theory test

Reply #9

woody4591 — Fri, 18 Mar 2016 18:18:52 GMT

Thank you all for your feedback. My searches have taken me back to 2002 when NJ went to twice a day draws. Unable to use March as starting point however. Hopefully April and May will give me a chance to test my theory. Feedback will be posted in June...

Luck to all

Reply #8

Wisconsin3054 — Mon, 07 Mar 2016 16:28:11 GMT

If you are testing something that is based on a condition that only happens once a month, like the first day of the month, then you probably should go back much further. 5-10 years maybe

Reply #7

Tialuvslotto — Mon, 07 Mar 2016 12:48:07 GMT

I prefer a longer back-test to see how the system performs under all types of conditions. We know that the game can be streaky. So, what happens when your system suddenly gets hot? What happens when the system goes cold? Do they balance each other out, or do you still profit? How can you maximize the impact of the first and minimize the impact of the second?

And, Stack, you are right: Those numbers won't be drawn in that order again. But numbers that are awfully similar will be. Same balance... [ More ]

Reply #6

grwurston — Sun, 06 Mar 2016 22:03:58 GMT

I usually go 6 months to a year. But I feel the longer the better, only because with the longer history it gives you a much better chance to observe how your particular game works (behaves) over time.

Also try tracking the same thing different ways, mid-mid, eve-eve, and combined. You may notice something on the combined draws that you wouldn't notice on separate draws for example.

Anything you can think of can be tracked, so use your imagination. Don't just step out of the box, crus... [ More ]

Reply #5

destinycreation — Sun, 06 Mar 2016 20:27:24 GMT

For Pick 4 - 3 months is sufficient - 6 months maximum. In my opinion, unless your Pick 4 system is based on very complicated calculations, it really is Not necessary to back test from ancient history.

Reply #4

butch2030 — Sun, 06 Mar 2016 17:35:47 GMT

For the IL pick4 game - Midday Evening - I went back to 01/01/2011 I am updating my file weekly. Each state will have different results.

I am only doing Illinois as I am living there at the present time. As of yesterday, there have been 3,782 combined drawings (Midday Evening).

Here are Straight numbers to play in IL that have hit more than 3x.

MY Numbers GRP

2011-01-22 M 9 1 2 7 37

2014-06-13 E 9 1 2 7 37

2014-07-06 M 9 1 2 7 37

2015-11-26 E 9 1 2 7 37

MY Numb... [ More ]

Reply #3

Stack47 — Sun, 06 Mar 2016 14:34:40 GMT

It really depends on the type of game and of course what type of data is collected.

How much history for a pick-3 game might include how long it takes for every straight combo to be drawn, but that could take over 10,000 drawings. By using digits, the time is much shorter. With 75 numbers and almost 18.5 million combinations, there won't be enough drawings to even a sample. It takes 15 drawings before it's possible for every number to be drawn so it's reasonable to use a history that include... [ More ]

Reply #2

woody4591 — Sun, 06 Mar 2016 14:30:00 GMT

Thanks for your input. I've researched from 2010 just to be safe. I now have the information I want. Now the hard part is sifting through it, to make sure it is profitable. I've seen and read a lot of ideas on LP since I've been a member, but most require many many combinations in order to get a hit. I've come across your replies many times and must say you seem to know your stuff.

Not quite ready to post anything yet tho. It has something to due with pick-4 doubles drawn on the first day of... [ More ]

Reply #1

Tialuvslotto — Sun, 06 Mar 2016 12:24:59 GMT

Combined or separate data? How often do you expect a hit?

If combined, I would shoot for a year -- this might be tedious if you don't have it automated.

If separated into Midday, Evening draws I would shoot for 2 years.

If you are just looking from one draw to the next you could probably get a good indication with 1 month (combined) or 2 months (separate

theory test

woody4591 — Sun, 06 Mar 2016 03:48:21 GMT

how many years is good when back testing a system