hypersoniq's Blog

Planning a big test on the Pick3 PA eve. AND Midday (NOT a live playing test). THis is me vs. history. Going to revisit the 1-1-1 system, for all 1,000 combos from

0-0-0 to 9-9-9 (Where you add 0,1,2,3,4,5,6,7,8 or 9 to the previous night's number)

using the "rolling 10" ... 8-9-0-1-2-3-4-5-6-7 etc...

Once I have the data, I will attempt to program this test in VB and compare the manually obtained results...

THEN apply it to the numbers AFTER LRS was run.

Remember, in actual practice, LRS was the only system I had work for real money, but 1-1-1 backtested TWICE the hits last time I ran it.

I need to start looking at what's next... LRS is a great tool, but there has to be more to pick3 life than a few hits every decade... 

Check the blog for further updates...

My own research lab

HyperSoniQ Innovations

and the first order of business is research into ideas I have about alternative energy vehicles, and alternative energy in general.

This would be the springboard for every idea I have had that would require research and development to get to market (or just to get to a patent which can be licenced to a manufacturer).

Once a decent management team is in place at the skatepark, here is where I would spend the majority of my time that I currently have to waste working for someone else.

Come ON MUSL, I want to see the press dilute the lab's contributions with gem headlines like "Lottery Winner perfects next generation storage cell technology for Electric Vehicles"...

I would open a huge Skateboard and BMX bicycle action park.

I would incorporate Ideas I have drawn up for the ramps and flooring that would allow them to ride like normal but if they fall and impact, it would reduce the chance of broken bones.

I would create areas that would match rider's skills and budget...

Public area($3 all, $2 straight B's, $1 Straight A's)... I need to win the PB to keep this admission artificially LOW so more people get the opportunity to participate. There would be plenty of challenging obstacles but done safely, not like those public concrete pool parks. Here I would work with local school districts... ONE by getting kids to bring in copies of their report cards with discounts for all A's and B's, and TWO to get kids in high school who need to use community service as part of their terms of graduation to teach others basic tricks. That both helps the teacher AND the others.

Training area($5 additional, $4 straight B's, $3 Straight A's)... This will cost more to get into but will have PAID instructors. Here riders can learn more difficult aspects of riding like big Vert ramps or Backflips into foam pits (for safety, and it's just fun). Also a unique indoor BMX dirt racing track for practice in any weather.

Core area($2 additional, $1 straight B's, FREE Straight A's)... This will be the X-games sized ramps, park courses and dirt jumps... Visitors will need to be "checked" by instructors to qualify to enter (for safety). I would line up pros to do clinics and have pro-am competitions in here. No amount of money will get you into the core area if you are not signed off as capable by a staff instructor.

So access to the core area would be $10 max ($4 if you have straight A's)

on the cheap, a straight A student could ride for $1 a day.

these are just estimates... but I couldn't hold these kind of prices if I wasn't already a millionaire...

The whole park can be built in phases, with Public being phase one. I am looking into doing this anyway once we head south next year... but I would rather not have a huge loan and high admission prices...

come on MUSL... it's a good idea!

Some seeds took as long as 4,000 loops to pick their number, some took 7, several took 1 (The same 9 from the last test).

there are only 1,000 combos... so there is a definite problem. The code is right on the money, I am just wrong with my seed assumption (whole date number, INT process)

I have 3 more variations...

Date+0.5,INT

Date,Round

Date+0.5,Round

I am currently running Date+0.5,INT... Still lets me browse, a definite improvement over it's ancestor the "Coincidizer" which held this PC hostage for upwards of 11 hours... This takes about 60 minutes at normal priority. ONLY because some numbers require several hundred and several THOUSAND loop iterations.

I already have VB's RANDOMIZE sensitivity threshhold identified at 0.5, and I have seen a marked improvement over LRS versus this CPU generated game...

looking for about 200 hits of the 1597 draw history before I start playing for real.... A number I am NOT likely to see considering so far the best is 9...

and no apparent practice draw patterns have emerged.

Unfazed by the initial noise, I move onward... (hopefully) to Victory!

had a cash5 ticket where LRS called one number... three of the other four were ONE OFF. all higher.

