United States
Member #37,169
April 11, 2006
341 Posts
Offline
I downloaded your wonderful work but am unable to put numbers in where it says use dropdown box to put in numbers. I appreciate any guidance you can provide
United States
Member #62,270
June 23, 2008
204 Posts
Offline
Quote: Originally posted by fashohits on May 31, 2019
I downloaded your wonderful work but am unable to put numbers in where it says use dropdown box to put in numbers. I appreciate any guidance you can provide
I assume you're referring to the Root Sums input. The options are ALL, LIST, and 1 through 9 individually. To put in a custom list, select LIST and input a string of numbers you'd like to evaluate in cell U12.
By string I mean, for example, input 14589 to have the macro include only Root Sums 1, 4, 5, 8, and 9.
United States
Member #59,352
March 13, 2008
5,626 Posts
Offline
Quote: Originally posted by mitachoo on May 30, 2019
Greetings everyone! I’ve been working on a system for Pick 3 based entirely on trends of “digits common to the previous drawing”, and while I’ve come up with a nifty tool, I’m struggling with how exactly to use the information the tool spits out to narrow down the field for the next drawing. Every strategy I’ve come up with is ALMOST, but not quite, repeatable.
The premise of the system is relatively simple: narrow the field of possibilities for the next drawing to combinations that meet the requirements of three flavors of “digits common to the previous drawing” filters. These filters are:
CWP (the number of digits in common with the previous Pick 3 drawing)
CWX (the number of digits in common with the previous Pick 4 drawing)
OUT (the number of digits in common with a list of numbers that have not appeared in the last 2 draws)
All three filters range from 0 to 3 and I concatenate them into a Structure Code that looks like, for example, 1-2-0. Using this example, for a Pick 3 combination to be deemed “viable” for the next draw, it must have exactly 1 digit in common with the previous Pick 3 draw, exactly 2 digits in common with the previous Pick 4 draw, and exactly 0 digits on the “Out Numbers” list.
With three filters that range from 0 to 3 there are 64 possible combinations but I limit my analysis to only 20 of the 64. Based on backtesting in my state (Connecticut), these 20 are the most frequently occurring Structures and account for 90% of all hits in the last 10 years (I initially tried tracking all 64 but the Excel Macro I use to do all the legwork did NOT like it). Here’s the list of 20:
Now I run a Macro that calculates the number of combinations that are “viable” for each of the 20 Structures, and further breaks down how may combinations each of the 9 Root Sums contribute to the total for each Structure. See the Matrix below (only the top 5 Structures are shown, the Matrix goes off to the right to cover all 20 Structures). What does this Matrix say? For example, Structure 1-1-1 has a total of 144 combinations in its bucket, with 18 coming from combinations that have a Root Sum of 1, 36 from a Root Sum of 2, etc. Or you can read it horizontally: combinations with a Root Sum of 1 have 18 in the 1-1-1 bucket, 15 in the 1-0-1 bucket, etc. (Side note: the number 937 highlighted in blue is a cross-check to confirm the Root Sum and Structure buckets add up to the same number, it’s not 1,000 because I used 20 Structures instead of the full 64 as noted above).
Now what the %&$$@!! do I do with this? My first thought (and the reason I broke down by Root Sum) was to narrow down the field by my making an educated guess as to the 2 to 3 most likely Root Sums to be drawn next, then re-create the Matrix to see which Structure(s) were most likely to hit on the next draw based on the reduced fields. Taking the matrix above and re-running for Roots Sums 1, 3, and 5, here’s what we get:
Now it gets interesting. If I’m convinced Root Sums 1 or 3 will come out, and if I’m further convinced Structure 1-1-1 will come out, 30 combinations (18 + 12) will cover all possibilities. And the numbers get better as you go off to the right of the Matrix (the Matrix is sorted by the number of combinations generated by each Structure, descending, left to right) . . .
. . . but now I’m stuck. Looking for any insights as to how this may be useful, other than the obvious (play the top 1 – 3 Structures and/or the top 1 – 2 Root Sums). Any thoughts?
PS - If anyone would like to tinker with my Excel File feel free to let me know, I'm more than happy to share.
