Tyrone, PA United States
Member #198,217
May 15, 2019
231 Posts
Offline
Hi witty,
I have mixed good news looking at the PA backtesting. It started off like a solid gold lock that numbers that showed up in the mid-day TTT1 and mid-day 1-Off & Mirror TTT were coming up the following mid-day consistently. I saw that happen January 21, 22, 23, and 24th, so I'm thinking jackpot. But then when I went back beyond 1/21, it only happened one more time during the dates where backtest data is available. Still might be worth watching, but when I clicked on the link for today's PA predictions, the mid-day 1-Off & Mirror TTT fails to populate with data. The pop-up window comes up and the heading is there but no prediction data.
The mid-day 1-Off & Mirror Rundown for today does the same thing.
Out of the last 13 P3E drawings, numbers that appeared in both the TTT1 Evening and 1-Off & Mirror TTT Evening for the previous day's prediction only once. There was a handful of other combinations that happened, but nothing I'm seeing that showed investment-worthy consistency in the P3E, at least just focusing on those two prediction systems.
More later when I have time to dig deeper into other systems.
Honolulu, Hawaii United States
Member #3,263
January 6, 2004
2,253 Posts
Offline
Quote: Originally posted by OhSoClose on Jan 25, 2020
Hi witty,
I have mixed good news looking at the PA backtesting. It started off like a solid gold lock that numbers that showed up in the mid-day TTT1 and mid-day 1-Off & Mirror TTT were coming up the following mid-day consistently. I saw that happen January 21, 22, 23, and 24th, so I'm thinking jackpot. But then when I went back beyond 1/21, it only happened one more time during the dates where backtest data is available. Still might be worth watching, but when I clicked on the link for today's PA predictions, the mid-day 1-Off & Mirror TTT fails to populate with data. The pop-up window comes up and the heading is there but no prediction data.
The mid-day 1-Off & Mirror Rundown for today does the same thing.
Out of the last 13 P3E drawings, numbers that appeared in both the TTT1 Evening and 1-Off & Mirror TTT Evening for the previous day's prediction only once. There was a handful of other combinations that happened, but nothing I'm seeing that showed investment-worthy consistency in the P3E, at least just focusing on those two prediction systems.
More later when I have time to dig deeper into other systems.
That's a lot of analysis. Thanks for doing it.
As we all know most of the states follow one patter for few straight days and then switch. This might very well be case in your analysis also. We should wait till we get at least 1 full month of data available to see the pattern.
On the other note, part of this might also be coz' in those old TTTs, the date sum number might not be there, so even the workout has the combination it will not show up in back-testing results . Given TTT only has 9 digits max with repeating digits also, may be I should put all digits as key digits in TTT. After doing that , you might see some better success off those not working TTTs also. My initial guess, but I will look deep in it and if that is the case, I will remove the Date sum to get numbers for those TTTs to get better results.
Date Sum filter is good for other workouts where they produce big list of combinations but not that useful for simple TTTs in my opinion.
Tyrone, PA United States
Member #198,217
May 15, 2019
231 Posts
Offline
I agree on the question of following it for a month to see what patterns emerge. I'll continue to harvest the data daily to watch the patterns.
As to your datesum question, I admit I don't understand it well enough to offer an opinion. I read on your site about how the 1-Off & Mirror TTT works and even built a variable version of that system for myself in Excel. Couldn't get a consistent enough prediction system out of it to make it worth risking money on it though.
I have no idea how you get the data for the TTT 1.
As far as where I was seeing prediction data not populating, it was in the prediction post for today. A couple of those were coming up blank. It just updated after the mid-day drawing and they're working now, so unsure what happened. Should be interesting to see if those populate eventually in the backtesting data.
Honolulu, Hawaii United States
Member #3,263
January 6, 2004
2,253 Posts
Offline
Quote: Originally posted by OhSoClose on Jan 25, 2020
I agree on the question of following it for a month to see what patterns emerge. I'll continue to harvest the data daily to watch the patterns.
As to your datesum question, I admit I don't understand it well enough to offer an opinion. I read on your site about how the 1-Off & Mirror TTT works and even built a variable version of that system for myself in Excel. Couldn't get a consistent enough prediction system out of it to make it worth risking money on it though.
I have no idea how you get the data for the TTT 1.
As far as where I was seeing prediction data not populating, it was in the prediction post for today. A couple of those were coming up blank. It just updated after the mid-day drawing and they're working now, so unsure what happened. Should be interesting to see if those populate eventually in the backtesting data.
That momentarily blank out for latest workouts is expected behaviour around the time when the latest results of that particular state are published.
As soon as the latest results come for any state, the system re-builds a new workout based on latest winning number and replaces the older one.
It takes a while roughly around 5-8 minutes, during which time you will see for those states, the midday workout area shows blank. You will see that mostly every hour of the clock for e.g. between 1;1:10 PM for 1 PM, since i run the process every hour of the close to update results.
I fixed the TTT - No date sum issues. Now going forward those false triggers will not be an issue.
