Welcome Guest
Log In | Register )
You last visited January 17, 2017, 7:50 am
All times shown are
Eastern Time (GMT-5:00)

Neural Networks

Topic closed. 16 replies. Last post 10 years ago by tntea.

Page 1 of 2
PrintE-mailLink

United States
Member #51869
April 28, 2007
1 Posts
Offline
Posted: April 28, 2007, 10:59 pm - IP Logged

I'm gambling with a Neural Network that is trained to find patterns in lottery history using a form of artificial intelligence called 'back propagation'.  I love neural networks because that is what they are designed to do, exactly.  They find patterns in anything, and can even be trained to do mathematics or solve real world problems.  The program I am using is called Lotto Sorcerer, and it is for Mac and Windows.  For $5 a month or less if you subscribe for a year, you can get up to date lotto history for every game that's played, and it automatically loads into the program.

 

Here is some sample output of the program for the interstate Mega Millions game.  If anyone wants me to generate output for a game they play let me know.  The system  the AI generates does not change that fast, and it takes many games before it settles on a new pattern.  There are sophisticated settings you can use to train the neural network to look deeper into the game.  I've found that it comes up with about the same thing even looking at just current history, and doesn't vary much from day to day.  Still, the odds seem to go up using neural networks.

 

NON-BONUS BALL SECTION ANALYSIS
Analysis of the prior draws for the non-bonus ball section calculates that the best strategy is by picking numbers from a pool of 2 hot numbers, one cold number and 2 non-hot/non-cold numbers.

Hot number pool: 1, 3, 4, 5, 6, 7, 8, 10, 13, 17, 21, 22, 24, 25, 29, 32, 35, 36

Non-hot/non-cold number pool: 2, 9, 11, 12, 14, 15, 16, 18, 20, 26, 27, 28, 30, 31, 38, 39, 40, 42, 46, 47

Cold number pool: 19, 23, 33, 34, 37, 41, 43, 44, 45, 48, 49, 50, 51, 52, 53, 54, 55, 56

BONUS BALL ANALYSIS
Analysis of the prior draws for the bonus ball calculates that the best strategy is by picking the number from the pool of cold numbers.

Hot number pool: 1, 3, 4, 6, 7, 8, 10, 12, 15, 17, 20, 21, 22, 24, 34

Non-hot/non-cold number pool: 2, 5, 9, 13, 14, 16, 18, 19, 23, 25, 27, 29, 30, 33, 35, 36

Cold number pool: 11, 26, 28, 31, 32, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46

SUMMARY
Average Standard Deviation: 18.329
Average Arithmetic Mean: 25.487
Average Median: 37.457
Average Sum: 152.924
% Even: 50.59%

If the next drawing follows the suggested strategy, the odds of winning the highest prize (jackpot) has improved from 1:175,711,536 to 1:37,911,360, an improvement of 78%.

The suggested numbers displayed are based on the suggested strategy.

12-13-17-39-45 BB:38
05-08-19-20-38 BB:11
05-13-19-28-38 BB:43
05-13-14-33-40 BB:26
11-13-14-36-44 BB:46
07-12-14-22-45 BB:26
03-09-22-27-49 BB:43
01-07-09-12-50 BB:11
16-21-25-30-51 BB:44
14-24-25-37-47 BB:40
09-17-24-31-50 BB:11
04-12-15-36-44 BB:45
10-30-33-35-39 BB:28
05-07-26-38-50 BB:43
05-06-16-40-50 BB:38
12-13-35-39-48 BB:43
03-07-28-44-46 BB:41
25-28-30-32-51 BB:40
05-09-36-39-53 BB:26
08-13-18-38-56 BB:40
08-17-18-47-54 BB:40
01-21-42-45-46 BB:46
04-06-26-40-56 BB:31
10-26-29-42-50 BB:44
13-36-39-42-48 BB:39

    psykomo's avatar - animal shark.jpg

    United States
    Member #4877
    May 30, 2004
    5142 Posts
    Offline
    Posted: April 28, 2007, 11:48 pm - IP Logged

    Jackoof: 

    Get with the Lottery Post program.......or

     just stay........................................."NU TE"!

    Post your MEGA's & PICK 6-5-4-3-2 .........AHEAD of POST TIME!!!!!!

    cause we ain't L@@K'g too PAY a "STINKING" .........SCHEEM-SHAM

    NUTTIN "HONEY"....for KRAP!!!!!!!

    RantLeaving               PROVE WHAT uuuuuuuuuuSAY!!!!!!!

    THANKsssssssssssssssssssss..............wish uuuuuuuuuuLOL!!!!!!

