I presume this post is addressed to me!
markov.exe was/is intended as a proof of a concept (markov chains in lottery) it was coded in the back of an napkin (so to speak, actually it was adapted from another program I saw in a book) and since then I have discovered that it has severe design problems!
If you still want to give it a go, my advice would be to create a file with say the last 100 draws of your lotto, oldest draw at the top newest at the bottom. The lines can be in draw order or sorted (it should make a difference from a systems point of view, but from the lotto predictions point of view it doesn't).
Now the final trick to solve some of the design problems is to add a line with just two numbers at the top of the file, this two numbers should be exactly equal to the last two that are at the end of the file....
Something like:
35 77
10 37 70 57 25 19
13 90 69 25 46 30
39 37 55 12 74 80
18 02 27 50 35 77
After just save the file to say input.txt and run the command:
markov < input.txt > output.txt
It will try build lines of 6 numbers based on the input file!!
Every run should display different numbers because the program selects each following number based on the past number and on some random probability... in the example above 25 is followed by 19 and 46 so the program will select one of these randomly!
Which run is best... none and all.... there is no way to tell! What I used to do when I was looking at the performance of this program was to just check them all against the next draw and see the results!
Have fun!