I am not fixing existing workouts and all existing back testing results as that's a hell lot of data cleanup effort for me , Almost like 50K workouts already present over the time (at least) LOL.
But the next future ones will be much more realistic TTTs as they are independent of DATE SUM limitation.
Give another 10 days, and we will have much better data to analyze from TTT perspective.
Once I am done these last few pending items, I am going to make it much easier for you to do analysis by providing some visual charts to look at for patterns finding in back testing data. That will make your pattern hunting job much easier. So stay tuned for new improvements soon in the system.
Tyrone, PA United States
Member #198,217
May 15, 2019
231 Posts
Offline
Quote: Originally posted by witty on Jan 25, 2020
That momentarily blank out for latest workouts is expected behaviour around the time when the latest results of that particular state are published.
As soon as the latest results come for any state, the system re-builds a new workout based on latest winning number and replaces the older one.
It takes a while roughly around 5-8 minutes, during which time you will see for those states, the midday workout area shows blank. You will see that mostly every hour of the clock for e.g. between 1;1:10 PM for 1 PM, since i run the process every hour of the close to update results.
I fixed the TTT - No date sum issues. Now going forward those false triggers will not be an issue.
I am not fixing existing workouts and all existing back testing results as that's a hell lot of data cleanup effort for me , Almost like 50K workouts already present over the time (at least) LOL.
But the next future ones will be much more realistic TTTs as they are independent of DATE SUM limitation.
Give another 10 days, and we will have much better data to analyze from TTT perspective.
Once I am done these last few pending items, I am going to make it much easier for you to do analysis by providing some visual charts to look at for patterns finding in back testing data. That will make your pattern hunting job much easier. So stay tuned for new improvements soon in the system.
That all sounds excellent. Any chance showing us the backtesting data on the PA Pick4 is on your timetable?
Honolulu, Hawaii United States
Member #3,263
January 6, 2004
2,253 Posts
Offline
New Addition: State level Back Testing Results enabled for all Pick 3 & Pick 4 games.
Hi Everyone,
The State Level Back Testing has been enabled for all Pick and Pick 3 workouts for the state of your choice.
In each of the State Workout page (Pick 3 & Pick 4), there will be a Green Button , that will take take the users to view how each of the State Level Workouts are performing in predicting the next winning numbers.
Please feel free to post any additional questions or information you need regarding the Back Testing results for all State Workouts.
Tyrone, PA United States
Member #198,217
May 15, 2019
231 Posts
Offline
Thanks, witty! I spent a couple hours today going through the Pick4 backtesting for PA. I identified 5 prediction systems that contained the winners for mid-day far more often than not and tried to search for a common subset of candidate digits to play. Unfortunately, the common subset contained an unworkably high number to play, so I'm exploring other angles.
Very useful system! Having the backtesting available really opens it up to exploration.
Honolulu, Hawaii United States
Member #3,263
January 6, 2004
2,253 Posts
Offline
Quote: Originally posted by OhSoClose on Jan 26, 2020
Thanks, witty! I spent a couple hours today going through the Pick4 backtesting for PA. I identified 5 prediction systems that contained the winners for mid-day far more often than not and tried to search for a common subset of candidate digits to play. Unfortunately, the common subset contained an unworkably high number to play, so I'm exploring other angles.
Very useful system! Having the backtesting available really opens it up to exploration.
Glad it is helping to you and others.
Like I said earlier, very soon I will try to add some additional tools and charts, so by few clicks itself half of your analysis part will be done. And you don't have to do lot of excelling (LOL for the word Excel'ing)
But it will be a while before I add those additional helpful tools. Being a new system (just a month old right now), first I am trying to wrap what all I targeted for the phase 1. Almost there but not yet done.
The back-testing was the biggest effort so far and most tricky to automate for all states and all workouts together.
Thankfully its done and hopefully not much glitches in the first round of its testing but let me know if you find any discrepancies so I can fix them as we find.
Honolulu, Hawaii United States
Member #3,263
January 6, 2004
2,253 Posts
Offline
Quote: Originally posted by OhSoClose on Jan 26, 2020
Thanks, witty! I spent a couple hours today going through the Pick4 backtesting for PA. I identified 5 prediction systems that contained the winners for mid-day far more often than not and tried to search for a common subset of candidate digits to play. Unfortunately, the common subset contained an unworkably high number to play, so I'm exploring other angles.
Very useful system! Having the backtesting available really opens it up to exploration.
You will have to apply additional overdue ROOT SUM and overdue TRIADS on top of common set for the Pick 4 combinations to bring down under 75 when it becomes profitable.
and play doubles only for better payout. for all states, double usually comes once every 3 draws and sometimes 3-5 doubles in a row.
Honolulu, Hawaii United States
Member #3,263
January 6, 2004
2,253 Posts
Offline
Quote: Originally posted by Rakster on Jan 25, 2020
Thank you.
@Rakster,
I have added Pick 3 and Pick 4 Workouts for Western Canada. You can find the link under Daily Workouts.
You might find a little delay in updating the results and workouts with latest numbers, since I will be updating the results manually for Western Canada every night till it is automated.