Markov Software

[GillesD]

To thornc

How do you define "the lexicographical order of the combinations"? I am not sure if it means the same thing as in French. I would like to see what it gives with Lotte 6/49.

Sure, many distributions are possible to fit data from a lottery. One example is the DELTA numbers (quite famous at one time) which gives a nearly perfect exponential distribution.

 

[Dennis Bassboss]

Try looking at the lexicographical order of the combinations...
In the existing 1000+ draws of the PT 6/49 lottery it's starting to resemble a normal distribution!

Heck, but don't forget that there are more complicated probability distributions out there... The Normal is just the one that most people know how to handle, or the first that they resort to!

In fact I would be surprised if someone someday would relate lottery drawing to Binumial or perhaps Chi Square distribution...
(just as an example...)

Resemble a normal distribution.....
Well we still have to wait for a while to see...But let me tell you this ...If we seperate all the possibles draw by 1,000 chunks we would get for some of these distributions or (rather small chunks) results way off the overall average created by the total population of distributions (or the addition of all these small chunks) having a median travelling or to the right or to the left side of a graph so having only one sample or two of these is not a very reliable data for such an analyze...
As Gilles D pointed out the overall distribution will strongly tend to equalize at one point after an enormous quantity of draws....But the way that it will is not the same for all the 6/49 lotteries...I mean by that that the numbers that hit mostly in the early stages shouldn't be the same in any lotteries....
or let me state that another way... the numbers with the less hits that would be the last to come close to that equal overall numbers of hits wouldn't be the same in any lotteries...
I tend strongly to agree with Gilles D about the Gauss Distribution not being a very reliable tool for individual numbers in the lotteries...

 

[thornc]

Brad
Care to explain those results of yours?

Dennis
Off course, it isn't... if it was then surely every statics teacher in the world would have won the lottery by now! And off course over the long run all the numbers will tend to equalise their hit count. That's easy to understand. The point is that in a small scale we can consider that drawings follow a probobility rule... the problem is finding which one!

Gilles
"lexicographical order" is the order in which all of the possible 14M- combinations can be sorted, so
1 2 3 4 5 6 -> #1
1 2 3 4 5 7 -> #2
1 2 3 4 5 8 -> #3
And so on.

The funny thing about using that Excell function is that you're using a Uniform distribution, in a pseudo-random software, made by a pseudo-perfect software developer on a pseudo-good computer ... Almost the same as using my Markov software, which uses the same Uniform distribution, it's also pseudo-random because I seed the generator according to the clock, I'm a pseudo-inperfect software coder and my computers are good !

Anyway, hope that others share their results...
BTW this is somehow related to that announcers thing Dennis spoke about on my other post!

 

[Brad]

Thornc,

re: Markov

I selected 10, 15, .... 250 line inputs (hist.lines) from my history file going back from draw#1149, generated output for each, then checked for matches (hits) and tabulated results for draw#1150.

Reason for having 3x 20 line inputs ... got only 1 output line 1st run. Noticed that each run has different output --> random selection?

Am using leading zeros cuz my chart format does not paste legibly to this BB.

Anything else?

 

[thornc]

Brad

Yap, there's a random selection, that's why the number of lines changes from run to run (for the same input file)!

I take that the most recent draws in history file(input file) where at the top! Because those then to influence the outptu more...

Anyway thanks for your input... Saliu thinks that input files with N/2 lines may have better results (N being the largest lotto number!)...

I don't expect the program to hit good results for a single draw, but I would hope to be able to improve it to get good results for say 10 draws (5 weeks in PT!).

 

[Brad]

Thornc,

yes to most recent on top.

So with a 6/49 lotto like BC49 would you recommend 24 or 25 input lines and then multiples of those?

 

[thornc]

Brad

I did some tests using N=50(close enough to 49 ) and checked using 25,50,100,150 and 500 input lines... against the next 10 draws. My results where far from good... but on one occasion I did get a 5in6, problem was that I had close to 1000 lines to play in 10 draws... to much COW (cost of winning!)

Do try to run to program several times with the same input file... believe it or not, after a few run the results are better! I think it has to do with the cache coerrency of the computer!(just joking!
)

 

[Brad]

Thornc,

OK I'll try that, my intentions were to go back 10 draws as I understand your point re: 1 draw being insufficient but I ran out of time ... maybe later today.

Cut my teeth so to speak in the first runs and must say that I agree with Ion's '5 point plan' to improve Markov's usability, as it stands now it's not very friendly. Me thinks many would be more likely to give it a go then ... kind'a like doing dishes by hand when you have a dishwashing machine .

