Independent research by Dylan Siegel

Trading algorithms.
Tested in public.

I build neural network trading algorithms from tick data, test them out of sample, and publish what holds up. And what doesn’t.

Follow the research Start the free course

BTC · Standby

What this site is

The research, the course into it, and the systems that pass.

Free

Follow the research

Neural networks trained on real tick data, tested out of sample, and written up as each run finishes. What passed, what failed, and why. Many notes have a companion video.

Read the latest → Free

Take the course

Start from zero: how markets work, how to build and test a signal, how to write the program with AI coding tools. Seventeen short lessons on one running example. No experience needed.

Start at part one → Paid

Run a finished system

Trading algorithms built from research that passed every test. Closed source, with tuned settings, execution rules, and monitoring. Paper trade each one in your browser first.

See the systems →

Latest

New notes, as the work happens.

All research →

Sample draftMemory

How long should a signal remember?

A proposed comparison of short and long memory windows using the same input signal.

Sep 13, 20261 minIntermediate

Sample draftExperiments

What happens after a burst of trades?

A study outline for separating a sudden increase in trading activity from a directional prediction.

Sep 12, 20261 minIntermediate

Sample draftExperiments

When two signals tell the same story

A proposed redundancy check before adding another feature to a research model.

Sep 11, 20261 minIntermediate

How the work is done

Every claim comes with its evidence.

Test before you believe.An idea is a guess until it has been run against real market data with a stated sample and a stated result, net of costs. Ideas that fail get written up too.
Let the evidence decide.I choose the model to fit the problem, then check how it behaves on unseen data, in fast markets and quiet ones. Its results have to justify its complexity.
Open research, closed systems.The course, the experiments, and the code are public. The finished, tuned algorithms are the only thing you pay for.

Systems

The strategies that passed.

Closed-source trading algorithms built from the open research. Every method they use is taught in the course or written up in the notes. Paper trade each one against live market data in your browser before a dollar is involved.

All systems →

BTC perps · ETH perpsLive preview

QD·1