Ohio United States
Member #179,976
February 15, 2017
1,283 Posts
Offline
Nice work parlayman! Your method is looking for past draws that closely matches current draws, is that correct?
My numbers to play come from a matrix that combines various hit sequences, skips, and a few other pieces of data, then puts each number into a row/column in that matrix. The backtest then attempts to find the hot spots in the matrix for numbers to play. I usually select around 20 numbers to play, and play for straight wins only. To be honest, I have not played Pick 3 here in quite some time since my interest is more in our 5/39 game.
Just for fun, here are 22 quick numbers for tonight's OH Pick 3 that my program came up with:
bliss United States
Member #176,317
July 31, 2016
22,052 Posts Online
Quote: Originally posted by Mr-B 216 on Sep 6, 2021
Nice work parlayman! Your method is looking for past draws that closely matches current draws, is that correct?
My numbers to play come from a matrix that combines various hit sequences, skips, and a few other pieces of data, then puts each number into a row/column in that matrix. The backtest then attempts to find the hot spots in the matrix for numbers to play. I usually select around 20 numbers to play, and play for straight wins only. To be honest, I have not played Pick 3 here in quite some time since my interest is more in our 5/39 game.
Just for fun, here are 22 quick numbers for tonight's OH Pick 3 that my program came up with:
A 31-day frame of past draws that matches the current draws, yes.
I use the Lottery Post Predictions page scoring method to score to each frame and then pick the highest scoring frame with the most-recent high score breaking a tie.
I hear you about only playing straights. I'm going to try a mod to my Correlation program to only consider straight hits and see how that goes. Thanks for the idea!
UPDATE: The Ohio Evening draw was 661 following a Midday draw of 660, so my workup didn't find a hit. I ran tomorrow's reports and the Correlation workup did not budge, exact same frame as yesterday. I'd like to think that's telling me something, so I'm going to stick my neck out and try one more prediction choosing Correlation workup numbers by intersecting my other Ohio reports for tomorrow with the Correlation workup to get - 096 103 119 275 312 516 723 900 for Tuesday.
east cost United States
Member #209,336
August 29, 2020
113 Posts
Offline
Quote: Originally posted by parlayman on Sep 6, 2021
A 31-day frame of past draws that matches the current draws, yes.
I use the Lottery Post Predictions page scoring method to score to each frame and then pick the highest scoring frame with the most-recent high score breaking a tie.
I hear you about only playing straights. I'm going to try a mod to my Correlation program to only consider straight hits and see how that goes. Thanks for the idea!
UPDATE: The Ohio Evening draw was 661 following a Midday draw of 660, so my workup didn't find a hit. I ran tomorrow's reports and the Correlation workup did not budge, exact same frame as yesterday. I'd like to think that's telling me something, so I'm going to stick my neck out and try one more prediction choosing Correlation workup numbers by intersecting my other Ohio reports for tomorrow with the Correlation workup to get - 096 103 119 275 312 516 723 900 for Tuesday.
"I ran tomorrow's reports and the Correlation workup did not budge, exact same frame as yesterday. I'd like to think that's telling me something, so I'm going to stick my neck out and try one more prediction choosing Correlation workup numbers by intersecting my other Ohio reports for tomorrow with the Correlation workup"
Glad to see that now we have two people that have programmed this data gathering method by dates. Now it's just a wait to see what their opinion of this method is. It's great to have the creators opinion but when it comes to our own creations we will always be biased especially when we feel it's the best.
This is what i'd like to know.
prediction time span (1st bet and total bets until prediction wins)
how much wagered subtracted from how much won over (prediction time span)
If i keep taking the same actions, same results occur.
timing plus luck plus quality - quantity = winner's circle
bliss United States
Member #176,317
July 31, 2016
22,052 Posts Online
Quote: Originally posted by AnythngPossible on Sep 7, 2021
"I ran tomorrow's reports and the Correlation workup did not budge, exact same frame as yesterday. I'd like to think that's telling me something, so I'm going to stick my neck out and try one more prediction choosing Correlation workup numbers by intersecting my other Ohio reports for tomorrow with the Correlation workup"
Glad to see that now we have two people that have programmed this data gathering method by dates. Now it's just a wait to see what their opinion of this method is. It's great to have the creators opinion but when it comes to our own creations we will always be biased especially when we feel it's the best.
This is what i'd like to know.
prediction time span (1st bet and total bets until prediction wins)
how much wagered subtracted from how much won over (prediction time span)
The answers to your questions are a work in progress, but I'll say this. I have NO desire to lose money, so I have been observing this workup as we go trying to find any kind of "tell" that will put me on the track of a winning frame of numbers and/or give me a hint when to wager. The next step, of course, is to put this workup in a giant backtest loop and let it work it's way back through time picking the winning frame(s) each day and putting those in a bin for closer scrutiny. My approach so far has just been to prospectively acquire data without bias and to see what the norms are.
