Do you use simulations to test systems?

Reply #11

aquariuslottery — Fri, 31 May 2019 06:30:35 GMT

The abbreviated wheels guarantee is the minimum ROI. When I matched five numbers but only has a three numbers match the wheel guaranteed, it became obvious the order the numbers are entered into the wheel determines if it can produce more than the wheel guarantee. Used a 3 if 4 wheel and matched four numbers several times by only matching four numbers.

That's why the distribution of numbers among lines in a wheel is so important. Balanced wheels will distribute each number in lines approxima... [ More ]

Reply #10

Stack47 — Thu, 30 May 2019 04:27:44 GMT

Have a wheeling program where I can apply just about any kind of filter I can think of. Found out lots of times the number I chose just didn't play well together. Those results showed numbers at the top of my list only appeared one or two times. IMO the ideal would be a wheel that filtered first and then find whatever guarantee the player wants, but I never found one.

The abbreviated wheels guarantee is the minimum ROI. When I matched five numbers but only has a three numbers match the wheel... [ More ]

Reply #9

aquariuslottery — Wed, 29 May 2019 05:46:37 GMT

How exactly do the results help anyone test the earnings potential of systems and/or wheels?

You have to measure their potential ROIs. Systems may have different earnings at different times and the only way to compare is ROI. Wheels may also have very different potentials. Overall potential ROI of a wheel may also be used for evaluating wheels. I use 4 parameters for defining wheels, eg. 3W 5H 16N 40T which translates into: you have guaranteed 3-number win if you hit 5 numbers in the pool of

Reply #8

aquariuslottery — Wed, 29 May 2019 05:00:31 GMT

I never use simulated data for testing systems. I only use actual drawing data. IMO, using a computer RNG to simulate a drawing is not as genuine as using actual drawing data.

I fully agree with that. Use only original data (draws) to backtest. Why bother with separate RNGs when you have the original data ready?

You can backtest in 2 ways. 1) Start from the latest and go back say 100 draws for jackpot games. This may give you lots of false positives (wins) which in reality may not have hap... [ More ]

Reply #7

Mr-B 216 — Tue, 28 May 2019 00:32:35 GMT

I never use simulated data for testing systems. I only use actual drawing data. IMO, using a computer RNG to simulate a drawing is not as genuine as using actual drawing data.

Let's say I have 3000 drawings to analyze and test. What I do is analyze drawings 1 thru 2900, and then test my results against drawing 2901, which is not a part of my data file. Since drawing 2901 is not included in the analyzing process, it can be considered virgin data. From there I simply continue to analyze drawing... [ More ]

Reply #6

AllenB — Tue, 28 May 2019 00:05:31 GMT

First Applying a System to the Draw History Means that the System is Automated. If Filters are applied they must be done by formula not input. Systems requiring Manual Selection of any kind are the First Step toward Automation and can be Practiced on Past Games but I don't call that a Back Test nor do I call that a System. It is a Tool that you Practice with on Old Games. Either by Practice with Repeated Runs for Fixed Setting on the Tool, you can find if a Setting or Code was a winner and when... [ More ]

Reply #5

Stack47 — Mon, 27 May 2019 22:33:45 GMT

I would first apply it to the Draw History.

How exactly do the results help anyone test the earnings potential of systems and/or wheels?

Just wondering.

Reply #4

cottoneyedjoe — Mon, 27 May 2019 17:09:16 GMT

If a person's system is based on trends or patterns observed in past draws to begin with, testing the system against the draw history is more apt to give them a false positive confirmation that it works, while testing it against newly generated random draws could show that the system fares no better than quick picks. Maybe the latter is something systems creators don't want to face, and so that is is why the discussions of testing systems focus only on back-testing.

Reply #3

Stack47 — Sun, 26 May 2019 23:38:55 GMT

Never thought it was necessary to test a guaranteed 13 line 3 if 5 of 15 numbers wheel until I matched 5 numbers. What I learned is while matching 5 numbers is difficult enough, it's the order of how the numbers are placed into the wheel that determine if the matched 5 numbers are on the same line. And for the same cost I probably should have played a 2 if 2 wheel.

Losing $3 after matching all 5 numbers isn't exactly my idea of fun. The LP wheels are user friendly making testing very easy.... [ More ]

Reply #2

Todd — Sun, 26 May 2019 21:12:24 GMT

Reply #1

AllenB — Sun, 26 May 2019 20:45:28 GMT

I would first apply it to the Draw History. That is after all a Random Set and the Product of the Game you are playing. Playing against an Imaginary Set of Random Numbers generated by say the Randbetween Function in Excel is something that I have been Hot on for the Past few Weeks.

I think You should do both but I am becoming more in favor making my selections more random and less logical. There is a balance to be discovered I think.

Do you use simulations to test systems?

cottoneyedjoe — Sun, 26 May 2019 19:39:13 GMT

If you select the lines you are going to play using systems and/or wheels, do you test the earnings potential of the lines using simulated random drawings before you go out and buy them