right now I am using a weighted average of statistics.
I have the following stats for match6... MODE, Average Deviation, Mean, my own set of "lucky" numbers, standard deviation and finally a conglomerate average of all the other stats...
They are generated based on the previous draw history to see if they help indicate where the next numbers will come from. Beside each set of 6 columns of stats (it is broken down by position) is a delta number, or how each generated number deviated from the actual draw results... at the BOTTOM of each column of deltas, I keep a running MODE, the most common number... I use an addition to "zero the mode" and create an error correction mask for each set of stats...
AN example of a correction mask...
Match6 Standard Deviation... by position = 04-07-08-08-05 ... not a playable set of numbers... to get each column's mode to zero, the following correction mask is applied... +2, 0, +16, +17, +22, +33... when combined, this generates a "pick" for the next drawing of 04-07-24-25-30-38 ... since I have the backtesting above I see that the best it has ever done was match 4 on drawing #2... I did not have the correction mask for drawing#2 so I did not match 4 ot 6 then... the main problem with the mask is that it adds a "hindsight" component... it can change with the stats over time but not dynamically enough.
now, for each stat listed above, I have a "pick" for a total of 6 picks for the next draw... each set over time has a higher success rate in it's respective columns than others... I use a zero count to find out which... now I have a resultant matrix of zero counts...higher being better... for each pick in each position... I also have the 6 picks stacked to see the reference numbers... Using a weighting scheme giving the nod to the higher performers in each category, I get an average for each position...
an example formula for generating one pick from 6...
=(((L50*3)+(L51*5)+(L52*3)+(L53*6)+(L54*3)+(L55*4))/24)
this takes the best pick in L53 and gives it a weight of 6 and the 3 least producing performers a weight of 3 each... as you can see it is a simple average.
I use conditional formatting to show me zeros in the delta columns... I have alot of them... but not alot on any one line. Rather than abandon the work I put into the stats, I decided to use the weighted average to get what I was after from day 1... a single pick with ties to draw history that consistently gets closer to the draw results than a quick pick... Since Match6 goves you 2 QP's, PA had a few tries to beat my one line and my match6 numbers come closer... but not enough to cash in on 4/6 5/6 and 6/6 prizes... yet.
I will continue to tweak the formulas, the next thing I need to add is short-term trend analysis to that one pick... I'm not there yet tho...
Sure it's more complicated for the same losing ticket I could get easilyu with a QP... but it is fun putting it all together... because it is in excel... all i need do is update the draw history to add the new draw and the rest is automatic... a "quick" pick ;-)