is no problem, i have that covered)
                  I've been reading up on all this stuff the past weeks and the formula's pretty much got me in a permanent dazzling state of mind.
                  There's more potential to it, i think, when this last scenario could be reality. Once the formula's have been implemented (and i'm breaking my head over them, to no avail) the rest of the software (crosschecks for one) would be less dificult.

                    NemeSys's avatar - Nemesis
                    Harrogate
                    United Kingdom
                    Member #16589
                    June 1, 2005
                    107 Posts
                    Offline
                    Posted: August 24, 2005, 7:24 pm - IP Logged

                    The time-consuming aspect I can live with. It's consistent results that need pinning down. Automating the search would speed it up, but you would still have lots of results to go through.

                    Somewhere there might be recurrence analysis software which does cycle through values automatically and implement other features; I'm always looking for new programs that implement RA, but so far there doesn't seem to be much (a few very interesting programs I've seen only run on other platforms - I've been tempted to install LINUX more than once!)

                    I don't want to say "we're at a dead end", but what we have so far is the best we have so far.

                    "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

                      paurths's avatar - underground
                      Switching between Fairfax, VA and Belgium
                      Belgium
                      Member #19287
                      July 29, 2005
                      2254 Posts
                      Offline
                      Posted: August 24, 2005, 7:29 pm - IP Logged

                      Well, i considered the same thing, installing linux!
                      Today i've spend quite some time on www.sourceforge.net where everything is opensource (and much only for linux), and so far i have not found anything that pins it down, but i guess there might be some things there. If i find something thats interesting i'll surely post it here.
                      (Hopefully it ain't linuxplatform only...)
                      Provide me with the formula's, Nemesys, and i'll write the "damn" program!!! Type 

                        NemeSys's avatar - Nemesis
                        Harrogate
                        United Kingdom
                        Member #16589
                        June 1, 2005
                        107 Posts
                        Offline
                        Posted: August 24, 2005, 8:32 pm - IP Logged

                        The formulae are readily available, but most of the really useful ones have already been written into software, namely VRA, TISEAN, AVTS, Cint, Fractan, among others. Enter 'recurrence' 'analysis' 'software' into AltaVista will find most of the best software that's been written so far (I don't like Google, btw - how it got so popular is a mystery to me...)

                        I'll take a look at sourceforge, too.

                        "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 27, 2005, 5:42 pm - IP Logged

                          New sub-page added to my site:

                          http://zarnia.250free.com/rng.htm

                          "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
                            MA
                            United States
                            Member #21195
                            September 3, 2005
                            6 Posts
                            Offline
                            Posted: September 3, 2005, 11:47 pm - IP Logged

                            Wow, you guys take lottery seriously! I thought the jury was out -- the lottery results are random! If not, wouldn't there be a few smart people who figured that thing out to become billionaires and thus proving it otherwise? Well, anyway, here are a few points that I'd like to make:


                            1. Visual Recurrence Analysis (VRA) assumes the data points are arranged in time. The most recent points should be at the end of the data file.

                            2. If the lottery numbers are generated by a mechanical device, it better be a single device. If VRA detects some sort of pattern associated with that device, be my guest, make a killing. But if there are multiple devices, used in different order, VRA would be useless.

                            3. If the lottery numbers are generated by a computer that uses a faulty random generator (that is, there is structure in the results), VRA can detect it. However, I would think the rnd creators put a lot of thought in that algorithm, and I don't expect any weaknesses in it.

                            4. The best way to test VRA is to use the historical data and run the predictions. VRA pretends that it doesn't know anything about the data specified in the "test" time window. All it does is trying to extract the patterns from the "train" window and extrapolate them to the test window. If you have the data that covers, say, a few thousand samples, you can compare VRA performance to "random guess" and see if there is any value in VRA. The result of this comparison is reported as "Normalized Squared Mean Error" in VRA. The higher the NMSE, the worse is VRA prediction as compared to the trivial predictors. If NMSE is equal to 1, VRA prediction is as good as the prediction by the trivial predictors. If NMSE is greater than 1, VRA prediction is worse than the prediction by one of the trivial predictors. If NMSE is 0, the prediction is perfect. 

                            5. There are quite a few prediction models and parameters in VRA, and it can literally take years to test them.  However, there are some methods (such as Average Mutual Information, False Nearest Neighbors, and Spatio-Temporal Entropy, all implemented in VRA) that may help choosing the optimal parameters.

                            6. Lottery is somewhat unique when trying to predict -- you've got to hit one or more numbers precisely. Compare that to say, stock market. If today's price is $100 and VRA predicts $103 for tomorrow, and the actual price tomorrow is $102, VRA would make money for me because I would have bought the stock and made a profit. The analogous situation in lottery would make me no richer than I was today.  I guess what I trying to say is that although VRA doesn't know anything about the nature of the data, it would be easier to profit from its predictions if the patterns can be characterized in terms of "close", "higher", and "trend", as opposed to "precise" and "all or nothing" as it is in lottery.

                            Finally, I'd like to say that I used to write software systems for lottery, when I was young and restless. I was an atheist believing there is a hand of God in the lottery results. Nowadays I feel like God himself, but I don't have the faintest idea what the tomorrow's lottery numbers will be.

                            Eugene Kononov, Author of the Visual Recurrence Analysis software.

                              Hyperdimension's avatar - latest trace_171.gif

                              United States
                              Member #9059
                              November 26, 2004
                              128 Posts
                              Offline
                              Posted: September 4, 2005, 12:21 am - IP Logged

                              Hi Eugene,

                              Your program VRA is excellent to detect patterns, I was trying to predict a fractal Sierpinski Triangle pattern using 0 and 1 in VRA and it predicted correctly, the data:

                              1
                              1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 1 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 1 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 1 1 0 0 0 0 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0
                              1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0
                              1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 0 0 0 0 0 0 0 0 0
                              1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 0
                              1 1 0 0 1 1 0 0 0 0 0 0 0 0 0 0 1 1 0 0 1 1 0 0 0 0 0 0 0
                              1 0 1 0 1 0 1 0 0 0 0 0 0 0 0 0 1 0 1 0 1 0 1 0 0 0 0 0 0
                              1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 0 0 0 0 0
                              1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 1 0 0 0 0
                              1 1 0 0 0 0 0 0 1 1 0 0 0 0 0 0 1 1 0 0 0 0 0 0 1 1 0 0 0
                              1 0 1 0 0 0 0 0 1 0 1 0 0 0 0 0 1 0 1 0 0 0 0 0 1 0 1 0 0
                              1 1 1 1 0 0 0 0 1 1 1 1 0 0 0 0 1 1 1 1 0 0 0 0 1 1 1 1 0
                              1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 1
                              1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 1
                              1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1
                              1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
                              1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 1 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 1 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 1 1 0 0 0 0 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
                              1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0

                              The VRA results:

                               

                              Thank's for feedback,

                              Regards

                              El pensamiento ordena el caos..

                              http://1x2quinielas.blogspot.com

                                 
                                Page 7 of 12