hypersoniq's Blog

Even though they give 2 quick pick lines, I am focusing on my picked line. Breaking the 28 draw window into one week steps, those QP lines will be changed every 7 days.

They let you play for up to 26 advance draws, but the 7 day / 4 week window gives 2 extra draws and it stays on the weekly budget.

I have, in the past, held 26 draw tickets. I did that once with follower data and hit $1k for 5 of 6, but it was on a QP line. For the purpose of this trial, I am only counting matches on my picked line, the neutrals. The N of 18 prizes are not being factored in... only counting the number of neutrals picked per draw. 

Any N of 18 hits will help defray the next week's purchase, but if the neutrals are not showing up, then it is time to look for a new way to pick from the script output.

This first test involved picking from the central transition, which is the median (Q2). Numbers were picked that had a frequency just below this level, and will cross over to the higher side of neutral if selected.

With 2 draws in, the first night had 0 matches on the line of interest, and last night my line matched 1. 

You need 3 on one line for $2, also having 4 of 18 also pays $2. No cash prizes hit in the first 2 nights.

Outside of the top tier hit of 6 out of 6, the pay table is kind of weak...

Top prize is the rolling jackpot (currently $830,000)

5 of 6 on one line only pays $1,000

4 of 6 on one line pays $20

3 of 6 on one line pays $2

The N of 18 game pays based on all 3 lines

10+ pays $2,500

9 pays $1,000

8 pays $50

7 pays $25

6 pays $10

5 pays $5

And 4 pays $2

If all 18 numbers on a ticket are unique, your ticket in the N of 18 can only be worth up to $10 on that part of the game (unless those 6 are all on the same line...)

The high N of 18 prize would require 5 of 18 numbers to have a match on the QP lines, which you do not get to pick. This is why I decided to go with a 4 week window (28 draws, changing the QP lines 4 times) rather than one $52 26 draw ticket. The N of 18 game is kind of like the double play on power ball, only they don't charge extra for it and the prizes are weak.

Don't get me wrong, any hit will be taken, but the absolute focus is on that line of neutrals I picked, because that is what I had to work with.

You need 4 pairs for a shot at the 10 of 18, I have never bought such a ticket, but there are the occasional winners of that prize. They DO pay on both versions of the game... if you have 3 on one line ($2) and match one other number on the other lines for 4 of 18 ($2), you win $4. When I hit that $1,000 they paid $1,025 because I matched 7 of 18 as well.

I know in advance the cost of this first attempt, $56... but not all up front. It replaces the $14/week that I was already using to play the pick 3. Also, any small prize hits along the way will lower that $56 cost.

2 draws in, no cash prizes. My current week ticket has 3 pairs, so highest possible N of 18 = 9.

Onward...

The "strategy" is play one combo for a week, the same combo for 28 draws (4 weeks) for a cost of $14 per week.

Why not just play for 26 draws at once? Because I get 2 new QP lines each week.

Best case scenario for the pick 3... IF day and night hit in the same week is $580 to $660. If this game works out, best case (min. Jackpot, post tax) is $299,650 (per winner).

Same basic concept.

Now if they will pick the right numbers for a change..

Got the draws out counter working on the Cash 4 life version of the script. This was more difficult because not all 60 numbers appear in each column. I had to wrap the counter in an if statement to first check if the value appears, then process as normal if it does, and replace with a dummy value (999) if it does not.

In this update I also added the display of neutrals per column.

Last update will be adding the chi square goodness of fit and P value... then porting to Match 6 will be as easy as changing the game parameter values (mere minutes of work).

I may be on my last pick 3 week for a change of focus to the Match 6 for the exact same budget.

Maybe it would be helpful to jot down what the components are and how they are obtained.

The first is the "model". Simple to put together, it is the last 150 draws pf the PA pick 3. There is a separate model for day and night.

Next is the classified distribution over that model. This is all digits 0 through 9. Looking at raw frequency, the count is the number of times each was drawn in that 150 draw model, separate columns per position. Along with that is the percentage of the whole... digit frequency divided by 150, then multiplied by 100. This shows at a glance how near or far the raw frequency was from the 10% expectancy. It is this percentage that is the basis for the other statistics. 

