Markov Chains

[Omega71]

Hello Everyone,

I'm sorry I've been away for awhile but I had to take care of a personal problem. Anyway, all is well now. Frank I hope you didn't think I was no longer interested in your spreadsheet because I most certainly am. I haven't had much success with it, though, but I'll keep trying. Thanks again Frank for your hard work. Take care everyone.

 

[bloubul]

Hi All

Where are you who partake in the Markov Chain. Well I'm getting at least 3# every draw. How are you doing.

Once again a BIG THANK YOU to FRANK

BlouBul

 

[HalfBee]

Hi All
Where are you who partake in the Markov Chain. Well I'm getting at least 3# every draw. How are you doing.
Once again a BIG THANK YOU to FRANK
BlouBul

How are you selecting the numbers with the information you extract?
I've a list of followers for each number currently drawn but haven't been able to overlap those lists to pull out anything useful yet.

Haven't worked on it in a while since nothing jumped out.

BTW: am not using spreadsheet version, but VB program I wrote myself for extra slice and dice abilities.

HalfBee

 

[bloubul]

Hi HalfBee

The secret lies in the number of draws you use. You must find the correct number of draws, to little nothing... to much the same.... Well with help I have found it....

BlouBul

 

[doug_w]

Frank You Are A Scholar And A Gentleman

Spent this Sunday afternoon creating the spread sheet from your terrific instructions. After few minor mistakes like no numbering the draws from 1 onwards, got the whole thing up and running. The time and effort that you put into this is very much appreciated and has been a great way to learn about Markov Chains. Now just need to learn to interpret the data calculated correctly.

 

[Frank]

I'm pleased you enjoyed the exercise Doug, as Bloubul says you have to tweak it to look back the optimum number of draws, decide whether to go for balls that follow the most, or balls that follow the least. Also whether you want the top 3 or just the top balls etc. You will probably then end up with a large pool of numbers from the columns corresponding to the last draw result. Duplicated or better - of any numbers helps to produce an order of priority to reduce your shortlist.
I'm currently using a look back of 600 draws and picking the top 3 balls following the balls of the last result. I get a pool of between 25 and 36 balls, many duplicates in the list , the highest counts from this list are my favourites.

 

[ChesterDennis]

Spent this Sunday afternoon creating the spread sheet from your terrific instructions. After few minor mistakes like no numbering the draws from 1 onwards, got the whole thing up and running. The time and effort that you put into this is very much appreciated and has been a great way to learn about Markov Chains. Now just need to learn to interpret the data calculated correctly.

Sounds like a worthy exercise to learn Excel. I just might try it.

CD

 

[FlourishingLight]

I have never looked into the Markov Chains, and really don't know what they are. I also am not familiar with Excel spreadsheets, so likely will be mind boggled to try to do what you guys are doing here. BUT, I appreciate any leads I can get, so will look into what Markov Chains are.

I would like your input one of my methods....and the best part is...it is consistent, and seems to work across many lotteries. I'd like to know how it works in yours.

Here it is. No graphs or spreadsheets necessary. For every number that came out in the last drawing, play 7 away. Look here at the last Lotto drawing from last night in Florida. For almost every number in the past two draws, you can find that the previous 2 draws have exactly 7 numbers away from the other numbers. I have been watching this for awhile. And it happens A LOT! Though, not for every single number, every time. (that would be too easy!)

SAT 06/20/15: 10-14-28-35-42-49
WED 06/17/15: 12-16-17-19-33-39
SAT 06/13/15: 07-08-13-15-28-50
WED 06/10/15: 30-31-32-41-50-53

In line one, you have 10, and right below you have 17.
In line one, you have 14, and two below, you have 7.
In line one, you have 28, and in same line, you have 35.
In line one you have 35, and in same line, you have both 28 and 42.
In line one, you have 49, and in same line you have 42

In line 2, you have 12, and in same line you have 19.
In line 2, you have 16, and that has no matches in two lines.
In line 2, you have 17, and that has no matches in two lines.
In line 2, you have 19, and that has 12 in same line.
In line 2 you have 33, again with no match.
In line 2, you have 39, and two below, you have 32.

In line 3, you have 7, and one line below, you have 53.
In line 3, you have 8, and in same line you have 15.

16-21-23-29-32-38 would be the drawing from 6/6/15

So that means that number 28 on the third line, in above post, also has match, two lines down with number 21.

I haven't gone further with the examples, but you might get the idea. You may not get all 6 with matches of seven away, but if you look for it, this is the way it seems to happen in many lotteries, including the 5 of 36 games and even Powerball, sometimes. I haven't looked into too many others.

Please tell me if it works in your lottery....and if you think this is useful. The problem I see with it is having to wheel 24 numbers, unless you leave out the numbers that already aligned with the numbers, before....and also the problem is that sometimes the 7 away is in the same drawing. I am wondering if there is a way to perfect this method?

Blessings!
FL

 

[FlourishingLight]

Number 16 had no match in line 2, until three draws down, with number 23.

 

[doug_w]

Frank - Have One Question When Generating Chains

The following is small sample of draw from our Saturday Lotto Draw here in Australia

42 36 41 18 31 28
39 5 38 1 36 13
44 21 11 17 6 16
19 8 9 21 6 41
16 43 33 39 20 7
39 21 18 35 38 43
25 4 19 27 34 24
30 33 16 10 1 41
34 10 30 12 44 42
36 19 26 9 15 33
33 30 9 19 41 44
42 22 35 16 37 24

Looking at these results the number 36 follows 42 in the first draw.

Then in draw 10 the number 36 is the first ball drawn and 42 is the last ball drawn in the previous draw number 9.

When generating do you include both of these as 36 follows 42 twice, or do you exclude where they go over draws and just count it as 36 follows 42 once only ?

