The ideal number of past draws for systemsPrev Topic Next Topic

New Topic New Poll

What is the "sweet spot" in the number of past draws when performing any type of analysis?

All of the game history

[ 9 ]

[ 56.2% ]

The last x years

[ 2 ]

[ 12.5% ]

The last x months

[ 0 ]

[ 0.0% ]

The last x weeks

[ 1 ]

[ 6.2% ]

The last x days

[ 2 ]

[ 12.5% ]

The last draw

[ 2 ]

[ 12.5% ]

Total valid votes

[ 16 ]

Discarded votes

[ 0 ]

hypersoniq

Pennsylvania
United States
Member #1,340
April 6, 2003
2,939 Posts
Offline

Oct 10, 2024, 7:17 am

Each system is different, but for those following past draws for prediction or forecasting, how many past draws do you work with?

I have usually worked with full histories, but that seems to water down any recent trends.

Have an EXCELlent day!
LottoBux

25

Ontario
Canada
Member #109,236
April 9, 2011
61,309 Posts
Offline

Oct 10, 2024, 7:31 am

Quote: Originally posted by hypersoniq on Oct 10, 2024
Each system is different, but for those following past draws for prediction or forecasting, how many past draws do you work with?

I have usually worked with full histories, but that seems to water down any recent trends.

Guess it depends on what lottery you're playing

Pick 3 /Pick 4 , 14/21/30 draws.

I don't play Jackpot games often enough to track nowadays but it used to be 50 draws minimum
tokecap

Sao Paulo
United States
Member #156,153
June 10, 2014
11,639 Posts
Offline

Oct 10, 2024, 11:57 am

I think it depends on which lottery you are playing and what strategy you want to apply.

TN
580 058 13 3 4 8 141 10/08/24 Morn

1
5 8 0 ( +1 and mirror)
5
-----------------------
= 551 (10/10/24 Morn)

TX
919 199 19 9 1 8 525 10/09/24 Nite

9 1 9 (mirror 1=6 and 9+1 = 0)
6 0
----------
= 690 (10/10/24 Morn)
Pick3geniu$

New York
United States
Member #203,701
January 14, 2020
380 Posts
Offline

Oct 10, 2024, 12:48 pm

I use patterns as a basis for determining what group of numbers to play next, but patterns are short-lived which means that the most recent draws are used right up to the current draw. However, this isn't enough to go on to warrant a play.
Stat$talker

700 light yrs West of Milky Way Galaxy's Center
United States
Member #200,634
September 1, 2019
5,506 Posts
Offline

Oct 10, 2024, 4:55 pm

Quote: Originally posted by hypersoniq on Oct 10, 2024
Each system is different, but for those following past draws for prediction or forecasting, how many past draws do you work with?

I have usually worked with full histories, but that seems to water down any recent trends.

"how many past draws do you work with?"

Azzz meny azzz da Math sayzz to..

Don't be like dat cotton-pickin guy in KHaLiPhOrNyA..$pendz a ton of time advising Playaz how to bet, but cain't win a Pick3 game himself...
Cmoore50

United States
Member #194,171
December 3, 2018
621 Posts
Offline

Oct 10, 2024, 6:10 pm

You need a 1000 draws or more for jackpot games. Track them in some type of group. My patterns are based on 20 draws back. So every draw I use 19 draws of data and try to predict what the 20th draw is going to look like.

GET IT DONE
BobP

Dump Water Florida
United States
Member #380
June 5, 2002
3,678 Posts
Offline

Oct 11, 2024, 1:32 am

Use draw history from the last time they threw out the old ballsets and replaced them with new balls and or machines.

Keep in mind they swap out the balls and machines every draw, so you have to be very specific what you're asking should you email the lottery and ask, if they say last draw, it's more likely that's the swap out and they're playing with you.

BobP
Kola

Rookie Time Traveler

Milky Way Spiral
United States
Member #28,944
December 25, 2005
1,657 Posts
Offline

Oct 15, 2024, 11:53 pm

Quote: Originally posted by hypersoniq on Oct 10, 2024
Each system is different, but for those following past draws for prediction or forecasting, how many past draws do you work with?

I have usually worked with full histories, but that seems to water down any recent trends.

Hello hypersoniq,

Pick 3 game player here. I mostly use the last two draws to make my predictions. I lean heavily on the idea that the most recent draws are reflections or current snapshots of a game's recent trends and history as it circles through time, so using just two draws works well. Occasionally, I may use the last three or more, but never beyond the circle of the last 6 draws. If I played the higher matrixed games, I would "generally" apply the same strategy.