    PSYKO.....No Pity!Thinking of...

      RJOh's avatar - chipmunk
      mid-Ohio
      United States
      Member #9
      March 24, 2001
      19894 Posts
      Offline
      Posted: April 29, 2007, 12:21 am - IP Logged

      I've been testing a trial copy of Lotto Sorcerer and if your data is in a text file as the following data then you can import it into the program and save the update fee.

      07/04/06  24 32 34 36 39  06
      07/04/10  07 15 16 19 28  10
      07/04/13  11 14 21 25 26  33
      07/04/17  01 04 11 31 47  37
      07/04/20  01 09 10 23 53  40
      07/04/24  08 10 35 36 43  14
      07/04/27  05 07 26 38 56  15

      I think they sell each game separately, so you could end up paying a monthly fee for each game you want to update.  Lotto Pro will update all the games it supports daily for $25/three months.

      I don't see any value in its neural network since it divides all the numbers into three pools hot, cold and non hot/cold and suggest you have numbers from all three pools in each combinations that you play, it like saying you should play all 56 main numbers and all the mega ball numbers.

      The one thing good about it you can test it before you decide whether to buy it.

       * you don't need to buy more tickets, just buy a winning ticket * 
         
                   Evil Looking       

        psykomo's avatar - animal shark.jpg

        United States
        Member #4877
        May 30, 2004
        5142 Posts
        Offline
        Posted: April 29, 2007, 1:13 am - IP Logged

        I'm gambling with a Neural Network that is trained to find patterns in lottery history using a form of artificial intelligence called 'back propagation'.  I love neural networks because that is what they are designed to do, exactly.  They find patterns in anything, and can even be trained to do mathematics or solve real world problems.  The program I am using is called Lotto Sorcerer, and it is for Mac and Windows.  For $5 a month or less if you subscribe for a year, you can get up to date lotto history for every game that's played, and it automatically loads into the program.

         

        Here is some sample output of the program for the interstate Mega Millions game.  If anyone wants me to generate output for a game they play let me know.  The system  the AI generates does not change that fast, and it takes many games before it settles on a new pattern.  There are sophisticated settings you can use to train the neural network to look deeper into the game.  I've found that it comes up with about the same thing even looking at just current history, and doesn't vary much from day to day.  Still, the odds seem to go up using neural networks.

         

        NON-BONUS BALL SECTION ANALYSIS
        Analysis of the prior draws for the non-bonus ball section calculates that the best strategy is by picking numbers from a pool of 2 hot numbers, one cold number and 2 non-hot/non-cold numbers.

        Hot number pool: 1, 3, 4, 5, 6, 7, 8, 10, 13, 17, 21, 22, 24, 25, 29, 32, 35, 36

        Non-hot/non-cold number pool: 2, 9, 11, 12, 14, 15, 16, 18, 20, 26, 27, 28, 30, 31, 38, 39, 40, 42, 46, 47

        Cold number pool: 19, 23, 33, 34, 37, 41, 43, 44, 45, 48, 49, 50, 51, 52, 53, 54, 55, 56

        BONUS BALL ANALYSIS
        Analysis of the prior draws for the bonus ball calculates that the best strategy is by picking the number from the pool of cold numbers.

        Hot number pool: 1, 3, 4, 6, 7, 8, 10, 12, 15, 17, 20, 21, 22, 24, 34

        Non-hot/non-cold number pool: 2, 5, 9, 13, 14, 16, 18, 19, 23, 25, 27, 29, 30, 33, 35, 36

        Cold number pool: 11, 26, 28, 31, 32, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46

        SUMMARY
        Average Standard Deviation: 18.329
        Average Arithmetic Mean: 25.487
        Average Median: 37.457
        Average Sum: 152.924
        % Even: 50.59%

        If the next drawing follows the suggested strategy, the odds of winning the highest prize (jackpot) has improved from 1:175,711,536 to 1:37,911,360, an improvement of 78%.

        The suggested numbers displayed are based on the suggested strategy.

