Welcome Guest
You last visited December 4, 2016, 1:15 pm
All times shown are
Eastern Time (GMT-5:00)

# Nonlinear regression to predict lotteries

Topic closed. 119 replies. Last post 2 years ago by lottoswe.

 Page 7 of 8
Tahiti- Polynesia
Tuvalu
Member #34524
March 4, 2006
54 Posts
Offline
 Posted: January 6, 2015, 7:38 pm - IP Logged

Hi,

For the last drawing Euromillions - Tuesday 6th January 2015 : 14 20 30 38 49

I didn't played but my test give : 49 4 14 20 36 (first 5 numbers predicted on sorted history file).

Bob

United Kingdom
Member #162012
December 19, 2014
23 Posts
Offline
 Posted: January 7, 2015, 8:02 am - IP Logged

Hi,

Are those statistics calculated on real tickets you bought, or they calculated on a theorical basis? How do you get them?

I don't calculate the statistics of my results. If "Sweepstakes" mean "ready made tickets", then I'm sure that my results are far better than random tickets sold by Euromillions.

5/15 happen 1, or more rarely, 2 times a year. Drawings happen 2 times a week, that's 104 times a year. So, success percentage is closer to 1 or 2%. 6% is huge.

Bob

Are those statistics calculated on real tickets you bought, or they calculated on a theorical basis?

It is the theoretical performance of my system. My cash in 2014 ROI was a + 5% of the total profit of the money invested (after paying the fees of 21% of profit) . Two years ago in Spain there were no gains taxes on lottery winnings , so it was much more profitable.

Explain the concepts :
Profit: Gross cash prize you win by hitting a draw.
Benefit : Budget Amount after paying taxes on the prize and then amortizing payment of investment.

How do you get them?

Once I have developed my calculation model, I submit to several performance tests consisting of several modes of play, on samples of 100 drawings (100 random draws, and from the draw 400 to 499 and the last 100 draws). The results of these bursts of evidence, recorded in database tables to check the deviation between what actually came out and what the system predicted.

When I indicate for example that the skill of 5 numbers is 0-6 %, want to express that in the performance tests can give results from 0% (0 sweepstakes) where they got the 5 numbers, up to a maximum (and the best) 6% (6 draws) where the system itself matched all five numbers in the previous test.

Usually the pool of numbers is between 15 and 20 numbers, so then total combinations may vary in each draw.

I don't calculate the statistics of my results.

Measure -> Control -> Management -> Continual Improvement

I have a log of my systems/models with a snapshot (I use subversion) and the %performance and other economics data. I have implemented transversely business processes to monetize the system and methods/protocols scientific to control/improvement; If these steps are not followed is very difficult to have all these matrices testing results on the head !!!

If "Sweepstakes" mean "ready made tickets", then I'm sure that my results are far better than random tickets sold by Euromillions.

`I completely agree ! I also bought 10 tickets sweepstakes machine to perform comparisons with my bets . For example , if I make 10 bets EuroMillions (20 € ) , I also do 10 bets ( another 20 € ) automatically by the vending machine. After 100 draws, I have no doubt that applying scientific method has competitive advantage.  I have 3 levels to operate my system, and you can see it en this post (my old username anubina): http://www.lotterypost.com/thread/279051/38941835/15 happen 1, or more rarely, 2 times a year. Drawings happen 2 times a week, that's 104 times a year. So, success percentage is closer to 1 or 2%. 6% is huge. 6% is the best case scenario and investment in such systems is highly risky and economically inadvisable (I only gained 5% limpid), so it is a very high risk to only 5%.  Although consistency try to improve profit by not paying rates of 21% (3,000€ prizes lower pay no fees)... `
Tahiti- Polynesia
Tuvalu
Member #34524
March 4, 2006
54 Posts
Offline
 Posted: January 7, 2015, 5:59 pm - IP Logged

Are those statistics calculated on real tickets you bought, or they calculated on a theorical basis?

It is the theoretical performance of my system. My cash in 2014 ROI was a + 5% of the total profit of the money invested (after paying the fees of 21% of profit) . Two years ago in Spain there were no gains taxes on lottery winnings , so it was much more profitable.

Explain the concepts :
Profit: Gross cash prize you win by hitting a draw.
Benefit : Budget Amount after paying taxes on the prize and then amortizing payment of investment.

How do you get them?