Quartiles are calculated for the percentages, with Q2, the median, always being at or near 10%. Those percentages that fall within the inter quartile range (Q1 to Q3) are classified as Neutral. Any percentage below Q1 is classified as Cold, and any percentage above Q3 is classified as Hot. This used to be one standard deviation below or above expectancy, but the inter quartile range is more stable and less influenced by outliers.

Finally, each digit has a number that is how many draws since it's last appearance. This is calculated by counting the rows from the bottom (newest) until that digit appears.

Seems straight forward enough.

Now, we look toward that 7 draw window in which we are trying to guess which all neutral numbers will appear. I wrote the script with a window parameter. When set to 0, it just gives the list above. When set to 7, it goes back 7 draws before calculating the model and then proceeds to classify the last 7 draws. After 1 week of play, the 7 draws are added to see how they were classified, and where I went wrong with the guess.

The original observation was that an all neutral draw happens at least once in any given 7 draw window about 70% of the time... back testing confirmed the observation.

We can not peer into the future. We still do not know which numbers will or will not be drawn, but we Do know which will drop off the list by the end of the week... it is the first 7 draws of 150... so the script will run the full 150, then run with a smaller model of 143. Both lists are printed, so we can see if a number changes classification at the transition points (hot to neutral, neutral to cold).

So then over the weeks of play, I started focusing on the Q2 transition area. Numbers slightly below expectancy transitioning to slightly above expectancy.

The current week focused on these transition numbers specifically. In simple form, numbers drawn less than 15 times that could come up and hit the 15 mark or even exceed it.

The good part... these numbers do end up appearing when looking over the next 7 draws

The bad part... they do NOT appear on the same draw.

Remembering that not only are draws independent of previous draws, they are also independent by position, I find myself stuck for how to tie them together.

I had learned through observation and back testing that when eliminating hots and colds as outliers, numbers in a random system tend to fill from the middle of the frequency range. This holds true for both RNG draws (PA Pick 3 MID) and ball draws (PA Pick 3 EVE). Yet even with this knowledge, it is still difficult or even impossible to get the right combo. (Outside of a single coincidental win, just like every other system so far.)

That is where I begin to look at an automated phase 2. Feeding these statistics into an agent that can analyze what data is present, based on the constraints provided by a Markov Decision Process. The problem I am having with this is... how do I set up rules when I can't even figure the steps out on my own?

I think it would be worthwhile to take a break from the Pick 3 again and brainstorm some new ideas. So, once I convert the Cash 4 Life script to work with the PA Match 6, I will switch to that for 4 weeks (28 draw window, 735 draw model) and drop the pick 3. Same cost ($14/week), and playing the same combo 4 times over 4 weeks will allow for 2 different QP lines each week.

Probably won't be ready for the change over for another week or 2, but I am coming up dry now trying to get that elusive second pick 3 win. Even though the Match 6 lower tier prizes are not very exciting, that minimum $500,000 top prize is better than I can get within the budget on any of the pick N games... and if that only works once, it is good enough!

Of course, since there is at least 1 or 2 weeks left to play Pick 3 before switching to Match 6, the all neutral QP generator could help out by reducing the number of neutrals loaded into it... just loading neutrals that are near that Q2 internal transition area. It is these numbers I notice most as showing up in the 7 draw window. If there is no deterministic way to select, just use an RNG to select from the leading candidate digits in each position... worth a shot!

Thinking about the wait on wins for the pick 3 brings up another option... the Match 6. 

If it were played in 7 draw increments, it would cost the same as the pick 3, $14 per week... a window of 28 days is just playing the same combo 4 weeks in a row. Sure the Quick Pick lines would change each week, but it fits the budget perfectly... park the pick 3 attempt for 4 weeks... if only the lower tier prizes were better...

As I approach the halfway point of week 9, It is getting a bit dull seeing the same thing week after week.

It will take a bit of time to convert the script for the 6/49 game, but it might be a welcome change up. I can still paper play the pick 3 for those 4 weeks to keep trying to refine the process. This will force my lazy self to finish the Cash 4 Life variant that I can not seem to find the motivation to do.

