Cracking a Pseudorandom Number Generator (PRNG) using another PRNG is mathematically possible and forms the foundation of modern security analysis...
That's exactly what it is. You can crack any lottery PRNG using .... another PRNG.
There are 2 reasons for that:
1) Output from lottery PRNG is limited.
It's the output that matters: 13.9M combos in 649 is statistically minuscule.
But that's all that lottery PRNG can generate.
It does not matter in the least what algorithm is used to generate these combos.
They are always within certain, and limited, range.
2) Both outputs are transparent.
Which means you can statistically observe what each PRNG is generating in real time, draw after draw.
You think there will be no matches between them? Think again.
I call them proximity matches: how many numbers from groups of combos from each PRNG match.
Make this comparison in 3 different time frames, see what happens.
I look at this from the point of DPC (Data Processing Cycle).
You take some data (raw or already preprocessed), apply some algorithm to it (simple or complex depending on need) and you will end up with some result(s).
A complete cycle: data -> algorithm -> result(s).
Why is it important?
Because what you see and don't see in randomness depends on DPS (Data Processing Scale).
And DPS depends on the number of DPCs.
For small scale a few dozen DPCs may do.
They will not reveal much.
But to see the real workings of randomness you need a very large DPS.
How large?
For regular processing I use 300M DPCs and I tried Bs (that's billions, not BS).
What are the benefits of such a scale?
You see a feature that is not visible in smaller scales: cyclic repeatability.
Yes, you read it right: cyclic repeatability.
And when there is cyclic repeatability there is possible predictability.
And what is at the other end of predictability?
They call it cash.
I was able to use it already and have tickets to show.
But the whole system still needs refining.
I'm posting this not to convince anybody about anything.
Convincing is not even on the list of my LP interests.
I post this to demonstrate that there are whole statistical universes for lotteries, ready to explore.
And you don't need AI to do that (I don't).
In fact, AI may totally mislead you into belief that some lottery PRNGs could be unbreakable.
They are because of 2 condition I specified above.
Needles to say, small lotteries like 535 or 539 are quite vulnerable.
This includes core PRNG for PB.
Not long ago I stated: you don't try -> you don't see -> you don't know.
You can add this one to it: you don't try -> you don't know -> you don't win.