Aligrithm is an independent research publication on systematic trading, quantitative research, market microstructure, and adaptive systems. Long-form essays, code notebooks, and architecture breakdowns across eight pillars, built for traders who care more about how markets behave than about hype.
More about the publication →Read in order. Each pillar walks you out of one trap with one new ability. The newest article is the worst place to begin.
Ranked systems churn at the slice boundaries, where names flicker across on noise and you pay for each swap. A no-trade buffer holds them, and only net-of-cost performance tells the truth.
Trading ideas go absurd, familiar, inevitable; the edge is largest at absurd, gone at inevitable. Hunt in uncomfortable places but verify ruthlessly, because most absurd ideas are wrong, not early.
A market is a complex system where crashes emerge from millions of decisions, leaving power-law fingerprints. Fat tails and clustering are permanent; the lens explains markets, it can't time them.
If prices were random they'd diffuse like ink, spreading as the square root of time. Markets diffuse anomalously: faster trending, slower reverting. That exponent is the variance ratio in physics.
Entropy measures how unpredictable a series is, catching nonlinear structure the variance ratio misses. Maximizing it under market constraints produces fat tails: the tails are natural, not a glitch.
The Lévy distribution fits the body of returns far better than the Gaussian, with real power-law tails, and still misses the crashes. No single elegant curve captures the tail that kills you.
The bell curve calls crashes impossible, and they keep happening, because real returns have fat power-law tails. Gaussian models fail in the crisis they exist for. Size for the move you can't model.
For a random walk, variance grows linearly with time. The variance ratio compares real growth to that benchmark: above 1 trends, below 1 mean-reverts. It measures the past, not the future.
Random walk and efficiency are different claims, one statistical, one economic, neither implying the other. Reject the random walk without calling the market inefficient, and that's where edges hide.
Removing yourself is engineering, not willpower: inventory where you intervene, close each point with pre-committed rules, automate what you can, and hand authority from the live you to the calm you.
Equal dollars isn't equal bets; the loudest instrument owns your P&L. Size inversely to volatility so each position carries the same risk, and the volatility you scale by is only an estimate.
The discipline premium is the return for not making the mistakes everyone else makes. It can't be arbitraged away because instinct doesn't learn, but it multiplies a sound edge, not replaces one.