The Ancient Lottery LoreKeeper replied, "A past draw lies at the centerpoint of a circle - a black hole. Two overlapping circles whose centerpoints sit on the circumference of the other share a Common Radius. This Radius is a Wormhole, a Time Machine. It allows the past draw (info) to safely travel into and beyond the present draw's event horizon without negative distortion, thereby creating an entanglement (synchrony) between the two communicating draws. This entangled space or portal is the Vesica Piscis - the white hole or Stargate through which the next draw is born."
hypersoniq

Pennsylvania
United States
Member #1,340
April 6, 2003
2,939 Posts
Offline

Oct 16, 2024, 10:15 am

Quote: Originally posted by Kola on Oct 15, 2024
Hello hypersoniq,

Pick 3 game player here. I mostly use the last two draws to make my predictions. I lean heavily on the idea that the most recent draws are reflections or current snapshots of a game's recent trends and history as it circles through time, so using just two draws works well. Occasionally, I may use the last three or more, but never beyond the circle of the last 6 draws. If I played the higher matrixed games, I would "generally" apply the same strategy.

Interesting!

I would say that of all the games available here in PA that the pick 3 is my absolute favorite. Mostly because I can paper play and not be bothered if I missed a win.

My approach, mostly spreadsheets (for the last 20 years) and Python scripts (a more recent upskill) has always considered the full draw history because it is not really much more labor intensive to process the whole history than it is to process a few draws. Even when working with short term approaches, having the whole history available allows for back testing... and that is where systems usually fall apart, they were not profitable over the long term.

There were stretches of up to 5 years at a time where I played nothing, but kept the history file and tried ideas with failed back tests. I am probably too cheap to be a legit contender, but the things one learns along the way help out in other areas. Lately I have been running live tests for a few weeks here and there, but I lack the persistence to continue when there are no positive results.

I will occasionally go through ideas on the lottery systems thread, figure out how to automate the workouts and just run that against the history file, did that once with vtracs a few years ago... nothing seems to be profitable over time. But I have always kept the columns separated in pick 3 (and other games) because my hypothesis is that the pick 3 is just 3 consecutive games of 1 in 10 and each position is an independent variable. It was only last week that I started applying what was learned so far to combos. The results of things like frequency (which I feel well versed in) and now adding permutations (uncharted territory for me) is starting to turn up results that nearly satisfy back tests for profitability, but more work needs to be done. I used to not play box because it meant I was wrong with the straight guess, but I am seeing the value in spending the same dollar as a 0.50 straight/ 0.50 box.

It is interesting to see the shorter duration strategies.

Have an EXCELlent day!
dixiejoelotto

Itajai
Brazil
Member #216,592
July 24, 2021
4 Posts
Offline

Oct 16, 2024, 11:36 am

Very good insights, hypersoniq!

Thanks a lot!

I still need to try and learn Python. Newbie to Python.

I mostly like to use Excel but I am working on a software for Pick3 (Brazilian draws - Rio de Janeiro, mainly) with a lot of filters built-in.

I learned how to program in Delphi for this task. I am not a programmer. But can work out for my study needs.

Would appreciate to exchange some ideas regarding your combo approach. And also regarding frequency analysis.
hypersoniq

Pennsylvania
United States
Member #1,340
April 6, 2003
2,939 Posts
Offline

Oct 16, 2024, 3:01 pm

Quote: Originally posted by dixiejoelotto on Oct 16, 2024
Very good insights, hypersoniq!

Thanks a lot!

I still need to try and learn Python. Newbie to Python.

I mostly like to use Excel but I am working on a software for Pick3 (Brazilian draws - Rio de Janeiro, mainly) with a lot of filters built-in.

I learned how to program in Delphi for this task. I am not a programmer. But can work out for my study needs.

Would appreciate to exchange some ideas regarding your combo approach. And also regarding frequency analysis.

"Would appreciate to exchange some ideas regarding your combo approach. And also regarding frequency analysis."

Yes, we should definitely do this!

For frequency, it is as simple as counting the appearance of each ball in each position, so for example, 0 through 9 in a pick 3.

Since random numbers in that domain tend to exhibit a uniform frequency, we end up with an expectation that the total appearances of each digit would be 10%. In actual draw history analysis, we see that some numbers do in fact come up more than others, but not by a great amount. You might see that over 1,000 draws a 9 comes up 125 times (over 10%), but a zero comes up 98 times (under 10%).

Note, I am of the belief that the numbers themselves hold no mathematical properties, and can just as easily be represented by letters or pictures on the balls. They are selected from a discrete set, so you would never, for instance, see a drawn number of 4 and a half.

For moving beyond the whole set frequency, I have also tried the follower frequency, which is to say that, IF a 6 was drawn, how many times was each possible number drawn after it.

Moving into combinations, I am moving away from "which digit will be next in each spot?" to "which combination has historically paid out the most?"

I still believe that each digit is an independent variable which is why the set shows resistance to some machine learning processes, but when looking at combos, you have the concept of permutations, such that 456 as a straight is also a box win if it shows up as 465, 546, 564, 645 or 654. Similar concepts of frequency, but now I am facing the whole 1000:1 rather than 3 consecutive shots at 10:1.