I am adding a tweak tothe LRS formula to check and count this phenomenon as one more fine-tuning step for LRS, particularly noticeable on the cash5.

rather than re-run a sheet copy I am just going to add to the LRS formula line to add the contents of designated cells and fill them with numbers from -5 to +5 so I can count the correct hits of each modification.

I caught and fixed a logic error in my new CR code, so hopefully I can update and run tomorrow.

Need to code a new button to run the test.

This one will make several assumptions...

1. The integer value of the DATE is part of the seed

2. They do pre-draw iterations of the loop

This will be a time-consuming run. One variable picks how many times it will iterate... going to start with 10

For each number in the history, it will randomize based on that date's integer value and run thru 10 iterations (each iteration is just one step on the list)

output will be the 10 RNG numbers, a "Yes/No" indication of a match, the number of the iteration that matched (1 thru 10) and the seed value that was used.

round 2 will change the variable from 10 to 100 and eliminate the output of picked numbers (to speed run time).

I said it before, it probably CANNOT be done, but if it can, I WILL DO IT THIS YEAR!

LRS continues, Cash5 went cold... hoping weekly re-scan changes things a bit... powerball has flatlined on LRS and Match6 is suffering a severe case of "one-off-itis"

Switched from full history to new matrix only in the PB... First play 2nite!

Looking for an 11 in the red ball slot... THIS may be the jump start LRS needed for powerball.

Test one involved seeding RANDOMIZE in a loop of the numbers 1 to 1,000,000 (one at a time)... generating a random list of numbers equal in length to the draw history... comparing for matches, then printing out the highest ones found.

UNFORTUNATELY there was no great news.

There are 1,570 draws so far on PA mid-day pick3 and the best list only matched 9. YES technically this beats the game on paper (cost to play taken from 9 straight hits = a profit of $2,930) but it is NOT what I was hoping for.

It is better than LRS, which picked 2 straight hits for the mid-day...

but I must move on.

trying different seeds next, ROUND vs INT for the pick, decimals etc...

I am looking for 227 hits, that averages one a week.

So I did show good results if it were a ball drawn game, but BAD results trying to synch 2 RNGs... Ideal would be of course 1,570 matches...

not there yet, but NOT giving up either, the potential returns are too great!

still haven't spent a dime so it's all good.

LRS gave me almost everything I wanted in a system...

one pick

used draw history

backtested well (better than any other system I have tried) without being a system where "hindsight" can adversely color the data

the ONLY thing missing is a win!

Now, Pick3 had 1 $500 win... but to play until the next backtested hit average I would have to spend over $450 of it to "maybe" hit again... I quit that game and I can say I beat one!

Pick4 had zero straight hits and only one box, I dropped that one too mostly because the backtest was poor compared to others.

Which leaves me with June's rag-tag collection of Cash5, Match6 and Powerball. Not even a single $1 hit on cash5 this month so far. Unlike the smaller games, the jackpot system only needs to work once...

July's test is finalized already... it's

Powerball every draw

Cash5 only on a roll-over (should save at least $15 or more next month)

Match6 ONLY when the after-tax prize is about $1,000,000 ($1,550,000 advertised)... this means if it's hit I may not play at all, and if it rolls thru june I could be in all month because it's almost that now.

In august I look at the cash5 and match 6... I will tally the money spent vs. the money won (if any) and the highest one stays for august... if they both go cold for june and july,they BOTH go... then it's back to just the powerball for me.

I am re-running each LRS chain program for C5, M6 and PB... just to make sure I have the latest data.

I have removed the p3/p4 LRS sheets to the thumb drive archive and I am quite pleased with the $60 I will save by not playing pick3/pick4 this month.

with Chain Reaction, I ran the million seed test for the first number, took only 2 minutes  to complete on this old PC (ran comparison on the "family pc"- the good one everyone else uses...that took only 21 seconds and gave the exact same results, so I guess the game is on...) now I have to find a uniform count increment to try and find the best match. I will run each seed through a 2000 count loop to generate a list of picks, this will be compared to the draw results... There are 973 integer seeds from 1 to 1,000,000 that generate the first number drawn on the mid-day pick3... NONE of them correspond to a date or system time variable, so it's a blind hunt from here on in. I am going to load them into an array and prequalify which ones match the second number, then run each of those (if any) for the 2000 draw test.

looks to be a looong day

No positive results from the june test yet.