Also is it best to have the draw results in drawn order or numeric order ?

Cheers

Doug

 

[Frank]

Firstly it is important that the most recent draw is at the top, and has the highest draw number. The oldest draw is at the bottom and has the lowest draw number. So in the example you give, draw number 12 would be at the top (42,36,41,18,31,28). Draw number 1 at the bottom is 42,22,35,16,37,24.

The draw you refer to as draw 10 is actually draw 3 and what you called the previous draw, ( draw 9) actually is later (draw 4) so 42 follows 36, not the other way around.

You also said " looking at these results the number 36 follows 42 in the first draw." No it doesn't.
All the numbers in the last draw ( remember this is the most recent draw, at the top) were drawn in the SAME draw, so you can't say a ball follows another in the same draw - we are looking for which balls follow named balls in the NEXT draw - not the same draw! This is so that after the number crunching is over, we can infer which numbers will follow the most recent draw result , in the next draw to come, in the future.

In reality, number 36 doesn't follow number 42 at all in the NEXT draw, starting from the bottom working upwards. However number 9 follows number 16 twice in the NEXT draw.

The order of the balls in each result does not make any difference to the counting process.

Does this make it any clearer?

 

[doug_w]

Apologies Frank

Firstly it is important that the most recent draw is at the top, and has the highest draw number. The oldest draw is at the bottom and has the lowest draw number. So in the example you give, draw number 12 would be at the top (42,36,41,18,31,28). Draw number 1 at the bottom is 42,22,35,16,37,24.

The draw you refer to as draw 10 is actually draw 3 and what you called the previous draw, ( draw 9) actually is later (draw 4) so 42 follows 36, not the other way around.

You also said " looking at these results the number 36 follows 42 in the first draw." No it doesn't.
All the numbers in the last draw ( remember this is the most recent draw, at the top) were drawn in the SAME draw, so you can't say a ball follows another in the same draw - we are looking for which balls follow named balls in the NEXT draw - not the same draw! This is so that after the number crunching is over, we can infer which numbers will follow the most recent draw result , in the next draw to come, in the future.

In reality, number 36 doesn't follow number 42 at all in the NEXT draw, starting from the bottom working upwards. However number 9 follows number 16 twice in the NEXT draw.

The order of the balls in each result does not make any difference to the counting process.

Does this make it any clearer?

I had worked that example on an online website which goes from draw 1 on top to current draw on bottom.

The excel spread sheet I created from your instructions follows your rules.

Ok so where the 9 follows the 16 in the next draw (in the draw results I posted) is counted. I just wanted to confirm that you do count it this way and not just where it follows a ball in the same draw .

Cheers

Doug


I just reedited this post as I have noticed that the way they do it on the website is slightly different to the way you are doing it. I need to do further investigation.

 

[Frank]

Never when it follows a ball in the same draw, Doug, just to be clear.

 

[doug_w]

You are online now Frank ?

Never when it follows a ball in the same draw, Doug, just to be clear.

Are you in the forums at this moment

 

[doug_w]

Never when it follows a ball in the same draw, Doug, just to be clear.

OK Got it in my thick skull now Frank

 

[Frank]

I am just now, I'll be off soon

 

[Frank]

I just reedited this post as I have noticed that the way they do it on the website is slightly different to the way you are doing it. I need to do further investigation.

I would not advise looking at anyone else's methods whilst using my spreadsheet. I made it clear at the beginning of this thread that this is my interpretation of Markov chains and uses my methods. Trying to mix them is bound to lead to confusion.

 

[doug_w]

I would not advise looking at anyone else's methods whilst using my spreadsheet. I made it clear at the beginning of this thread that this is my interpretation of Markov chains and uses my methods. Trying to mix them is bound to lead to confusion.

Thanks Frank, will stick with your interpretation

 

[doug_w]

Hi Frank

Could you point to any documentation you may have read in coming up with your interpretation of Markov Chains. I have looked at some of the pdf's put out by some Universities, but they go way over my head.

I need the Dummies Markov Chains pdf

 

[Frank]

Well the problem here Doug, is that there are markov chains and there are lotteries. As you probably discovered, the Markov process is a mathematical matrix method of calculating probabilites where there is some cause and effect between a starting condition and the next thing to happen. Any course on Markov will talk you through tree diagrams of probabilites and then put the figures into Matrices. If you have not done Matrix calculations before it can get a bit heavy.

However I'm not calculating the probability that if a number six is drawn today what is the probability that number 14 will be drawn next time ?. Lotteries are random so there is no actual relationship between a ball drawn today and a ball drawn tomorrow. That complicated mathematics would be pointless.

If you want to learn it, be my guest :- http://techeffigytutorials.blogspot.co.uk/2015/01/markov-chains-explained.html

And there are lots of Youtube videos which would talk you through it, but they will not be talking about lotteries, so what's the point ?

All you need to read are the first six lines on the page I just linked to, in fact I will quote them here:-

Markov Chains is a probabilistic process, that relies on the current state to predict the next state. For Markov chains to be effective the current state has to be dependent on the previous state in some way; For instance, from experience we know that if it looks cloudy outside, the next state we expect is rain. We can also say that when the rain starts to subside into cloudiness, the next state will most likely be sunny. Not every process has the Markov Property, such as the Lottery, this weeks winning numbers have no dependence to the previous weeks winning numbers.

Bearing in mind the above, one can still ask the question " which ball(s) were drawn the most in the past which followed named ball ?

That's all my programme does. It answers that question. I does not work out probabilities, but there just might be a physical reason why balls 12,17,27,44 follow ball number 15 the most over the past 600 daws. You dont need maths to count them, you need a computer. If you think there is a reason for this result you can use that information to your advantage. That's all I'm doing Doug. Its not actual Markov, its it's cousin.