I don't know if I would abandon it! Lets do a little analysis here.
Looking back 279 drawings, this is the order of the numbers most frequent to least frequent (if my calculations are correct):
And the drawing was: 01-12-14-25-35, Mega Ball: 38
So the drawing was fairly all over the place. If we were to play 75% of the most and least frequent numbers, we would be playing 42 (56*.75). That leaves us with the top 21 and bottom 21, which would look like so:
Top 21 most frequent: 53,46,5,36,7,40,24,14,12,30,51,43,42,25,48,17,54,44,22,13,16
Top 21 least frequent: 9,33,29,8,21,3,55,50,1,4,27,23,11,49,34,15,10,41,28,37,6
If we included the same method for mega balls, we would be playing 34 (34.5) of them, but we'll assume around 20 tickets. That would have left us with:
Top 10 most frequent: 22,2,42,10,7,35,15,39,44,36
Top 10 least frequent: 23,20,19,18,16,3,28,45,31,46
Since the mega ball drawn was the 13th most frequent, we would have missed it while playing with this method.
Let me know if those numbers match yours RJ. Is this similar to how you set up your picks?
The method I like to use for picking Mega Millions & Powerball numbers are with 'due'. I use similar means, taking the most due and least due, which seems to encorporate alot of the most and least frequent numbers.
Numbers organized from most due to least due:
Note: I used a pool size of 266 for my picks, this is with a pool size of 279. I use 266 because I am able to back check those with a few previous drawings, where the pool size of 279 can not be back checked because that is the entire size of my database.
Playing 75% of the numbers (with 20 mega balls) would have resulted in:
Top 21 due: 17,7,55,32,30,53,38,40,2,20,56,11,42,12,35,29,36,8,24,19,18
Bottom 21 due: 39,3,34,50,21,41,52,22,33,1,5,27,15,49,23,6,28,4,37,31,48
Top 10 due: 9,15,24,45,41,18,4,25,12,44
Bottom 10 due: 30,42,33,40,34,1,3,13,28,46
Using top/bottom due vs. frequency methods here would have produced similar results, but we would have picked up one more white ball with the frequency method (4 out of 5).
We just need to refine some filtering methods, maybe with taking both methods into account.