"Every strategy I’ve come up with is ALMOST, but not quite, repeatable." That's because the
game is random. If you do the math you will find that the results are a product of the matrix
and nothing more. For this method to be useful the data needs to be predictable. Your first
step should be to test the randomness of the 3-digit patterns to see if they can be predicted
to some degree, you must consider that all64 patterns are in play every game. Years ago
winsum asked about writing a program which was based on an idea he had and a old filtering
method that I had used in a older program. The filter was called "Stacks" at first but has been
renamed several times over the years. That tool does what you are doing here but expands
it to 20 levels. The user could set a range for how many numbers/ digits would repeat from the
previous game, then the previous previous game etc all the way up to 20 games. The data to
the right shows the history for each stack as to the number of digits/numbers came from those
draws. Steve and I had different views on the end product, his being it was a great filter which I
agreed but I had my doubts as the data was not predictable enough. I offered the tool here at LP
several times as a additional filter packaged with other tools but it never got much attention. The
historic data looks good but I found it almost impossible to predict without setting the stacks to
cover a wide range. The user could set any numberof the stacks or bypass as needed looking for
the ones that showed / looked the most predictable forthe next game. Each setting allowed a
range to be set. For example, lets say the first positionwas thought to be a 1 or 2 then both could
be played by setting the low range to 1 and the highrange to 2.
Bottom line.
Unless the randomness breaks down when the method is applied then it's possible that the
odds will increase not decrease. Please don't think that I am trashing your attempts here,
I just think that you are not considering the combined odds for each step in the process. I am
getting old and the gray matter is not what it use to be but take it from someone who has
traveled down similar roads many many times. If the data is not predictable the method will
United States
Member #62,270
June 23, 2008
204 Posts
Offline
"Please don't think that I am trashing your attempts here"
Not at all RL, I appreciate your input! There's a reason those of us who have been part of LP for 10+ years are STILL here . . . that <snip> randomness will NOT be foiled!
This post has been automatically changed by the Lottery Post computer system to remove inappropriate content and/or spam.
United States
Member #62,270
June 23, 2008
204 Posts
Offline
Quote: Originally posted by RL-RANDOMLOGIC on May 31, 2019
"Every strategy I’ve come up with is ALMOST, but not quite, repeatable." That's because the
game is random. If you do the math you will find that the results are a product of the matrix
and nothing more. For this method to be useful the data needs to be predictable. Your first
step should be to test the randomness of the 3-digit patterns to see if they can be predicted
to some degree, you must consider that all64 patterns are in play every game. Years ago
winsum asked about writing a program which was based on an idea he had and a old filtering
method that I had used in a older program. The filter was called "Stacks" at first but has been
renamed several times over the years. That tool does what you are doing here but expands
it to 20 levels. The user could set a range for how many numbers/ digits would repeat from the
previous game, then the previous previous game etc all the way up to 20 games. The data to
the right shows the history for each stack as to the number of digits/numbers came from those
draws. Steve and I had different views on the end product, his being it was a great filter which I
agreed but I had my doubts as the data was not predictable enough. I offered the tool here at LP
several times as a additional filter packaged with other tools but it never got much attention. The
historic data looks good but I found it almost impossible to predict without setting the stacks to
cover a wide range. The user could set any numberof the stacks or bypass as needed looking for
the ones that showed / looked the most predictable forthe next game. Each setting allowed a
range to be set. For example, lets say the first positionwas thought to be a 1 or 2 then both could
be played by setting the low range to 1 and the highrange to 2.
Bottom line.
Unless the randomness breaks down when the method is applied then it's possible that the
odds will increase not decrease. Please don't think that I am trashing your attempts here,
I just think that you are not considering the combined odds for each step in the process. I am
getting old and the gray matter is not what it use to be but take it from someone who has
traveled down similar roads many many times. If the data is not predictable the method will
fail.
RL
" . . . the user could set a range for how many numbers/ digits would repeat from the
previous game, then the previous previous game etc all the way up to 20 games"
I was originally thinking along those lines and at one point I had 10 flavors of each of my three filters, looking at digits in common to draws 30, 60, 90, etc, days back, but I ended up ditching it to simplify the analysis.
MA United States
Member #171,823
January 12, 2016
255 Posts
Offline
Quote: Originally posted by RL-RANDOMLOGIC on May 31, 2019
mitachoo
I made a few updated to the stack program and ran a mock setup setting 5 stacks to single values
for my pick-3 game. The program reduced the matrix from 1000 to 36 lines and the second picture
shows the Ultra-34 check option with the results. Not bad at all provided one could hit the stack
values.
RL
Here are the results of the 36 lines, 1ea straight and 5-box hits.
to RL so u are saying you run your stacks and then the values u get is what you feed into the rogues on ultra? or your first program gave you the 36 numbers that dont have to be fed into ultra 34?
United States
Member #62,270
June 23, 2008
204 Posts
Offline
Ah, fudruckers. As I was re-doing my spreadsheet to add linkage to Cash 5 I found an error. The OUT filter was initially intended to capture Root Sums that were out for the last two draws; being Root Sum based, 0 was muted. In other words, as configured now, 0 will never be flagged as an Out number, even when it is.
The fix is simple; copy the formula from cell GL7 into cells GK7 to GK734.