Cheers

 

[GillesD]

Normal distribution with lexicographical order

Thornc,

With Canadian Lotto 6/49, I looked at what would be the distribution based on lexicographical order.

With 1986 draws, this lottery is quite representative of what could be expected of a 6/49 lottery.

Using 15 sub-groups (roughly 932,255 posible combinations per sub-group), I calculated the number of occurence in each sub-group for the 1986 results. Here are my results:

- from 01-02-03-04-05-06 to 01-08-19-22-24-28: 142 times
- from 01-08-19-22-24-29 to 02-03-25-36-37-48: 134 times
- from 02-03-25-36-37-49 to 02-12-24-31-34-39: 124 times
- from 02-12-24-31-34-40 to 03-07-17-28-43-45: 124 times
- from 03-07-17-28-43-46 to 04-05-09-41-42-43: 120 times
- from 04-05-09-41-42-44 to 04-16-27-33-43-49: 134 times
- from 04-16-27-33-44-45 to 05-13-21-28-30-38: 125 times
- from 05-13-21-28-30-39 to 06-13-15-16-17-47: 117 times
- from 06-13-15-16-17-48 to 07-14-21-23-41-49: 132 times
- from 07-14-21-23-42-43 to 08-19-21-26-33-43: 129 times
- from 08-19-21-26-33-44 to 10-12-19-22-31-41: 141 times
- from 10-12-19-22-31-42 to 11-26-32-42-44-48: 145 times
- from 11-26-32-42-44-49 to 14-16-30-39-45-46: 133 times
- from 14-16-30-39-45-47 to 17-28-29-31-40-41: 130 times
- from 17-28-29-31-40-42 to 44-45-46-47-48-49: 156 times

To me, these results do not look like a pseudo-normal distribution but more like a real uniform distribution. Repeating this analysis with 10, 30 or 50 sub-groups, I still found the distribution to be uniform and never normal (or Gaussian).

Finally, using a 10,000 randomly generated combinations (by my pseudo-random software) and the same 15 sub-groups, I still found a uniform distribution with groups having betwee 646 and 705 combinations.

And by the way, I think most of us here in Quebec will agree on this: given some time (and it will not be that long), Markov will be recognized as one of the best in his peers.

 

[thornc]

Gilles

Very interesting, our PT lottery has around 1100/1200 draws and guess what it resembles more Gaussian then Uniform...but I guess that with more draws it will tend to that.

And don't be suprised if your computer generated combinations look Uniformly distributed, most "Random Generator" functions in programming languages are Uniform! If you want I can look up some procedure on how to transform create Gaussian Random Variables from Uniform Random Variables, I have it somewhere I used it a lot in my signal processing days!

Finnally do you mind explaining what do you mean by:

given some time (and it will not be that long), Markov will be recognized as one of the best in his peers.

You see since English is not my main language, I some times have problems understanding some things!! Worse yet is that I learned British English in school and sometimes the American and Canadian Englishs are somehow different!

 

[GillesD]

A few comments

A - I believe all 6/49 lotteries should come up as an Uniform distribution. For me to verify this, I would like to get a look at the results of your lottery. Can you provide me the information on a site that would list all results of your lottery?

B - I sure hope that a large amout of randomly generated numbers is uniformly distributed. This is what it should be unless you program it otherwise

C - No need for a complex formula to generate other distribution, my pseudo-random software can easily generate numbers with Uniform, Normal, Poisson, Bernouilli distributions as well as a few other ones.

D - About my quote you do not understand, sorry about this. I had not noticed that you are located in Portugal. I was talking about A. Markov, a defenseman with the Montreal Canadians (#79) in hockey

 

[thornc]

Re: A few comments

A - I believe all 6/49 lotteries should come up as an Uniform distribution. For me to verify this, I would like to get a look at the results of your lottery. Can you provide me the information on a site that would list all results of your lottery?

You can get an excell file with all of the PT results here!
Watch out that the number of numbers changed from 45 to 49 somewhere in the late 80s ealy 90.... I alsways have to check and can't seem to remember exactly when!

 

[Dennis Bassboss]

Re: Re: A few comments

You can get an excell file with all of the PT results here!
Watch out that the number of numbers changed from 45 to 49 somewhere in the late 80s ealy 90.... I alsways have to check and can't seem to remember exactly when!