Open questions are as follows:
1) What is the ideal frame length? I blindly chose 31 days simply to see what a calendar month looked like. This is not fixed in stone.
2) Is it even worth looking at boxed hits? I chose yes JIK there was paydirt hiding in those, but knowing full well that a $41 dollar bet on a $40 win is a fool's errand.
3) How to handle numbers which have already hit? Should I look for repeats or should I exclude those from future predictions or just ignore them?
4) Most importantly, is there any possible "tell" in any of this? In the Date workups, we found that some states tend to draw straight Date workup hits in clusters a day or two apart. I keep hoping there might be an analogue of that strategy which could be applied to this workup.
5) Is there any possibility that a another workup intersected with this workup would yield a "tell?" You may have noticed that in my predictions for Ohio, I mentioned that I intersected my other reports with the Correlation report to see if that would yield a result. This is something I just started looking at, but I'll say that I less optimistic about this possibility from experience trying to intersect other workups. This method reduces losses because it produces smaller sets of numbers to play, but it never seems to produce a winner. If a win comes, it always seems to be in the larger original set.
6) Last and certainly not least, as you asked, When during a frame, do I jump in and start playing? Again, I have NO desire to lose money. That's why I'm looking at these trying to figure out when to step in (and of course, how to play - STR BOX 50/50). Also, this has direct bearing on whether or not to consider boxed workups in the scoring of frames. Should I or not?
There are a lot of variables here and we are shooting at a moving target, which makes things way more difficult. That's why I thought I would throw this workup out there where folks with real skills might take a look at it. And please don't be afraid to tell me if this is a bad idea that you have already explored. I'm a reasonable person and I have no desire to waste my time drilling dry wells.
P.S. Apologies to the OP for hijacking this thread. I just thought if ever there were folks with an interest in this workup, this thread would be the place to find them.
bliss United States
Member #176,317
July 31, 2016
22,052 Posts Online
Quote: Originally posted by parlayman on Sep 7, 2021
The answers to your questions are a work in progress, but I'll say this. I have NO desire to lose money, so I have been observing this workup as we go trying to find any kind of "tell" that will put me on the track of a winning frame of numbers and/or give me a hint when to wager. The next step, of course, is to put this workup in a giant backtest loop and let it work it's way back through time picking the winning frame(s) each day and putting those in a bin for closer scrutiny. My approach so far has just been to prospectively acquire data without bias and to see what the norms are.
Open questions are as follows:
1) What is the ideal frame length? I blindly chose 31 days simply to see what a calendar month looked like. This is not fixed in stone.
2) Is it even worth looking at boxed hits? I chose yes JIK there was paydirt hiding in those, but knowing full well that a $41 dollar bet on a $40 win is a fool's errand.
3) How to handle numbers which have already hit? Should I look for repeats or should I exclude those from future predictions or just ignore them?
4) Most importantly, is there any possible "tell" in any of this? In the Date workups, we found that some states tend to draw straight Date workup hits in clusters a day or two apart. I keep hoping there might be an analogue of that strategy which could be applied to this workup.
5) Is there any possibility that a another workup intersected with this workup would yield a "tell?" You may have noticed that in my predictions for Ohio, I mentioned that I intersected my other reports with the Correlation report to see if that would yield a result. This is something I just started looking at, but I'll say that I less optimistic about this possibility from experience trying to intersect other workups. This method reduces losses because it produces smaller sets of numbers to play, but it never seems to produce a winner. If a win comes, it always seems to be in the larger original set.
6) Last and certainly not least, as you asked, When during a frame, do I jump in and start playing? Again, I have NO desire to lose money. That's why I'm looking at these trying to figure out when to step in (and of course, how to play - STR BOX 50/50). Also, this has direct bearing on whether or not to consider boxed workups in the scoring of frames. Should I or not?
There are a lot of variables here and we are shooting at a moving target, which makes things way more difficult. That's why I thought I would throw this workup out there where folks with real skills might take a look at it. And please don't be afraid to tell me if this is a bad idea that you have already explored. I'm a reasonable person and I have no desire to waste my time drilling dry wells.
P.S. Apologies to the OP for hijacking this thread. I just thought if ever there were folks with an interest in this workup, this thread would be the place to find them.
The next step, of course, is to put this workup in a giant backtest loop and let it work it's way back through time picking the winning frame(s) each day and putting those in a bin for closer scrutiny.