Once I have developed my calculation model, I submit to several performance tests consisting of several modes of play, on samples of 100 drawings (100 random draws, and from the draw 400 to 499 and the last 100 draws). The results of these bursts of evidence, recorded in database tables to check the deviation between what actually came out and what the system predicted.

When I indicate for example that the skill of 5 numbers is 0-6 %, want to express that in the performance tests can give results from 0% (0 sweepstakes) where they got the 5 numbers, up to a maximum (and the best) 6% (6 draws) where the system itself matched all five numbers in the previous test.

Usually the pool of numbers is between 15 and 20 numbers, so then total combinations may vary in each draw.

I don't calculate the statistics of my results.

Measure -> Control -> Management -> Continual Improvement

I have a log of my systems/models with a snapshot (I use subversion) and the %performance and other economics data. I have implemented transversely business processes to monetize the system and methods/protocols scientific to control/improvement; If these steps are not followed is very difficult to have all these matrices testing results on the head !!!

If "Sweepstakes" mean "ready made tickets", then I'm sure that my results are far better than random tickets sold by Euromillions.

`I completely agree ! I also bought 10 tickets sweepstakes machine to perform comparisons with my bets . For example , if I make 10 bets EuroMillions (20 € ) , I also do 10 bets ( another 20 € ) automatically by the vending machine. After 100 draws, I have no doubt that applying scientific method has competitive advantage.  I have 3 levels to operate my system, and you can see it en this post (my old username anubina): http://www.lotterypost.com/thread/279051/38941835/15 happen 1, or more rarely, 2 times a year. Drawings happen 2 times a week, that's 104 times a year. So, success percentage is closer to 1 or 2%. 6% is huge. 6% is the best case scenario and investment in such systems is highly risky and economically inadvisable (I only gained 5% limpid), so it is a very high risk to only 5%.  Although consistency try to improve profit by not paying rates of 21% (3,000€ prizes lower pay no fees)... `

Hi,

Bob

New Mexico
United States
Member #86099
January 29, 2010
11116 Posts
Offline
 Posted: January 12, 2015, 2:04 pm - IP Logged

Hi all

Here are a few pics that shows linear regression.  The first pic shows the program for which I am using

regression to predict the next value for Rogue positions A to J for my 5-39 game.  Each rogue position

has 4 possible choices, 0-1-2-3 so each value is expected to show on average around 1 of every 4 plays.

The next pic shows a scatter plot which shows the number of shows for each value over 200 games.

The next pic shows the value plotted one at a time

The black line here shows the expected and as you can see the actual values varies over the plot.  This graph

is very small and one pixel change up or down indicates one value change which is hard for us to see here but

easy for the program.  Next the program will draw many new straight lines to find the one that runs through

the exact center of as many as possible.   There are a number of different methods used to plot the next x-y

position for the value to hit or to assign a probability for each value showing in the next draw.  Sometimes if

there are too few values the plots will show that each value has a very good chance of hitting where as stated

above if too many then we have to switch to non-linear regression.  These values are very linear as you can see

from the graphs.  Hope this helps.

RL

Wow!

Are you going to upload this?

United States
Member #59354
March 13, 2008
3962 Posts
Offline
 Posted: January 13, 2015, 10:54 pm - IP Logged

lakerben

For now I have to say no, the program is still under development but so far it's proving to be the best

I have ever written.  I have deleted all my older stuff from my computer.  I have a daily game version

that operates the same way and it will reduce a P-4 to 3 lines with 6 rogues.  Each rogue value has a

range of 0 to 3 and it outputs straight hits.  I am in the process of testing all my old predictions tools

and analysis stuff to see what I can find that helps pin down the rogue values.  P-3 can be reduced to

5 lines with 4 rogue values or 2 lines if 5 rogues are set.   The hard part right now is trying to decide

which game to work on, number games or daily games,  both are really tempting.   These programs

don't need extra filtering which was the motive behind them.  Anyway, thanks for your interest and

maybe at some point in the future I will release it.   I kind of overbuilt these programs as they have

a number of hidden options, wheels etc.. that give the user many ways to generate the lines and set

the rogue values..

Pic of P-4 setup using 6 values.