The changed from 6/45 to 6/49 occurred the 05/19/1990...And if we count only the 6/49 draws you have not reach the 1,000 draws psychological barrier in that lotto yet...But you are now getting close to it...

 

[thornc]

Re: Re: Re: A few comments

The changed from 6/45 to 6/49 occurred the 05/19/1990...And if we count only the 6/49 draws you have not reach the 1,000 draws psychological barrier in that lotto yet...But you are now getting close to it...

Sorry but you're wrong on this one... just checked the history file!
In the 3rd draw of 88 (16th January) the number 46 was the first to be drawed that was bigger then 45... I assume that the change happened in the pass of the year...

So we're well above the 1000 draws barrier

 

[Dennis Bassboss]

Re: Re: Re: Re: A few comments

Sorry but you're wrong on this one... just checked the history file!
In the 3rd draw of 88 (16th January) the number 46 was the first to be drawed that was bigger then 45... I assume that the change happened in the pass of the year...

So we're well above the 1000 draws barrier

Well then my database for that one is flaud ...could be I'm not playing with that one very often...At first I tough that I only had the 6/49 draws from the very first of them...In my database draw 01 is 05-19-1990...Where did you got your excel database from? If you know of an official site or somewhere that I could get it I'd like to verify my datas and update them..directly from there!!

 

[thornc]

Re: Re: Re: Re: Re: A few comments

Where did you got your excel database from? If you know of an official site or somewhere that I could get it I'd like to verify my datas and update them..directly from there!!

Believe or not the link above is from the Portuguese "lottery comission"!
www.djogos.scml.pt
djogos=Departamento de Jogos=Games Department
scml=Santa Casa da Misericórdia de Lisboa-> The institution to which the lottery is assigned!

They only supply that excell file with the history!!
The file includes all the draws ever in the PT lottery plus the new JOKER game (which is included in the lottery slips)

 

[Dennis Bassboss]

Thank you !

 

[Dennis Bassboss]

Hello all,

After looking at tonight's draw on TV I notived something that got me thinking. I'll present the problem first and put my questions/comments after!

It's all about the Portugueese lotto, we have two draws one on saterday and another on monday; the only way you can bet for the monday one is to bet for the saterday also with exactly the same bet but paying twice, one can play only for saterdays also.
The problem is that it had been a while since I had watched one live draw and to my suprise, they are using different machines! I had thought that where the same but they are different, the problem seems to arise that the draws are being transmitted live by two different TV stations and each station has a different machine on their studio, I thought they were using the Game Department's studio and machine but they are not!

So where does this leaves me/us....
Two different lotto games or a single lotto game with two draws like we're led to believe....I had been playing like it was a single game with not so good results but now I'm starting to think differently....

Comments are welcome...
BTW Dennis, on the "Markov Software" thread I give some wrong information, the PT lotto has started using 6/49 in 19/05/1990 draw 20 of that year, not before as I stated, before was 6/47!!

 

[Sérgio]

Thornc

I´m from Portugal also and as far as to portuguese lottery it started as 6/45, latter changed to 6/47 and finally became 6/49 in draw 20/90 I think.

Since you are a programmer and like totoloto I sugest you visit Artur Marques site and search for g-loto, it´s a recipe to built a lotto program based on genetic algorithms.

I plan to use it and built my own program somewhere in the future, it´s interesting that aproach.

 

[thornc]

Oi!

Thornc

I´m from Portugal also and as far as to portuguese lottery it started as 6/45, latter changed to 6/47 and finally became 6/49 in draw 20/90 I think.

Since you are a programmer and like totoloto I sugest you visit Artur Marques site and search for g-loto, it´s a recipe to built a lotto program based on genetic algorithms.

I plan to use it and built my own program somewhere in the future, it´s interesting that aproach.

Sorry to other folks going to post in Portuguese, if that's a problem LT let me know!

Primeiro que tudo obrigado pela dica do site, há algum tempo um amigo meu da universidade tentou fazer uma coisa do genero, mas acho que ele acabou por desistir... os GA não se adaptam bem a ambientes aleatorios, e no meu ver o loto é um no fundo um problema aleatório... claro que podemos sempre ver a coisa de outra forma e considerar que apenas temos que procurar a chave ou os numeros certos e ai os GA funcionam perfeitamente...

Back to english, sorry but it was a lot faster to me to write that paragraph in Portuguese... Sérgio since you visinting come back again and do post in the international forum, I'm fealling lonely being the only one posting about PT lottery!
Ok Dennis you do post also!

About the history changes I was aware of them, I only didn't recall the dates!