Resources

Free code from the One Person Trade Desk.

The repo from the videos: Sierra Chart files, ACSIL studies, Claude Code prompts, and runnable backtest-bias demos. Public, practical, and updated as each video ships.

Get the repo

One list. Every component, as it ships.

Drop your email and the repocomes straight back — plus a short brief each week: what’s working on the desk, what isn’t, and how to apply it to yours.

One click to leave the list whenever.

What's in it

One component per video, and every one of them runs. The method is always in the video, free.

Prop firms

Sierra Chart → Lucid (Rithmic) checklist

The click path for running five Lucid accounts off one chart: one lead account connected through Rithmic, four followers on the Trade Copier, and the per-view Trade Account rule that silently splits the route when you miss it.

Starts from the chartbook in the same repo. Carries the pass condition for proving the route on an evaluation account first, and the reason the copier is not atomic — it sends separate orders, so followers can fill late, at another price, or not at all. A click path, not a strategy.

Open the checklist
Prop firms

Sierra Chart → TopstepX mirror

Trade a Topstep account from the Sierra Chart chart and DOM you already read the market on. You click the order; the mirror copies it to TopstepX through the ProjectX API.

Its own repo, free, MIT, runs locally — Topstep requires trading activity to originate from your personal device. It contains no strategy and cannot open a position on its own. Which Topstep accounts allow this is stage-gated; the guide quotes Topstep's own rules.

Get the mirror
Prompt

Order manager starter prompt

The seven requirements that separate a signal from a system that can't run away from you — arming switch defaulting to OFF, bracket on entry, one position at a time, size as an input, emergency flatten, mismatch handling, bar-close gate.

Paste-ready. It's the chassis, not the car: hand it to Claude Code and it writes the order manager around your own signal. The reasoning behind each of the seven is the video.

Read the prompt
Sierra Chart

The chartbook + the ACSIL study

The chart every video is filmed on, as a Sierra Chart chartbook — ES and NQ, the opening-range study, a footprint and a DOM. Ships with the ORB signal study in C++ and a deploy script that drops both into place.

DEPLOY.md walks the install and names the four build gotchas that cost me an afternoon each. The one-shot prompt that wrote the study is in there too, so you can rebuild it for your own setup instead of using mine.

Open the Sierra folder
Runnable

The Fit / Fill / Fluke gauntlet

Three tests for the three lies every backtest tells: overfit entries, fills it never got, and a result the market handed it for free. One command runs all three.

Runs on synthetic random-walk data on purpose — data with no edge in it — so every dollar it reports as profit is provably fake. You watch each bias manufacture money, then point the same tests at your own results. numpy only, offline, a few seconds.

See the three tests
Runnable

Sierra Chart .scid tick reader

A teaching reader that parses the real .scid record format byte by byte, rebuilds bars from the ticks, and feeds them straight into the honest-vs-naive fill test.

The whole honest-fills argument turns on one bit per bar — did the high print before the low? That bit lives in the ticks, not in OHLC. This is how you get at it. Writes its own synthetic tick file, so it runs with no market data.

Read ticks.py
TradingView

ORB indicator + strategy (Pine)

The worked example the series starts from — an opening-range breakout indicator and its strategy in Pine Script, plus the one-shot prompt that builds both from scratch.

Paste into TradingView's Pine Editor for the fast 'is there a pulse?' read. Then run the same setup through the gauntlet to find out whether the pulse was real.

Get the Pine files
Structure

The desk layout

The repo's folders are the desk — research, backtests, indicators, sierra, journal — mapped to the ten roles a fund pays ten people to fill.

Clone it and you have the skeleton the rest of these pieces drop into. CLAUDE.md gives Claude Code the project context so it knows where things go without being told twice.

Browse the repo

Which platform should you actually run?

Side-by-side breakdowns of the platforms and prop firms this desk gets asked about most — what each one does, what it costs, and which one is worth the switch. See all comparisons.

What’s free and what isn’t. Everything above is the teaching layer: the method, on synthetic data, so you can see the lie with nothing hand-waved. The production harness — the real .scid reader over live exchange files, the parquet cache that replays years in seconds, the multi-gate validated kernel — runs on the live desk. The method is never gated. What’s paid is the data and the engine.

The repo is public either way — browse it on GitHub. The list is how you get told when it changes.