Ive already been throughout the site, the first reference was march 2001, then June 2001, then the revisited version June 17 & 26 of 2003, which is what I am referring to, where he an improved version after discovering the worst case does not work. However, I cannot make heads or tails of the improved last version to work, which is a mixture of top and worst pairs. See below an excerpt of this explaination, maybe someone has gotten some understanding from it:
"By the way, applying the concept to the pick-3 game: the behavior is different. Getting a 'wonder-grid' of the best pick-3 pairings results in 10 combinations. They offer acceptable results only as 'boxed' combinations. Playing the 'wonder-grid' to win 'straight' is inconsistent. The best I have gotten so far for pick-3 is ELIMINATING the 'worst 6' pairings. The result is some 50-70 combinations with a lot of 'straight' hits. There is also a larger number of 'boxed' hits. The best result, however, is playing 'straight only' (disregard all the 'boxed' hits!)
I eliminated also the 'worst 7 pairings'. For the most part, running POWER-3.EXE results in 0 combinations. Sometimes, I get 1 or 2 straight combinations. I haven't hit yet, eliminating the 'worst 7' pairings. But, who knows! I'll check more draws."
• That was then. Things have become much clearer now. In fact, playing the best 3 pairs while eliminating the worst 7 pairs is one proven strategy at beating not only random play handily, but also beating the monstrous house edge of 50%! I create the final grid and play it for the next 100 draws. This appears top be the optimal range for the fruition of a pick-3 wonder grid. There are two interesting triggers.
1) If the final grid consists of 18 or 19 combinations, it is best to play the same grid for the next 100 draws;
2) If the grid consists of 10 or 11 combinations, the probability is higher the grid will hit the very next draw; if the grid does not hit, discard of it.