        12-13-17-39-45 BB:38
        05-08-19-20-38 BB:11
        05-13-19-28-38 BB:43
        05-13-14-33-40 BB:26
        11-13-14-36-44 BB:46
        07-12-14-22-45 BB:26
        03-09-22-27-49 BB:43
        01-07-09-12-50 BB:11
        16-21-25-30-51 BB:44
        14-24-25-37-47 BB:40
        09-17-24-31-50 BB:11
        04-12-15-36-44 BB:45
        10-30-33-35-39 BB:28
        05-07-26-38-50 BB:43
        05-06-16-40-50 BB:38
        12-13-35-39-48 BB:43
        03-07-28-44-46 BB:41
        25-28-30-32-51 BB:40
        05-09-36-39-53 BB:26
        08-13-18-38-56 BB:40
        08-17-18-47-54 BB:40
        01-21-42-45-46 BB:46
        04-06-26-40-56 BB:31
        10-26-29-42-50 BB:44
        13-36-39-42-48 BB:39

        JACK:

        IF U kan prove your #'s...................sobeeit & THANK$$$$$$$$$$$$

        BUT & their is ALWAY's a ................>>>>>>.>>>>>> AH BUTTttttt!!

        PROVE-IT by an (simple) .................>>>>>>>>>>>>......EXAMPLE!!

        OTHERWIZE??????????CALL??????????????THE.........CABLE "GUYY"

        LOL>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>with/UR?/EXAMPLE$ 

          psykomo's avatar - animal shark.jpg

          United States
          Member #4877
          May 30, 2004
          5142 Posts
          Offline
          Posted: April 29, 2007, 2:51 am - IP Logged

          just wait........1-2-3-4-5-6-7-8-9-10-11-12=13>>>>>TILL!!!!!!

          pacattack05...get this bull

          HE will STRAT your SHAFT & REM YOUR BORE.....OUTT of neurtal if "U"

          don't post>>>>>>>>>>>winning EXAMPLES of this $HAMM-U-HEAR!! 

            RJOh's avatar - chipmunk
            mid-Ohio
            United States
            Member #9
            March 24, 2001
            19894 Posts
            Offline
            Posted: April 29, 2007, 2:02 pm - IP Logged

            I would think software using a neural network to reduce the number pool size by half and still have all the winning numbers 80% of the time would be more useful.  The way Lotto Sorcerer works, you can't tell if it really has a neural networks or just a formula for picking hot and cold numbers because it suggests the same mix for all pick5 or pick6 games.

             * you don't need to buy more tickets, just buy a winning ticket * 
               
                         Evil Looking       

              hypersoniq's avatar - 8ball
              Pennsylvania
              United States
              Member #1340
              April 6, 2003
              2450 Posts
              Offline
              Posted: April 29, 2007, 11:45 pm - IP Logged

              I went down the neural net road before, backprop not the best method.

              NNs have 2 phases, the first of which is training. Pattern recognition uses a separate form of NN altogether. I got to the point where one NN was set to find the patterns and the next was set to feed forward and "learn" to see them, it never worked out.

              I picked up abook on what I believe to be the "one", Bayesian Networks... but alas, I'm too dumb to understand it. I extracted the most basic algorithms and tried a theory on how NN's "predict" using weighted averages in Excel, that fell to the wayside also because in the end, the picked numbers just don't match the numbers the MUSL was picking.

              I don't want to discourage anyone from trying new approaches, but be forewarned, it's not the "holy grail"... most often the better NN programs complain about lottery data because it's too "noisy"

              Look up JOONE, a freeware NN program that runs in JAVA. There is a supportive user community but beware, the NN crowd doesn't believe NNs are useful in guessing random events, so choose your questions wisely...

              Playing more than one ticket per game is betting against yourself.

                BobP's avatar - bobp avatar.png
                Dump Water Florida
                United States
                Member #380
                June 5, 2002
                3112 Posts
                Offline
                Posted: April 29, 2007, 11:52 pm - IP Logged

                I would think software using a neural network to reduce the number pool size by half and still have all the winning numbers 80% of the time would be more useful.  The way Lotto Sorcerer works, you can't tell if it really has a neural networks or just a formula for picking hot and cold numbers because it suggests the same mix for all pick5 or pick6 games.

                It appears to work the same way as Lotto Challenger and Win Lotto Pro or whatever it's calling itself these days.

                The program instantly runs stats to see the percentage of winning numbers from hot, average, cold and due or whatever in the last 100 draws.  Determines current hot, average, cold, due based on frequency. Dumps all hot, average, cold and due numbers into hats creating  pools and then randomly picks numbers from those pools in the percentages the stats suggested based on the total number of numbers the user chooses to wheel.  There is nothing neural to this process and the program does not learn.

                BobP

                  Avatar

                  Canada
                  Member #2352
                  September 19, 2003
                  42 Posts
                  Offline
                  Posted: April 30, 2007, 11:00 am - IP Logged

                  BobP is right, I can't see any neural network analysis in Lotto Sorcerer.

                  Hypersoniq is right - the scientific and mathmatical groups will never acknowledge that it is possible to predict randomness.

