# Examining Ion Salui

#### Virbatem

##### Member
http://www.saliu.com/bbs/messages/638.html

The above web page delves into a two part system promoted as a 'possible killer lotto system' featuring selecting a 'favourite' number based on skips. Read the web page for more details.

After creating a program to analyse average and meadian skips for a drawn number over all previous draws of a lottery history I was disappointed to discover the variation in results were limited. For a 45 number lottery the range was from 6 to 8 and a 49 number lottery ranged from 6 to 9 with only one result being 6.

eg: a number has a summed total of skips which, when divided by the total number of times it was drawn, shows the average of the skips.

Increasing the detail of the results to include 2 decimal places did not help to show any immediately usable value in the numbers. Following a few hours of pondering I came to realize nearly the same results can be achieved with the division of the total number of games in a lottery history by the total number of times a number has been drawn.

eg: in a 6/45 lottery with 907 draws: #8 is found to have an average skip size of 6.34 by summing all the skips sizes and dividing by the number of times it was drawn: 139

It can also be calulated that 907 draws / 139 total occurrences equals 6.52: a very minor difference.

The maximum difference between the two calculations was 0.33.

So if you are able to use the average skip of a number then simply divide the total draws by the total times drawn for that number.

#### thornc

##### Member
You got the name wrong! "Ion Saliu"!

Now, what's interesting about using median instead of averages is that you now that 50% of the time something happens! The average doesn't tell you anything usefull like that!

In fact better than the median, is the 90 percentile!

Have you tried findling into the SD of the skip of each number? Making the calculation you can easily find beetwen which skip the number has hit 95% of the times!

This is something that I have been wanting to do, but never did!

#### andrew3698

##### Member
Personally I did not find Ion's 'killer strategy' to be a very profitable proposition. After partitioning the lotto field, I still had to deal with rather large pools of lotto numbers. To refine the strategy you need to define filters on smaller subsets of the lotto field.

It is unclear whether one should use the entire draw history of a lotto game, to draw conclusions about the skip patterns of individual lotto numbers. This is because the ball-picking machines and the ball sets are most likely to have been changed over a period of 900 draws.

I have some experience in playing the (6+B)/45 type of lotto game. May I suggest the following idea.

Examine the longest skip for each lotto number in the last 100 draws. Sort the skip frequencies from highest to lowest. This means the "hotter" numbers will be found at the lower end of the tabulation. Partition the field into three equal or near-equal sets (15-15-15, 16-16-13 etc). For simplicity label them A, B and C.

Now you have three pools to wheel, namely,

AB , AC or BC

Investigate the spread of the winners for the last 50 to 70 draws or more. You may be pleasantly surprised to find a high frequency of five to seven winners in AB, AC or BC.

The drawback is, you would still be wheeling large pools of numbers. For example, today I decided to pick 3 to 4 numbers from A and 2 to 3 numbers from B. The ranges were correct because AB contained 6 winners in it. I used Nick's Lotto Designer program to wheel out 36 lines using conditions that proved to be totally correct. Yet I had only two sets of winning triples embedded within the blocks.

All the best in your analysis!

#### Virbatem

##### Member
Thanks for the input thornc,

Median done for Canada 6/49 producing a range from 5 to 7.

SD done producing a range of 9.16 to 12.97.

I also tacked on a Max Skip for each number and, wow, #7 had a max skip of 80! The lowest max skip is shared between #41 and #44 with a period of 32.

Tell me how a 90 percentile is calcualted and I'll prolly add that in too.

So I am now compiling a huge assortment of statistics for each number of the entire history of the lottery. Who wants to sort it all out and organize it into some form of basis for a stategy?

thornc said:
You got the name wrong! "Ion Saliu"!

Now, what's interesting about using median instead of averages is that you now that 50% of the time something happens! The average doesn't tell you anything usefull like that!

In fact better than the median, is the 90 percentile!

Have you tried findling into the SD of the skip of each number? Making the calculation you can easily find beetwen which skip the number has hit 95% of the times!

This is something that I have been wanting to do, but never did!