Working on my Ph.D.  "University of hard Knocks"

I will consider the opinion that my winnings are a product of chance if you are willing to consider

they are not.  Many great discoveries come while searching for something else

Trump / 2016 & 2020

Auckland
New Zealand
Member #2247
September 4, 2003
33 Posts
Offline
 Posted: January 14, 2015, 12:59 am - IP Logged

dr san

I am kind of excited about this new program, Setting 10 rogue values will produce a 5of5 on one line

as compared to setting DMP's 10 digit filters to single values which produced 45 lines in last nights

game.   Most of the digit filters will range around 0 to 3 for most games so in a heads up comparison

the new program beats it  hand's down.  You can also see in the pic below that using the digit filters

in DMP that there were 24 non-prize paying lines.  If I set the new program to generate using just 5

rogue values it produces 563 lines and every line will at least break even.  It's a very nice reduction

tool and I hope I can build a nice predictor for it.   Anyway, only time will tell.

RL

Hi RL,

Can you explain more about a rouge value to a non-programmer?

http://bryntegict.co.uk/resources/computing/theteacher/newalevel/cp1_2_3.htm

spot9

Georgia
United States
Member #129908
July 1, 2012
200 Posts
Offline
 Posted: January 14, 2015, 1:09 am - IP Logged

lakerben

For now I have to say no, the program is still under development but so far it's proving to be the best

I have ever written.  I have deleted all my older stuff from my computer.  I have a daily game version

that operates the same way and it will reduce a P-4 to 3 lines with 6 rogues.  Each rogue value has a

range of 0 to 3 and it outputs straight hits.  I am in the process of testing all my old predictions tools

and analysis stuff to see what I can find that helps pin down the rogue values.  P-3 can be reduced to

5 lines with 4 rogue values or 2 lines if 5 rogues are set.   The hard part right now is trying to decide

which game to work on, number games or daily games,  both are really tempting.   These programs

don't need extra filtering which was the motive behind them.  Anyway, thanks for your interest and

maybe at some point in the future I will release it.   I kind of overbuilt these programs as they have

a number of hidden options, wheels etc.. that give the user many ways to generate the lines and set

the rogue values..

Pic of P-4 setup using 6 values.

RL, I am waiting with baited breath.

Just do it......

United States
Member #59354
March 13, 2008
3962 Posts
Offline
 Posted: January 14, 2015, 2:52 am - IP Logged

Spot9

The rogue values are generated by a algorithm / conversion process.  The only thing special about

the values are that they are somewhat more predictable than the actual numbers they represent.

Five rogues equates to 4*4*4*4*4=1024 which will more than cover a pick-3 game as each one is

ranged from 0 to 3.

The overall odds are the same so there is no advantage from that perspective but through analysis

I have been able to predict around 80% without much effort.  The program has a game-mode which

allows playing against real data from the games history and I hope I can work up to the 100% level.

If I can reach the level where I can predict the first 6 rogue values for my 5-39 then any line I play from

those generated will pay 10 to 1.  In the pic below you can see that playing just the first 6 will produce

3of5 or greater winners.   Six rogues reduces to 142 which if a person could achieve that level then it's

within the realm of buying every line.    Even if a person could only manage it one out of every 30 games

it would be a money maker.  For now I am working on picking 9 rogues for a 3 line play.

6-rogue play

Working on my Ph.D.  "University of hard Knocks"

I will consider the opinion that my winnings are a product of chance if you are willing to consider

they are not.  Many great discoveries come while searching for something else

Trump / 2016 & 2020

Economy class
Belgium
Member #123700
February 27, 2012
4035 Posts
Offline
 Posted: January 14, 2015, 1:55 pm - IP Logged

I like the idea with different colors for the numbers.

New Mexico
United States
Member #86099
January 29, 2010
11116 Posts
Offline
 Posted: January 16, 2015, 10:44 pm - IP Logged

lakerben

For now I have to say no, the program is still under development but so far it's proving to be the best

I have ever written.  I have deleted all my older stuff from my computer.  I have a daily game version

that operates the same way and it will reduce a P-4 to 3 lines with 6 rogues.  Each rogue value has a

range of 0 to 3 and it outputs straight hits.  I am in the process of testing all my old predictions tools

and analysis stuff to see what I can find that helps pin down the rogue values.  P-3 can be reduced to

5 lines with 4 rogue values or 2 lines if 5 rogues are set.   The hard part right now is trying to decide

which game to work on, number games or daily games,  both are really tempting.   These programs

don't need extra filtering which was the motive behind them.  Anyway, thanks for your interest and

maybe at some point in the future I will release it.   I kind of overbuilt these programs as they have

a number of hidden options, wheels etc.. that give the user many ways to generate the lines and set

the rogue values..

