To do controlled studies the approach must be algo

To do controlled studies the approach must be algorithmically automated. I've been playing around with that a few years but it's immensely hard so I just built some helper tools and still 'drive' by the eyes of myself and a mathematician trading partner. Automating it is almost as hard as making computer vision for a self-driving car. Humans are biased (hence my reluctance to call a drawdown to $1.60 given unbalanced portfolio but machine learning is even more biased by training datasets, and algorithms are always more brittle than human intuition. It only took Jim Simons and his Renaissance hedge fund team 20 years to get their algorithms right, make regular 70-90% gains per year and charge hedge fund customers 4% a year plus 40% of profits.

But if you follow along you can surmise our core technique and start looking for the nested fractal structures yourself. Convince yourself watching them evolve over again and markets will seem much less chaotic. The psychological benefit alone is priceless. How many newbie retail who bought in $3-10 range over the summer could keep buying instead of sell at $1.62? Even Jim Simons doesn't trust his machines and still has the urge to shit his pants and sell at major bottoms.