Pillar
Real prices are random numbers with memory, so an EMA of noise plus a cumulative sum makes a passable synthetic chart. Generate thousands to put error bars on a backtest, but never to prove edge.
Fixed-lag cross-correlation assumes the delay between two markets never moves. It does. Dynamic time warping aligns two series with a stretchable path, scoring shape similarity and recovering a lead-lag that varies through time.
Profiling a slow algo points at JSON decode, not your alpha. Put the simple hot path (decode, filter, forward) in Rust and keep the complex stateful brain (OMS, reporting, unwinding) in Python. Split at the JSON, forward binary, and match your backtest exactly.
CryptoMamba, a compact Mamba SSM, forecasts Bitcoin's next close and trades $100 into $262. But it never runs the naive or buy-and-hold baselines, on one bull year of daily bars.
The theory under the ML arc: why fitting the past predicts the future, what each parameter costs in variance, and why complexity's tax shrinks only as one over the square root of your sample.