May was a bust... $101 spent (from april's $500 fluke win) $4 won (all on PA cash5)

June gets cheaper... Tonight 1/5 on cash5, 0/6 on match 6... cost to play $3

tomorrow is another $3 day... then sun. and mon. only $1 ;-)

June cost will be $66 (again, from april's $500). Then for July I will be back to spending my own money, so more cutbacks will be made.

"paper" testing of scenarios on the Chain Reaction program are ongoing, again... no live tests planned unless I get AmAzInG backtest results (like 1 hit a month). This program is constantly changing... The first full run is testing the first 3 combos drawn in the PA mid-day daily number (pick3).... finding ALL integer seeds from one to a Million that produce that number... then they will be compared in excel to look for similarities.

If that produces desireable results, the program will be set to generate a year beyond the most recent draw. IF this works I will keep the output (a.k.a. "the List") as guarded as a winning powerball ticket. I want to beat them at their own game... I do not know if it can even be done, but if it can, I WILL do it, and most likely by summer's end. The actual live test criteria would be EVERY draw matches, but I might play around if it backtests 1 hit a month (rather than the 1 hit every 1.5 years of LRS). Think of the possibilities...

If the list works on a monthly basis, it would pay for month-long tests of ALL the games

If the list works on a weekly basis, it would fund "Jack"ing every powerball draw (like Jack Whittaker... play 100 QP's to win the PB)

Let's not even dare to imagine the list would exactly match... I would have to play musical retailers to keep making tax-free bucks

LOL

Seriously, it is keeping me sharp with VB AND excel, more so than LRS even.

I can't wait to see how LRS handles South Carolina pick3/4/palmettoC5 numbers.

Dusted off an old idea from back in the heyday... According to some, wishing is accomplished thru lucid visualization. So I saved a copy of the powerball numbers results page and tweaked the code so that my numbers appear with the date I will play them... Even made the scrolling marquee read "One winner in the state of Pennsylvania..." and reset the jackpot amounts to $15 Million Annuity and $7.1 Million Cash

Now before I update sheets or go online to "window shop" for Myrtle Beach direct oceanfront beach "palaces" that I could never afford... I point the browser to my local copy of the "dream results" page.

It's as lucid a visualization as I can muster. Glad my HTML skillz aren't going to waste :-(

I might try this with the palottery page to "wish" those Cash5 and Match6 numbers. And that is about as Mystical as I get... Sounds stupid,I know... but there are no negative effects to channeling positive energy... right?

oh well... back to my spreadsheets...

no hits... st. or box (on big 4) and no P3 hits.

Cash5 won a total of $4 (unless tonight adds more, and I doubt it)

Powerball won zip, but has the dream prize, so that's in for June

June test will be Cash5, Match6 and Powerball

July test will be Cash5 on at least one rollover, Match6 on at least 1,000,000 and powerball

and august it's back to just the PB.

Was fun to play all the games, glad I don't do it all the time tho.

Contemplating LRS, waiting for another win... Here is what I see as the benefits of this particular system...

A) Auto-ranging. Because it comes directly from past draws IN POSITION you don't have to filter out numbers that have never been drawn in that position. It will never call 01 as the second number or 54 as the third number.

B) Built-In Randomness. It uses the history, therefore the same source of observed /entropy as the game itself.

C) One Line. It only picks one line to play, very economical.

D) Already Optimized. No error-correction or delta needed.

E) It embraces Coincidence. That's what it counts! How many times has X happened.

F) It is back-testable. I can see how it did already, a key indicator in when NOT to play as much as what to play and when to play it.

G) I created it from a clean sheet. Leverages both VB6 and Excel skills, so I don't get too rusty in either one. NOT based on someone else's idea or commercial software.

H) It already worked! $503 to date won, $185 spent since adopting LRS... that's positive gain! AND it will get cheaper as I Eliminate games like the pick3/4 next month.

