Josephus, these are good question, but hard to answer due to the varying ideas of what success means to each player. BUDGET often plays a major part.
One Pick 3 player may play 10 combos, $5, anticipating box hits, while another plays 20 combos for $10. Our second player will feel the increased number of box hits is worth the cost. And yet a third player, will play 10 combos at $10, in hopes of a .50/box, .50/straight win.
When you deal with occasional accurate predictions, the real question is: Is the payout worth the wait? We know that 72% of Pick3 draws are going to be single numbers. Even if they offer the smallest payout, they offer an opportunity of a hit more often. Doubles are drawn 27% of the time. Many people feel since doubles pay double, and are a smaller group, they are the moneymaker. And the third mindset is 'you can't lose playing triples.' If a triple hasn't shown in say, 80 draws, it is time to start playing triples.
Replication from one state to the next is possible with any system. Since each system is built along certain rules or a framework in which you insert the variables of your own game, it often depends on how well you know the game in your state as to success or failure. This is not set in stone as trends change from time to time, which is one of the biggest failures of all systems. Again, how well you know your game, and how quickly you identify trends can make or break any system. All lotteries are in flux.
Remember, the purpose of every every system is to find a universal key that works every draw, for any lottery. Backtesting has advantages in helping do that.
But backtesting can't account for the BRAIN.
From the pure numbers standpoint, backtesting can give an idea of how a system was producing 6 months, 6 years, or over the life of the game, based on the criteria you select, but it cannot account for what is registering in the brain at that time.
Examples of things that would register in the brain, at the time they were occurring, but can't be accounted for in pure mathematical backtests would be: doubles haven't hit in 15 draws, the digit 6 hasn't been seen in 24 draws, or triples have been out 118 days. Likewise you may have seen certain 'triggers' fall that would have led you to a hit that would be glossed over by pure math.
Inasmuch as it produces results showing IF one played in this manner, ignoring all other factors, THEN this would likely be the outcome, backtesting serves a purpose.
Creative systems intrigue me. I like people who think, who stand on the edge and look in, look out and both ways before crossing the street. Back in mid 2009, I'll guess August, WinD and/or paurths (forgive me if I appear to be giving more credit to one than I should, both are excellent thinkers!), presented a system for pick 3 players that (I feel) never received its due.
The idea is simple enough. One makes sets of 5, 6, or 7 digits. In the case of 6 digits, one would make every set from {0,1,2,3,4,5} to {4,5,6,7,8,9} and track those for longest sets out.
We then take that set, wheel the digits, and play those sets. Very simple, viable idea. Excellent, strong foundation. The drawback was not in the system itself, but in others understanding of the idea, coupled with the complexity of tracking each set. Many felt that the longest out set was supposed to hit that draw. If it didn't, the system was a failure.
At the same time, I was working from the oppsite side, based on the idea that 'a trend will continue.' I could see it no other way, so I was tracking sets that continued to hit. In doing so, I found a handful of sets that separate themselves from the pack by a couple of percentage points.
I realized WinD and/or paurths system was head and shoulders above mine when backtesting proved a set only needed to hit 2 or 3 times in the next 100 draws to maintain a leading edge, where their system did away with the headache of percentages completely.
I still use some of the routines written around their system.
1) In total, a system is only as good as the idea we use to build the framework and how well we intepret the output.
2) Backtesting has a purpose, but can never take us back to that point in time to allow us to know what the brain was thinking.
G