Few words of explanation:

1. Main idea for me was to see entropy data together with performance of Padovan numbers.

2. Entropy calculations were done as follows: ("reduced" lottery matrix set was used as per usual)

a) for every game event and each number, count number of frequencies going past 10 draws (thats why data points start from 10 and up for the entropy)
b) calculate probability: number probability = number frequency / 10 * 80 (80 is total numbers in NY P10 game)
c) calculate Shannon entropy = entropy += (number_probability * log(1 / number_probability))

3. Green represents Shannon Entropy and Red Padovan Sequence Numbers matched

Few observations:

1. it actually took some time to measure correct time window (10 in the NY P10 game) in order to arrive with well scaled trend line, believe it to be the correct measure

2. in many instances Shannon Entropy "predicts" (saddle points precede day or two in advance increase in the number of hits from the Padovan sequence) and this is exactly value of this chart.

3. Overall strategy to optimize entry into the game (not to optimize getting exact numbers!) appears to be at the lowest "saddle" points of entropy trend lines that is when uncertainty is at its highest, (this parallels findings in the RPS game! where betting during lowest entropy points brought best payoff!)

Similar chart can be constructed for entire prediction sequence (not just Padovan or some other arbitrary "well behaved" sequence)

Different lottery game require different scale (the above fits well with 20/80 game).

Immediate prediction for NY P10 game is that there will be >=7 hits in the Padovan sequence (even knowing that brings HUGE advantage into the game) timing this event is exactly subject of this blog entry!