In the past 8 weeks or so, I have seen one night number picked in the day, it was boxed. I have also seen at least 2 instances where a boxed hit would have come in if I played the wild ball.

Not a missed opportunity, however. 8 weeks @ $14 per week is $112. Those "opportunities" would have netted $112... BUT... playing day for night and night for day DOUBLES the ticket price, and adding the Wild Ball ALSO doubles the ticket price... that $112 would have cost an additional $336, for an 8 week cost of $448.

At that point it is not fun anymore...

Because of an earlier win, I am about to drop another $14 out of pocket for this next week, bringing my total out of pocket expense to $28... not bad for 9 weeks of daily play... PLUS the original win funded a few cracks at the pick 5 AND the initial attempt at the Cash 4 Life brute force gambit.

Another straight win will allow free play for the rest of the year, if there are 2 wins between now and December 31st, that will fund a second try at Cash 4 Life.

The point, gotta be rigid with the budget, no added bells and whistles like wild balls or playing the day/night switch. This lets it keep the entertainment value. 2 games (mid P3 and eve P3) for $14 per week is as modest as I can make the budget and still be in the game.

Now, I could always drop the box play and cut that in half to $7 per week, but what fun would that be?

The strategy this week is to pick frequencies that reside in between Q1 and Q2, with Q2 being the transition target.

In 150 draws, Q2 is that frequency of 15 appearances, so looking at the neutrals at or below 15... where being drawn crosses the Q2 line within the next 7 draws.

So far, not looking good. The first day (9/4) was probably the best shot at a box hit on mid, and a straight hit on eve...

My mid number... 393, Thursday mid draw 133,

My eve number... 172, Thursday eve draw 272.

I have no new ideas for how to interpret the output any differently for next week. BUT, I will try again.

The Markov Decision Process (MDP), this will be the framework for Phase 2, it will define the features, the goals and the constraints, the rewards and penalties.

The actual navigation of the framework will require an AI agent driven by the reinforcement learning (RL) framework described by the MDP.

So this is actually a 2 part mission. It may include giving the agent the ability to run the phase 1 script with changing variables. This is one of those things where the agent can output what settings were used so it is not a mystery.

Confusing to be sure, but the analogy is that the RL agent is the maze runner while the MDP describes the maze.

The week was like most weeks, no matches, even boxed... however next week requires back to the out of pocket funding.

The day draw pick, 063 had a sum of 9, which was in the wide range of 8 to 18. The night pick, however is 010, sum of 1.

After finding the sums between 8 and 18 happened in 74.468% of the last 94 draws of neutrals, it might be added to the process to at least make sure the sum falls in that range... just for a week. It does have to hit the other criteria as well... closest percent to the median, mid level of draws since last appearance, neutral classification...

The 063 did not show up, closest was to a box hit on 8/28 with 016 and on 9/2 with 366.

The evening has a draw left, only 1 box close call on 8/31 with 004. Shame because a 0.50 box hit with the pair of 00 would have paid $80...

Knowing the cost for the rest of the year still puts a single $250 straight hit with a $40 box hit in that place where it returns anything I spend plus covers the entire remainder of 2025. Hoping that can happen sooner rather than later. (Or not at all)

Who would have thought that 1 in 1,000 was so hard to figure out...

Now don't get me wrong, using sums violates 2 personally held beliefs...

1. Each position in a combo is as random and independent as previous draws.

2. The numbers have no numeric properties, they could have easily used A to J as 0 to 9.

But thinking about sums... a range from 0 (000) to 27 (999) that encompasses all 1,000 combinations.

A bell (Gaussian) curve over the distribution history, with sum 13 and sum 14 at the top of the curve. In contrast to the discrete uniform distribution created by taking pure frequency.

A one to many mapping, meaning many combos at the highest peaks.

Just like the digits themselves have no memory, the sums are just as random.

So, what use could they be?

My neutral selection process could have completely different neutrals in 150 draws, or even as soon as the next week. As I recently ran a full back test on the entire history, it might be worth a look to see what the observed sum range is. A tool to be used to help with the per column synchronization problem I have in all such systems. Roughly 10% of the data presents itself as all neutral in a back test, so that would be 1,700 sum samples to look at. Is there a range in which most would sit? Would any ranges NOT be present?