I spent $8 today... $1 on pick3, $1 on pick4, $2 on Cash5 (Phase1 vs. Phase2), and $4 on Powerball (Phase1 vs. Phase2)...I used to drop $10 on each PB drawing. I have made progress in cost-to-play!

the phases will be decided tonight, then I will be back down to $5 on a powerball or match6 day... After cutting the pick3/4, I will be down to $15 a week total! (7*Cash5($7), 2*Match6($4),2*Powerball/powerplay($4))

AND come july, it's the next phase of cuts... Cash5 only on jackpots of 200,000, match 6 only on draws of 1,000,000... I'll always have one ticket in the powerball. I am glad to be done with the QPs.

Maybe that will psych me up! 

trying to come up with new systems or variations on existing systems is a mentally draining experience... some days or weeks you are just FULL of ideas, then times like right now, zip..zero...nada.

I work with chain reaction but not whole-heartedly because I have no intention of playing the mid-day games. Seed Hunting is BORING.

LRS test will need to wait until the draw results for tonight (Phase1 vs. Phase2). Next month I will be down to Cash5/Match6/Powerball and the following month, since I will be using my own cash again, I'm back to just the powerball.

I have nothing else... I tried to optimize and error correct LRS, but that's basically built-into the system, it's already as good as it gets (and it hasn'tbeen that good lately).

I am contemplating a cheap wheel system, 5 lines (one PB betslip) using the LRS pick as the DN source... I might call it a "Training Wheel", but I don't really do wheels (the whole "one ticket per game" thing).

I don't know what to do next for system development... LRS was my magnum opus because it actually worked once (just once). I have been thru filters, statistics, skips, pairs, v-tracs, LRS, PRO-file, someof Ian Saliu's stuff, WINhunter, Neural Networks, Angles(using Auto-CAD), The Chief's "Bell Curve", A wacky system that never got anywhere based on the Periodic table of elements, tons of workout systems (such as 111), and a few others I have already forgotten (like the goofy MIDI based "numbers to notes" program I wrote to try and "Hear" the next sequence)...

I need new ideas, new stones to turn over, new places to look...

Anyone have any ideas? From the above list, NOTHING is too "out there"...

My only criteria is

A) must result in one playable ticket only

B) must be back-testable

but if it's not, it can be modified, so olet the ideas flow...

running the highest hits vs. most previous draws eliminated...

this weekend on PB and cash5.

percentage increases were not as high as I had hoped... another test tomorrow perhaps, but I'm tested out for tonight.

LRS phase 1 seems to indicate 06 will re-emerge as PB (back-to-back)...

LRS phase 2 gives the nod to PB of 02 (indicating a possible palindrome 02-06-02...)

BOTH are in agreement that 08 will be among the white balls (the ONLY thing they can seem to agree on)

Thumb drive has backup of BOTH phases.

I will begin working on the refinements for "Chain Reaction" for mid-day P3 tomorrow... no live tests are planned...

P3/P4 eve will not make it onto the June Test schedule... that will most likely be confined to Cash5/Match6/Powerball with a grand total cost-to-play of $66 for the month. (running out of that lucky pick3 win money... may be back to just PB for July)

Oh well, it was fun while it lasted...

on the bright side, no systems retirement for me in dec.

"Chain Reaction"... this will replace my "coincidizer" from years back...

The Goal: find the RNG seeding scenario that matches the MOST hits in PA's computer drawn mid-day pick3

The Challenges: picking a 3 digit number with =999*rnd() or a series of 3 digits with =9*rnd()

Done so far... using the numeric value of the excel draw date as a seed, the first scenario (999*rnd()) has already been run through a VB6 Seed-controlled RNG and compared in excel... there are 1500 draws... 4 hits were produced... this beats the game for backtesting, but I'm NOT satisfied with just 4... Seed Controlling means you call RANDOMIZE(<your number>) rather than RANDOMIZE(TIMER)... this will repeat the sequence on subsequent runs.

To do... seed hunting. THIS is what that whole "lotto synch" was SUPPOSED to accomplish. They used 3 draws, I will use the entire draw history (could take DAYS to run but if it works...) I am optimizing the code to run as clean and fast as possible. Coincidizer only used 5 past draws when seed hunting. I want to blend the individual programs so I can sweep seeds and generate test results from the same UI.

My early prediction is that it can't be done, but remember I like to try EVERYTHING.