                   Attrasoft had a pretty good free version that is easy to set up and work with and their support people did provide help with lottery analysis.

                  One reasearcher in Cellualar Automata worked up a model to sync the lotto stream but I was never able to find out how it worked. It did sound like a new and novel approach. From what I read, he was able consistently score all seven numbers for most lottos within 200K lines.

                  Not a practical playing field, but it would certainly beat random generation.  Intriguing stuff that.

                    savagegoose's avatar - ProfilePho
                    adelaide sa
                    Australia
                    Member #37136
                    April 11, 2006
                    3315 Posts
                    Offline
                    Posted: May 1, 2007, 12:35 am - IP Logged

                    if i send u about 10 years history, can u get it to generate some ai plays for about 12 lines?

                    2014 = -1016; 2015= -1409; 2016  = -1171; 2017 = ?  TOT =  -3596

                    keno historic = -2291 ; 2015= -603; 2016= -424; 2017 = ? TOT = - 3318

                      RJOh's avatar - chipmunk
                      mid-Ohio
                      United States
                      Member #9
                      March 24, 2001
                      19894 Posts
                      Offline
                      Posted: May 1, 2007, 2:55 am - IP Logged

                      Hypersoniq is right - the scientific and mathmatical groups will never acknowledge that it is possible to predict randomness.

                      Fortunately, scientists and mathematicians never have to acknowledges anything but facts.  The facts are no ones has ever shown that a random event can be predicted.  I doubt if anyone has ever been able to consistently pick all the winning numbers in any game with even 5% of the possible outcomes because most game have payouts that would make that profitable even if not practical.

                       * you don't need to buy more tickets, just buy a winning ticket * 
                         
                                   Evil Looking       

                        savagegoose's avatar - ProfilePho
                        adelaide sa
                        Australia
                        Member #37136
                        April 11, 2006
                        3315 Posts
                        Offline
                        Posted: May 1, 2007, 5:14 am - IP Logged

                        i have a game that isnt random , so any predictive software i want to know anout

                        2014 = -1016; 2015= -1409; 2016  = -1171; 2017 = ?  TOT =  -3596

                        keno historic = -2291 ; 2015= -603; 2016= -424; 2017 = ? TOT = - 3318

                          Avatar

                          Honduras
                          Member #20982
                          August 29, 2005
                          4715 Posts
                          Offline
                          Posted: May 2, 2007, 5:29 am - IP Logged

                          I'm gambling with a Neural Network that is trained to find patterns in lottery history using a form of artificial intelligence called 'back propagation'.  I love neural networks because that is what they are designed to do, exactly.  They find patterns in anything, and can even be trained to do mathematics or solve real world problems.  The program I am using is called Lotto Sorcerer, and it is for Mac and Windows.  For $5 a month or less if you subscribe for a year, you can get up to date lotto history for every game that's played, and it automatically loads into the program.

                           

                          Here is some sample output of the program for the interstate Mega Millions game.  If anyone wants me to generate output for a game they play let me know.  The system  the AI generates does not change that fast, and it takes many games before it settles on a new pattern.  There are sophisticated settings you can use to train the neural network to look deeper into the game.  I've found that it comes up with about the same thing even looking at just current history, and doesn't vary much from day to day.  Still, the odds seem to go up using neural networks.

                           

                          NON-BONUS BALL SECTION ANALYSIS
                          Analysis of the prior draws for the non-bonus ball section calculates that the best strategy is by picking numbers from a pool of 2 hot numbers, one cold number and 2 non-hot/non-cold numbers.

                          Hot number pool: 1, 3, 4, 5, 6, 7, 8, 10, 13, 17, 21, 22, 24, 25, 29, 32, 35, 36

                          Non-hot/non-cold number pool: 2, 9, 11, 12, 14, 15, 16, 18, 20, 26, 27, 28, 30, 31, 38, 39, 40, 42, 46, 47

                          Cold number pool: 19, 23, 33, 34, 37, 41, 43, 44, 45, 48, 49, 50, 51, 52, 53, 54, 55, 56

                          BONUS BALL ANALYSIS
                          Analysis of the prior draws for the bonus ball calculates that the best strategy is by picking the number from the pool of cold numbers.