Pic of P-4 setup using 6 values.

Wow! I hope you make this available!

Tn
United States
Member #54963
September 4, 2007
1164 Posts
Offline
 Posted: January 21, 2015, 11:26 am - IP Logged

Spot9

The rogue values are generated by a algorithm / conversion process.  The only thing special about

the values are that they are somewhat more predictable than the actual numbers they represent.

Five rogues equates to 4*4*4*4*4=1024 which will more than cover a pick-3 game as each one is

ranged from 0 to 3.

The overall odds are the same so there is no advantage from that perspective but through analysis

I have been able to predict around 80% without much effort.  The program has a game-mode which

allows playing against real data from the games history and I hope I can work up to the 100% level.

If I can reach the level where I can predict the first 6 rogue values for my 5-39 then any line I play from

those generated will pay 10 to 1.  In the pic below you can see that playing just the first 6 will produce

3of5 or greater winners.   Six rogues reduces to 142 which if a person could achieve that level then it's

within the realm of buying every line.    Even if a person could only manage it one out of every 30 games

it would be a money maker.  For now I am working on picking 9 rogues for a 3 line play.

6-rogue play

Hey RL Boot here!! Hope you getting it narrowed in!!! Looks good! Best of luck with it!!

WHEN IT FEELS THE WHOLE WORLD SUCKS!

RELAX.........IT'S ONLY GRAVITY

I think I can I think I can!!!!

From Denver, Rocky Mountain Empire,
United States
Member #49750
February 13, 2007
439 Posts
Offline
 Posted: January 26, 2015, 10:57 pm - IP Logged

Hi all

Here are a few pics that shows linear regression.  The first pic shows the program for which I am using

regression to predict the next value for Rogue positions A to J for my 5-39 game.  Each rogue position

has 4 possible choices, 0-1-2-3 so each value is expected to show on average around 1 of every 4 plays.

The next pic shows a scatter plot which shows the number of shows for each value over 200 games.

The next pic shows the value plotted one at a time

The black line here shows the expected and as you can see the actual values varies over the plot.  This graph

is very small and one pixel change up or down indicates one value change which is hard for us to see here but

easy for the program.  Next the program will draw many new straight lines to find the one that runs through

the exact center of as many as possible.   There are a number of different methods used to plot the next x-y

position for the value to hit or to assign a probability for each value showing in the next draw.  Sometimes if

there are too few values the plots will show that each value has a very good chance of hitting where as stated

above if too many then we have to switch to non-linear regression.  These values are very linear as you can see

from the graphs.  Hope this helps.

RL

Greetings, RL! Is this an extension of the 10 programs we have? If not, how is it different? Details please. Thanks!

Go Broncos!

New Mexico
United States
Member #86099
January 29, 2010
11116 Posts
Offline
 Posted: January 31, 2015, 2:01 pm - IP Logged

Greetings, RL! Is this an extension of the 10 programs we have? If not, how is it different? Details please. Thanks!

United States
Member #59354
March 13, 2008
3962 Posts
Offline
 Posted: February 15, 2015, 10:28 pm - IP Logged

Hi all

The 10 tools I uploaded in the other topic were for anyone who wanted to play around with digit

theory.

Most of my stuff comes with a bias search tool that is based on regression but it falls short of pinning

the tail on the donkey so to say.   I don't plan to release the rogue program anytime soon as it is still

being tested and developed.

With the Rogue-DG version I only need to set the first 4 values for a P-3 straight hit in 5 lines and I

currently averaging 3 out of 4 most attempts which is far better than the expected.  You can see from

the pic below that I have made several changes and added other analysis options.

For the number games, at least for now I am still using an old lexie program and playing one line each

game.  I have gotten close with this program a few times but it's very hard to hit all 6 values.   It seems

that the best predictor I have found so far is gut-instinct but I am not ready to stop looking.

Lexie program

Working on my Ph.D.  "University of hard Knocks"

I will consider the opinion that my winnings are a product of chance if you are willing to consider

they are not.  Many great discoveries come while searching for something else

Trump / 2016 & 2020

Tahiti- Polynesia
Tuvalu
Member #34524
March 4, 2006
54 Posts
Offline
 Posted: February 16, 2015, 12:42 am - IP Logged

bla bla bla bla...etc.....

 Page 7 of 8