Since there are 3 weeks, a different strategy will be applied each week.

Week 1: Taking the Markov Chain follower direct result, using the classifier as a tie breaker on equal frequencies.

Week 2: taking the 10 most recent followers per position, and finding the one highest on the list in the classifiers for each position.

Week 3: taking the digit in each column that is simultaneously highest on both classifier and follower tables, that also appears on the 10 most recent followers list.

They could all be way off, one may get closer than the others, maybe a dumb luck coincidental win... I should have mixed it up with the match 6 and pick 3 cycles, but I was too busy making the Android app work.

After this P5 cycle, I may consider going back to playing the games on days I have opportunity, which would reduce plays to 3 or 4 days a week and directly reduce the cost. But that depends on the ongoing lag test.

The idea I have for the classifier as a back test might finally help define that "churn", or how the numbers are random, but the frequencies are closer to the same. This would involve tracking the average time it takes for numbers to transition from hot to neutral, from neutral to cold and also from cold to neutral and neutral to hot.

The app was not the end game... only the beginning of understanding and interpreting the data it produces.

After the last draws of the pick 3, the tally stands at -$82, adding round one of pick 5 tomorrow takes that to -$96.

There is still that 1:1000 chance on the last mid and eve draws in the pick 3 cycle today, but that has yet to work into a win. One or the other hitting today would give me back the $82 spent so far and fund the "experiment" into September! A hit on the pick 5 would refund the money spent on the lottery over my entire play history! (I had plenty of "off" times, sometimes spanning years, and I am cheap, and don't always lose)

I have had better luck over the years on the pick 4, now that the app can handle all 14 games (except cash pop, they can keep that money grab) I may also switch up games in the cycle pattern... maybe a cash5, pick 4, pick 5 cycle?

5 days away from my first major app update, removing Cash 4 Life and replacing it with Millionaire 4 Life. I have to put all of the parts together and test on the windows app before pushing the Android update, including deactivating the classifier button on that game until there are 870 draws.

While this project bas been a rewarding learning experience so far and a win would just be the icing on the cake, I am about ready for dessert!

Extra update... curiosity got the best of me. I tried putting the app on my decade old galaxy note 8... it works! Slower, but not unreasonably so.