                          Hot number pool: 1, 3, 4, 6, 7, 8, 10, 12, 15, 17, 20, 21, 22, 24, 34

                          Non-hot/non-cold number pool: 2, 5, 9, 13, 14, 16, 18, 19, 23, 25, 27, 29, 30, 33, 35, 36

                          Cold number pool: 11, 26, 28, 31, 32, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46

                          SUMMARY
                          Average Standard Deviation: 18.329
                          Average Arithmetic Mean: 25.487
                          Average Median: 37.457
                          Average Sum: 152.924
                          % Even: 50.59%

                          If the next drawing follows the suggested strategy, the odds of winning the highest prize (jackpot) has improved from 1:175,711,536 to 1:37,911,360, an improvement of 78%.

                          The suggested numbers displayed are based on the suggested strategy.

                          12-13-17-39-45 BB:38
                          05-08-19-20-38 BB:11
                          05-13-19-28-38 BB:43
                          05-13-14-33-40 BB:26
                          11-13-14-36-44 BB:46
                          07-12-14-22-45 BB:26
                          03-09-22-27-49 BB:43
                          01-07-09-12-50 BB:11
                          16-21-25-30-51 BB:44
                          14-24-25-37-47 BB:40
                          09-17-24-31-50 BB:11
                          04-12-15-36-44 BB:45
                          10-30-33-35-39 BB:28
                          05-07-26-38-50 BB:43
                          05-06-16-40-50 BB:38
                          12-13-35-39-48 BB:43
                          03-07-28-44-46 BB:41
                          25-28-30-32-51 BB:40
                          05-09-36-39-53 BB:26
                          08-13-18-38-56 BB:40
                          08-17-18-47-54 BB:40
                          01-21-42-45-46 BB:46
                          04-06-26-40-56 BB:31
                          10-26-29-42-50 BB:44
                          13-36-39-42-48 BB:39

                          YOu are doing a great job Jackpot....If you are a programmer using Neural Network, i suggest you program the computer to learn Tntea's schooling e/o : http://www.lotterypost.com/thread/142913 (read all 18 pages of it)...

                          Specifically the E/O type, not the 2-1-2 type...Teaching a computer to recognize patterns using Tntea's e/o type is better than teaching them Neural Network....If you can use an algorithm to make the computer learn the e/o types it will be more better and so that it could find patterns...

                          if you are not a programmer trying to teach a computer to learn through algorithms, then i don't know what to tell you...

                            Avatar

                            Honduras
                            Member #20982
                            August 29, 2005
                            4715 Posts
                            Offline
                            Posted: May 5, 2007, 12:00 am - IP Logged

                            YOu are doing a great job Jackpot....If you are a programmer using Neural Network, i suggest you program the computer to learn Tntea's schooling e/o : http://www.lotterypost.com/thread/142913 (read all 18 pages of it)...

                            Specifically the E/O type, not the 2-1-2 type...Teaching a computer to recognize patterns using Tntea's e/o type is better than teaching them Neural Network....If you can use an algorithm to make the computer learn the e/o types it will be more better and so that it could find patterns...

                            if you are not a programmer trying to teach a computer to learn through algorithms, then i don't know what to tell you...

                            You can also let the computer recognize TnTea's 2-1-2 types...With the 2-1-2 is better to track the lottery than the e/o....But do not use both at the same time..Let the computer analyze 2-1-2 type OR the e/o type but not both at the same time...It will be too specific if you let the computer analyze both the 2-1-2 types & e/o types....

                            When i say 2-1-2 types i mean all kinds it could be 3-2-0 or 1-1-3 any kind...I'll post more about the 2-1-2 types later.....

                              psykomo's avatar - animal shark.jpg

                              United States
                              Member #4877
                              May 30, 2004
                              5142 Posts
                              Offline
                              Posted: May 8, 2007, 8:20 pm - IP Logged

                              You can also let the computer recognize TnTea's 2-1-2 types...With the 2-1-2 is better to track the lottery than the e/o....But do not use both at the same time..Let the computer analyze 2-1-2 type OR the e/o type but not both at the same time...It will be too specific if you let the computer analyze both the 2-1-2 types & e/o types....

                              When i say 2-1-2 types i mean all kinds it could be 3-2-0 or 1-1-3 any kind...I'll post more about the 2-1-2 types later.....

                              pumpi76>>>>>>>>>>>>this is PSYKO asking?? did JACK poost some GOOD

                              ##################################'ssssssYET???

                              for JACKPOTs or PICK 3 or 4 or 5????????????????????????????????

                              ALL??????????????? anyone can ASK is try to prove a SYSTEM HIT's

                              once in a short or long>>>>>time>>>>>maby jack E-mailed U dee ##'s)

                              ASK????????????????? anyone to PROVE a FEW HIT's if it is POSSIBLE?

                              LOL>>>>>>>>>>> LET us wait for OUR NEXT...................MEAL????