#### Virbatem

##### Member
Thanks for that suggestion andrew,

It's an interesting method. am not sure I can smoothly add it to the program I have in it's current form. The program is purely devoted to sorting numbers based on some statistic (which is what you suggest) and can be divided into three pools. But applying a wheel would require some alterations to the program... putting me back some few steps and many days in alterations. But I will keep what you have written in mind.

By the way, My local lottery has 6/45 with 2 bonus numbers.

See the Saturday Lotto section of the below link:

http://www.lotterywest.wa.gov.au/asp/index.asp?pgid=135

andrew3698 said:
Personally I did not find Ion's 'killer strategy' to be a very profitable proposition. After partitioning the lotto field, I still had to deal with rather large pools of lotto numbers. To refine the strategy you need to define filters on smaller subsets of the lotto field.

It is unclear whether one should use the entire draw history of a lotto game, to draw conclusions about the skip patterns of individual lotto numbers. This is because the ball-picking machines and the ball sets are most likely to have been changed over a period of 900 draws.

I have some experience in playing the (6+B)/45 type of lotto game. May I suggest the following idea.

Examine the longest skip for each lotto number in the last 100 draws. Sort the skip frequencies from highest to lowest. This means the "hotter" numbers will be found at the lower end of the tabulation. Partition the field into three equal or near-equal sets (15-15-15, 16-16-13 etc). For simplicity label them A, B and C.

Now you have three pools to wheel, namely,

AB , AC or BC

Investigate the spread of the winners for the last 50 to 70 draws or more. You may be pleasantly surprised to find a high frequency of five to seven winners in AB, AC or BC.

The drawback is, you would still be wheeling large pools of numbers. For example, today I decided to pick 3 to 4 numbers from A and 2 to 3 numbers from B. The ranges were correct because AB contained 6 winners in it. I used Nick's Lotto Designer program to wheel out 36 lines using conditions that proved to be totally correct. Yet I had only two sets of winning triples embedded within the blocks.

All the best in your analysis!

#### Virbatem

##### Member
Hi again thornc,

If you wish to examine what I've done please go to:

http://www.iinet.net.au/~htjs/lottoeval

The full download does NOT contain the statisical analysis we have been discussing. This lottoeval.msi file is required if you have not previously downloaded the program.

If you have taken time to look at the program previously then you can auto extract the version 3.0.4 update and examine all the stats.

nearly 1am here, going to bed.

[I wonder why the bottom of the forum pages states all times are within my prefered timezone and then proceeds to tell me it is currently 07:31 PM?]

thornc said:
You got the name wrong! "Ion Saliu"!

Now, what's interesting about using median instead of averages is that you now that 50% of the time something happens! The average doesn't tell you anything usefull like that!

In fact better than the median, is the 90 percentile!

Have you tried findling into the SD of the skip of each number? Making the calculation you can easily find beetwen which skip the number has hit 95% of the times!

This is something that I have been wanting to do, but never did!

#### johnph77

##### Member
thornc -

Before I launch myself into a long extensive response, in your reference to the 90th percentile are you referring to positional frequency?

ty & gl

john

#### thornc

##### Member
What I'm refering as 90th percentile is just like the median is the 50th percentile!

Let's take a lotto number, say #1 on a given lotto... we construct it's skip line and get something like this (from my head, for illustrative purposes only!):
3 0 5 2 0 10 4 8 1 3
We sort it to get:
0 0 1 2 3 3 4 5 8 10

The median (50th) percentile is: 3
The 90th percentile is: 8
The average is: 3.6
The Standard Deviation is: 3.3

So I can say:
1) 50% of the times #1 has hit within 3 skips
2) 90% of the times #1 has hit within 8 skips!
3) 68% of the times, #1 has hit within
0.3 and 6.9 skips!

Can this constitute a strategy?
Maybe, I haven't given it a real test! Still haven't had the time!!

#### johnph77

##### Member
thornc -

ty - no response from this end, then. Interesting theory, though.....

gl

john

#### thornc

##### Member
How do I get this update?

Virbatem said:

If you have taken time to look at the program previously then you can auto extract the version 3.0.4 update and examine all the stats.

##### Member
Virbatem said:
>>>>[I wonder why the bottom of the forum pages states all times are within my prefered timezone and then proceeds to tell me it is currently 07:31 PM?]<<<<

#### system13

##### Member
thornc said:
How do I get this update?