Strategy
I indicated in a post last week I would post a play strategy as developed from Lotto Architect for the lottery here this Wednesday. This is it. I trust it won't bore the pants off you. Some of the steps may be unsupported by logic (my logic) - in such cases it is just what I feel I should do. The lottery is a 6 ball from 40 with two bonus balls drawn from the same pile as the main balls. Apologies if you feel this is a bit of a eulogy for Lotto Architect. It is. I like what it does and how it goes about it. And I shall put my hand out to the author of the product for a promotional fee.
So I choose to play with 25 numbers. I like the idea of playing with 25 numbers out of 40, good chance of at least half of the prize balls being in the 25 and Lotto Architect can do the work to filter down from this wheel size to my budget level. I will select wheels where the guarantee is quite low, so should just about ensure a minor prize result if half of the drawn numbers are within my selection which is more than half the total numbers in the lottery, but I shall need the bonus balls to come into play because of the prize structure - 3 plus 1 bonus ball is minimum prize division. If more than 3 balls match I may get lucky.
Bring up the Hot Cold Due process for 25 numbers and observe the set. What sort of performance has the Hot Cold Due process had in previous draws for N=25? I observe the Hot Cold Due back test and find that in the last 100 draws, 3 balls or more have matched to the drawn balls 92 times.
(M=3 21 times, M=4 38 times, M=5 24 times, M=6 9 times) Not bad, 8% miss rate, 92% success rate.
Next I check the individual number statistics module for historical occurrences of the displayed Hot Cold Due numbers for this lottery. I see that two numbers have already been drawn twice in the last 4 draws. So what are the odds of these two numbers being drawn in this next draw, which would mean 3 times in the last 5 draws? Check the Bernoulli probability tables at the Lotto Architect forum and observe that such occurrence or probability of 3 balls in 5 draws is very low. So I reject these two numbers. I therefore decide to reduce the wheel size to an N=23 wheel where all the numbers are the same as the previous set except reduced by two numbers. Presumably the Hot Cold Due process with its algorithms doesn't include the Bernoulli tables as a reference. And if I use an N=23 wheel instead of an N=25 wheel there is less filtering. I suppose a sort of paradox with Lotto Architect; it has enormous capability with filtering, but then, of course the more filtering you do the more chance you have of removing the prize winning numbers. So I am keeping the filtering demand to a minimum and reducing it further by using the N=23 wheel instead of the N=25 wheel.
I open the wheel database and decide to 'bulk' up the wheel set by aggregating 4 wheels. I choose wheels that give (all N=23) a 3 if 3 match 104 lines, a 3 if 4 match 54 lines, a 3 if 5 match 26 lines and a 3 if 6 match with 17 lines. Total lines at 201, however aggregation finds a few lines the same so this becomes 196 lines. I have no idea what sort of expanded coverage this gives me, but I am going to filter it anyway so any coverage guarantee is lost.
Use Number Groups to perform first filtering. I set the filtering within number groups to reject entries from the wheel set that have regular gaps between the numbers for number sets like 2,4,6,8,10,12. I set the rejection to accept up to M=4 but to reject for greater than M=4 so number sets like 2,4,8,12 ie 4 of such a group will be accepted. I observe from the number group statistics that number sets occurring in such patterns have virtually no history of any success so I am confident to reject same. I filter for all gap sizes that the program is capable of. This filtering removes 12 lines from my aggregated wheel set to leave 184 lines.
I then set about constructing the Rejection Filters. There are literally dozens of these and they need to be created prior to the filter process. Once created they are reusable next week. These filters are not swappable between lotteries and so are unique to the lottery and its draw history. Some of the filters are understandable as to what they do, many others require a rather deeper appreciation of mathematics than I have to understand their mode of action, however this lack of understanding of mode of operation does not affect at all ones ability to use them, as in every instance the accuracy of the algorithm/filter combination can be assessed by report and so be rejected or accepted for use. One filter in particular however is quite understandable and that is Common Appearance where it can be set to remove every occurrence in your wheel of any number set that is identical to any previous draw in the lottery data history file. I choose to reject such number sets from my wheel. This filter can also be used to accept any matching conditions one might set between wheel number sets and the draw history, for example 3 numbers in a wheel number set identical to a past drawn number set are acceptable, more than three numbers are rejected. I chose to reject only for 5 numbers with 6 matching numbers being eliminated in the first usage I described earlier for this filter. Whilst the entire suite of filters took some considerable time to set up, their usage is instantaneous. Using all of these filters I removed 139 lines which left me with 45 lines. This is at my budget level. So I thus saved these 45 lines and printed out the tickets.
Anybody else prepared to post a strategy based on the tools or software program they use?
I shall advise results in due course.