If I were actually going to do this, how many days would you want to go back? My bias is always a year, 365 days, but, in the interest of time, I'm thinking about 100 days. The full output from 100 tests would already be quite a bit to look though and I'm thinking I'd like to have an answer while we're all still young. AnythngPossible, did you do anything like this? I'd be extremely interested to know how far back you backtested.
bliss United States
Member #176,317
July 31, 2016
22,052 Posts Online
Quote: Originally posted by parlayman on Sep 6, 2021
Interesting chart! My program doesn't make such a sophisticated chart, but after seeing your chart (for Ohio) I did a run of my correlation program on the Ohio data to see what it would find for Ohio today. The chart is below. The left-hand column is the current numbers and the right-hand column is the most-highly correlated frame (frame = 31 days) of Ohio numbers from the past. It's marked up showing the straight hits brightly highlighted and boxed hits lightly highlighted to make it easier to see what's going on. This method gives you a chance to see correlated frames developing (I have a giant file of all the correlated frames which meet or exceed a user-set threshold) so you maybe have a chance to spot a winner before it falls. Unfortunately, I think this chart has fizzled out as these things do, but I filtered my other Ohio workups for today (9/6) through the 2011 frame of numbers and came up with this small group of candidate numbers for tonight to compare to yours - 202 214 269 308 467 488 736.
bliss United States
Member #176,317
July 31, 2016
22,052 Posts Online
Quote: Originally posted by Mr-B 216 on Sep 6, 2021
Nice work parlayman! Your method is looking for past draws that closely matches current draws, is that correct?
My numbers to play come from a matrix that combines various hit sequences, skips, and a few other pieces of data, then puts each number into a row/column in that matrix. The backtest then attempts to find the hot spots in the matrix for numbers to play. I usually select around 20 numbers to play, and play for straight wins only. To be honest, I have not played Pick 3 here in quite some time since my interest is more in our 5/39 game.
Just for fun, here are 22 quick numbers for tonight's OH Pick 3 that my program came up with:
Ohio drew 897. That's a Straight Hit for your workup, Mr-B 216. You Win!!! Seriously, would your workup have also produced that result for today's draw? If so, that's very impressive!
east cost United States
Member #209,336
August 29, 2020
113 Posts
Offline
Quote: Originally posted by parlayman on Sep 7, 2021
The next step, of course, is to put this workup in a giant backtest loop and let it work it's way back through time picking the winning frame(s) each day and putting those in a bin for closer scrutiny.
If I were actually going to do this, how many days would you want to go back? My bias is always a year, 365 days, but, in the interest of time, I'm thinking about 100 days. The full output from 100 tests would already be quite a bit to look though and I'm thinking I'd like to have an answer while we're all still young. AnythngPossible, did you do anything like this? I'd be extremely interested to know how far back you backtested.
Yes i have tried something similar in excel but i went back as far as 360 draws combined and 720 draws combined 6 months and 1 year. I looked under the rock and i saw fragments of gold specks but not the nugget. I was more curious to see what you guy's find, Mr B and yourself. Testing out the original posters method through your lens will be a unbiased review. I have used cycle counts tied pairs and box combinations.
This poster's method in my point of view is an extend calendar date method that many use on LP but some folks may only go back 1 to 2 calendar years looking for combos that occurred back then on or around the same date +/-5 draws depends whom the user is using the calendar method.
You can use anything to look for certain combinations to meet your criteria like. Neighboring pairs, Flips, Mirrors, Ghost and the list goes on. The more you add the more your pool of combination numbers grow. I am hoping you or mr b looks under the rock and find the nugget that i may have missed.If so i will pick up were i left off some time ago.
Good Luck looking forward to you guy's reports
If i keep taking the same actions, same results occur.
timing plus luck plus quality - quantity = winner's circle
Ohio United States
Member #179,976
February 15, 2017
1,283 Posts
Offline
Quote: Originally posted by parlayman on Sep 7, 2021
Ohio drew 897. That's a Straight Hit for your workup, Mr-B 216. You Win!!! Seriously, would your workup have also produced that result for today's draw? If so, that's very impressive!
"Seriously, would your workup have also produced that result for today's draw?"
Well, yes and no. If I use the same filters day to day, the numbers to play pretty much remain the same for a few days at a time. Some numbers will drop off, some will be added.
The workup I did yesterday was done with next to no thought, and for Eve draw only. There was no data from the Midday draw being analyzed, so it's unlikely 897 would have been there. That's not to say those numbers would be unplayable for the Midday draw, I just wouldn't have been in on the Midday draw, so no win for me.
I guess it's time for me to revisit how my Pick3/4 program works. It's been a while so I'll need to refresh my memory a bit. For tonight's OH Pick3 Eve draw:
Ohio United States
Member #179,976
February 15, 2017
1,283 Posts
Offline
Quote: Originally posted by Speler on Sep 8, 2021
Why does this always end in endless posts about pick 3?
What do you mean by this? Your comment makes no sense. If you have no interest here and have nothing to contribute, feel free to ignore the thread and move on.