The aggregation of odd and even or high and low would still be per column and not of much use, but the sum could include a third group of statistic, we have per sample and per digit, this would be per draw... used on the observed history it could provide a loose guideline to apply when selecting from each category, which ones fall into a common sum range?

Of course any such constraint increases the chance of throwing away the winning combos, but all systems seem to do that even if it was not intentional.

Might be worth the look...

Turns out, in the last 94 neutral draws, 70 were within the sum range of sum 8 to sum 18, with the highest concentration between sum 11 and sum 16. That is interesting but not directly applicable.

What about going all in on coincidence? What about a path finding algorithm that takes actions to pick the next draw (per column, of course) and gets reward points for being correct and loses points for being wrong... however, it is given the freedom to use any method to get there. It could transform (+1/-1 etc), it could mirror, it could perform arithmetic and / or algebraic calculations, etc...

Training the model on past data to arrive at a best guess BUT also output what it used to get there... 

I will learn the majority of what I need to know by the time I finish the Markov Decision Process via reinforcement learning script for the current system... 

So if a system is given a toolbox of techniques to solve the problem and let loose to do it's thing, what kind of pick would that generate?

I still have the rest of 2025 to focus on the current system, but I need to start thinking about next steps far ahead of retiring the current system...

Nothing yet.

As I proceed with the PA Lottery Results RSS feed scraper, it looks like I have to manually update the big game files to within a few weeks to make it work... that is a pain, but... needs to be done. I am also exploring using their website, as you can get a whole year at once. This would be more powerful help for catching up on the games that were not updated in a long time. Why does PA have to stay in the stone age by NOT having an API? It is MUCH easier to extract data using JSON.

I am committed to this current system through December 31... already know it will cost $252 ($14 × 18 weeks). With 3 draws left on the free play, there is still a chance to avoid any of that being out of pocket, but in the most likely outcome, it would still be one straight hit in those 18 weeks that could erase the expense, or 7 $40 box hits...

Markov Decision Process using a "reward" system is looking to be quite an undertaking... I do not yet have even a flowchart together for the process... BUT, I can output the data to multiple files (one for the model data and one with the individual digits and their features), so that is a start for the MDP script, know your input. The next step is to know your process and finally to know your output. I know the first and last...

Moving slowly forward, but still moving forward!

Gonna have to get a ticket, gonna use the Python Mersenne Twister PRNG. The free MM QP was worthless.

When dealing with these games on a per column basis, there are probably 5 of the last 7 weeks where each pick DID show up in it's respective position within the next 7 draws, however Synchronization remains a problem, as they do not end up on the same draw.

Though I could see this to a some extent in every singe system I have ever worked with, it is the eternal plague of the pick N games. In order to win, they need to line up on the same draw. But since each draw position is as independent as the previous draws, I do not have a workaround... it is why each pick will always ever be nothing more than a "best guess".

Also, follow up hits have proved to be challenging as well. 90% of every big idea I have tried to implement has hit once, then that was it. In order to implement the rest of the plan (pick 5 "Quinto" and Cash 4 Life only played on won money from the pick 3) there needs to be more than just 1 hit. I am on my last week (through Wednesday of next week) of "free" play.

Now, moving to using the inter quartile range to classify as hot, neutral or cold has lowered the frequency of all neutral observations from 85% of all windows to 70%, but it is more stable because the range is not as prone to influence by the outliers as is the standard deviation.

So, even though this system looks quite different from previous ones, it still suffers from the same synchronization issues, which contributes to the low hit frequency issue. The Cash 4 Life only needs to work once... the pick 3 was always meant to be the engine that propels the system forward... I have yet to solve that problem.

The addition of the chi square statistic and the P value was a success. What is of interest is the P value, because it goes from 0 to 1, the higher it is, the more representative your observed distribution is to the expected distribution, which is a discrete uniform distribution. A P value of 0.05 or less means the distribution does not match... something is not right with the results... in 2,400+ runs in the sliding back test, the lowest observed was 0.13 and the highest 0.97, so that tells me the PA lottery draws DO fit the expected uniform distribution. The variance will be left to the next script to figure out.