Numbers are seemingly random and unpredictable - yet fractals are totally predictable!

This raises 2 questions:

1) What exactly is predictable about fractals?

In this fractal example we can observe that the dots form a predictable pattern in terms of what areas are likely to hit and what areas are likely to remain unhit, i.e. the small black upside-down triangle regions - and this pattern must form based on a trail that "snakes" around the big white triangle region.

If it were totally unpredictable then the enclosed area would form a solid triangle of equal distribution without any of the small black upside-triangles (representing mostly unhit areas) - and not all areas of the triangle are accessible at any one time - hence has more to do with the snake's intelligence for want of a better interpretation.

2) How to apply fractals to roulette?

Observation A: It seems the most obvious mechanism behind the fractal example is dependency: each trial forms a dot and draws a line based on where the previous dots and lines were drawn - forming a trail of dots and lines that are connected by their relationship = dependency. If the fractal was based on independent trials then the starting point would always be the same, and the dots and lines would never form a predictable pattern - unless they were not equally-likely. However, the fractal example is in fact using equally-likely outcomes.

How to apply this mechanism to Roulette? Answer: cycles! A cycle begins and ends on a repeat, which is carried over to the next cycle as the defining element:

1231

11

122

22

22

212

The cycle always has more chance of ending in order 1 because the outcome depends on the starting partition:

2... more chance finish on 2.

Observation B: The fractal example is based on 3 outcomes that are used to create multiple outcomes - representing the many possible spaces where a dot/line could end up being drawn within the enclosed area of the triangle.

With cycles we can use 3 outcomes to generate multiple outcomes too, i.e. 9 Dozen Options:

Order 1, Dozen 1

Order 1, Dozen 2

Order 1, Dozen 3

Order 2, Dozen 1

Order 2, Dozen 2

Order 2, Dozen 3

Order 3, Dozen 1

Order 3, Dozen 2

Order 3, Dozen 3

Observation C: some areas of the triangle are hit more than other areas of the triangle.

With Cycles, Order 1 is going to hit more than Order 2 or 3 - but just betting Order 1 doesn't result in any edge because risk/reward is the same ratio as betting Order 2 or Order 3 = equivalent.

Observation D: the most important! The snake cannot repeat the same outcome as the last outcome, i.e. it cannot return to the previous dot immediately - but it can return there later after several trials have taken place!

How to apply this to Roulette? In 9 Dozen Options (see above), Order 2 Dozen 1 cannot immediately follow Order 1 Dozen 1 - but it can return there later in a set of cycles.

What other observations are there and what further refinements must we make to our dozen cycles to make this applicable? Or if it's not applicable then why not?