Have an EXCELlent day!
dixiejoelotto

Itajai
Brazil
Member #216,592
July 24, 2021
4 Posts
Offline

Oct 16, 2024, 9:15 pm

Quote: Originally posted by hypersoniq on Oct 16, 2024
"Would appreciate to exchange some ideas regarding your combo approach. And also regarding frequency analysis."

Yes, we should definitely do this!

For frequency, it is as simple as counting the appearance of each ball in each position, so for example, 0 through 9 in a pick 3.

Since random numbers in that domain tend to exhibit a uniform frequency, we end up with an expectation that the total appearances of each digit would be 10%. In actual draw history analysis, we see that some numbers do in fact come up more than others, but not by a great amount. You might see that over 1,000 draws a 9 comes up 125 times (over 10%), but a zero comes up 98 times (under 10%).

Note, I am of the belief that the numbers themselves hold no mathematical properties, and can just as easily be represented by letters or pictures on the balls. They are selected from a discrete set, so you would never, for instance, see a drawn number of 4 and a half.

For moving beyond the whole set frequency, I have also tried the follower frequency, which is to say that, IF a 6 was drawn, how many times was each possible number drawn after it.

Moving into combinations, I am moving away from "which digit will be next in each spot?" to "which combination has historically paid out the most?"

I still believe that each digit is an independent variable which is why the set shows resistance to some machine learning processes, but when looking at combos, you have the concept of permutations, such that 456 as a straight is also a box win if it shows up as 465, 546, 564, 645 or 654. Similar concepts of frequency, but now I am facing the whole 1000:1 rather than 3 consecutive shots at 10:1.

Thanks for your explanations, hypersoniq

I agree with you in your considerations. Thank you!

I just did not understand this part:

Moving into combinations, I am moving away from "which digit will be next in each spot?" to "which combination has historically paid out the most?"

What is it that helps you consider which combination has historically paid out the most? Its frequency over the analysed period? Say how many times each combination (either straight or combo or the sum of both) has shown in the last 1000 analysed draws.

Or is it something different you consider in deciding which combination paid out the most?
hypersoniq

Pennsylvania
United States
Member #1,340
April 6, 2003
2,939 Posts
Offline

Oct 16, 2024, 10:13 pm

Quote: Originally posted by dixiejoelotto on Oct 16, 2024
Thanks for your explanations, hypersoniq

I agree with you in your considerations. Thank you!

I just did not understand this part:

Moving into combinations, I am moving away from "which digit will be next in each spot?" to "which combination has historically paid out the most?"

What is it that helps you consider which combination has historically paid out the most? Its frequency over the analysed period? Say how many times each combination (either straight or combo or the sum of both) has shown in the last 1000 analysed draws.

Or is it something different you consider in deciding which combination paid out the most?

The column analysis never seemed to be in synchronization, some numbers would come up, but the ones that were off were way off. I figured that looking at combos was going to trade solving the synchronization problem in exchange for lower frequencies... and that looks to be the result so far. The most popular combo in the PA evening pick 3 is 667, which happened 30 times in over 16,000 draws. To see how many times that may have hit adding $80 per 50 cent box brings the totals higher than just looking for a straight hit one column at a time. The least drawn combo has 5 appearances straight over the history. Because the top combos are so close together in number of hits, the data must be re checked frequently.

Have an EXCELlent day!
SunTribe

Zagreb
Croatia
Member #5,123
June 18, 2004
53 Posts
Offline

Oct 17, 2024, 2:30 pm

I think it's been over 20 years since Nick Koutras provided on another forum a mathematical formula for calculating the expected range of any lotto game. Personally, I did not notice any advantage of this expected range in my 5/50, 6/45 and 7/35 but maybe it can be of interest for some other format.

Range=(V+1)*(1-2/(K+1))

V - total numbers
K - numbers drawn

F.e.
5/50
Range=(50+1)*(1-2/(5+1))= 34.000 (34)

6/45
Range=(45+1)*(1-2/(6+1))=32.857 (33)

7/35
Range=(35+1)*(1-2/(7+1))=27.000 (27)

BR

"The Greatest Failure Of All, Is The Failure To Try!"
dr san

bgonÃ§alves
Brasil
Member #92,558
June 9, 2010
4,155 Posts
Offline

Oct 19, 2024, 5:18 am

Regarding previous draws, you have to take into account the following: analysis as a whole (all positions together) and by position, bet in ascending order, ok, e.g. number 32 is late, but in what position? We have delays with the position or neighboring number e.g. number (x) is late from the neighboring position, the seed for the next one is in the last draws

New Topic New Poll

Subscribe to this topic

The ideal number of past draws for systemsPrev TopicNext Topic

The ideal number of past draws for systemsPrev Topic Next Topic