About Dark Stone Capital

A small quant shop, written down in public.

Dark Stone Capital is a research-first, systematic capital firm. We trade our own book in US equities and crypto, and publish what we learn — model post-mortem's, regime studies, infra write-ups, and the occasional war story from running it live.

01 — What we are

One desk, a model registry, and a strong bias toward writing things down.

Dark Stone Capital is a small systematic shop. We run quantitative strategies across US equities and crypto, with a research practice oriented around regime detection and risk-conditioned position sizing. The work is the point. The publishing is a forcing function — if a study can't survive being written down, it doesn't deserve to size positions.

We are not a hedge fund. We do not manage outside capital. There is no subscription, no course, no paid tier, and nothing to sign up for. The journal is free because it costs nothing to publish, and because writing in public makes the work better.

02 — What we do

Research, execution, writing — in roughly that order, on repeat.

Research. Regime detection, pullback behavior, signal construction, backtest discipline, and the unglamorous statistics that surround all of it. We compare live strategy variants against report-only candidates before anything earns promotion. The question always picks the tool.

Execution. A risk-aware intraday engine, strategy code names, historical replay, cached market data, streaming market context, broker reconciliation, and a journal. We keep it in our own stack so we own the assumptions, the failure modes, and the decision trail.

Writing. The lab notebook is the public output. Long-form notes, published as the work happens. No newsletters. No cadence beyond "when something is worth saying."

03 — How we operate

Five rules that survive contact with markets.

• Small on purpose. Headcount is a feature, not a bug.
• Models drift; document anyway. A bad post-mortem written today beats a great one written six months from now.
• Risk before edge. Position sizing kills more strategies than alpha decay does.
• No black boxes for ourselves. Every signal is interpretable to the person trading it.
• No outside capital. What we trade is our own.

04 — The stack

Boring infrastructure, carefully maintained.

Most edges live or die at the seams. Ours is built in Python, sized on a ONYX risk engine, persisted in Postgres, and published through Ghost. Most things are tested, most things are monitored, and some things are even documented before they break.

05 — The journal

Long-form notes, published as the work happens.

The journal is the front door. If you trade systematically and want to compare notes, that's where to start. Expect regime studies, model post-mortems where the model lost, infrastructure writeups that nobody asked for, and the occasional war story from a bad day at the desk.

Things you will not find here: trade alerts, signals for sale, forecasts dressed as conviction, or anything that asks for your email in exchange for telling you what's already true.