Found two more patterns. Using either one could produce a small set of playable numbers that hit better than anything I've come up with thus far. Below is a random example. Chosen because that was the month I was on while testing and both methods hit in this example. As usual, I've only tested with Florida data. I'm curious how well this works with other stats. Both systems use doubles as triggers. So far I haven't noticed that one method works better than the other. I just noticed both so I'm sharing both. In any case, here's the system:
5/24/2012 |
MID |
5 |
9 |
3 |
5/24/2012 |
EVE |
2 |
0 |
9 |
5/25/2012 |
MID |
6 |
6 |
9 |
5/25/2012 |
EVE |
3 |
5 |
6 |
5/26/2012 |
MID |
7 |
0 |
6 |
5/26/2012 |
EVE |
8 |
9 |
8 |
5/27/2012 |
MID |
0 |
1 |
2 |
5/27/2012 |
EVE |
1 |
7 |
7 |
5/28/2012 |
MID |
2 |
9 |
3 |
5/28/2012 |
EVE |
0 |
0 |
7 |
5/29/2012 |
MID |
6 |
7 |
4 |
5/29/2012 |
EVE |
2 |
2 |
4 |
5/30/2012 |
MID |
2 |
3 |
7 |
5/30/2012 |
EVE |
8 |
6 |
9 |
5/31/2012 |
MID |
6 |
7 |
3 |
5/31/2012 |
EVE |
6 |
9 |
7 |
Method #1:
A double 6 hits on 5/25/13 in position one and two. From a testing point of view, we looks for any sets following that double that contain a 6 in any position. As luck would have it, the very next draw contains a 6 in position 3. Keep that in mind. Here's how this system would have produced that set:
Starting from when the double hit, take three (I go back three because it seems to give me a good hit/cost ratio. You could just do one and have a lower cost or do more and get more hits.) previous draws. So you'd have 5/25 MID, 5/24 EVE, and 5/24 MID:
Each one of these sets will produce three sets. You'll want to replace each position with the double, in this case, a 6, one at a time. So the 5-9-3 will produce the playable sets of 6-9-3, 5-6-3, and 5-9-6. Do that for all three of the original sets and you will have nine playable sets. In this example, the 5-6-3 hits boxed. $9 would result in a $80 win (or $150 if you play online). Now, typically, it takes a few days for one of the sets to hit (assuming it doesn't miss). There are misses but it hits enough to show there's a pattern. It's the best thing I've come up with so far. It's reproducable month to month.
Method #2:
This method is identical to Method #1 except that you generate new numbers every draw and follow the draws. So instead of just using the three draws that occurred before the double occurred, you would use the last three draws. In this case, on 5/31 MID, 6-7-3 hits. That's a boxed hit from 5/30 MID, using the double replace method.
Thoughts?