Skip to content

What It Does

Ryan Bayne edited this page Nov 9, 2026 · 3 revisions

What It Does

Status: active Category: E (product / user) Created: 2026-10-31 Last updated: 2026-10-31 Owner: TCS team

Trading Command Suite is a self-hosted trading application. It scores markets, gates those scores through explicit rules, and — if you let it — places orders through your own broker. Everything below runs on your machine, against your accounts.

👉 See it before reading about it: 17 screens from the alpha — each captioned in plain language, with a glossary for the words the app uses.

⚠️🛑 It is not configured to profit by default and no strategy it ships has been shown to be profitable. Read the Financial Disclaimer first.


The areas of the app

Area What it is for
Scoring The META.Stack engine. Protocols are built from directives; a protocol scores a symbol and everything downstream is gated on that score
Signals What the engine currently rates highest, with the reason and the ladder of every symbol it looked at
Protocols Build and inspect protocols — which directives they include, their thresholds, and their recorded performance
Trading Bots, positions, trade journal, exit signals. A bot binds a protocol to a symbol with sizing and limits
Portfolio & accounts Balances, equity, exposure and per-position state across every connected broker
Watchlists Which symbols are monitored, and by what
Research Symbols, fundamentals, earnings calendar, provider comparisons
Lab Backtesting, protocol and strategy comparison, theory registry, live data playground
Planner Trading style, objectives, capital — the constraints a bot must respect
Status Health of every subsystem: database, providers, bridges, queue, architecture maps
Settings Automation, risk, scoring defaults, appearance and accessibility
Development Engineering tooling: pipeline graph, trace inspector, QA checklist, page walkthrough

What makes it unusual

A decision trail, per trade. Every dispatched order can be traced back through the gate that allowed it, the score that produced it, and the directive that moved the score. GET /api/automation/decision-trail. Most retail systems cannot answer "why did it buy that?" — this one records it as a first-class object.

Explicit gating, not vibes. A score alone does not cause a trade. It must clear the protocol's gates, the bot's thresholds, the risk chain and the broker's capability check for that symbol. Each refusal is recorded with a reason rather than swallowed.

"Cannot judge" is a real answer. The app distinguishes fine from could not determine — an empty list, a failed query and a genuine zero are three different things, and they render differently. This is the single most repeated lesson in the codebase: plausible output with no error is the defect class that costs money.

Honest failure over convenient success. A prebuild gate list (39 validators, counted in docs/GENERATED-FACTS.md) fails the build on contract violations, duplicated implementation, undocumented decisions and stale generated docs.

Backtesting and A/B comparison as the measurement floor. Protocol variants can be compared side-by-side, and a verdict refuses to name a winner on a sample too small to support one.

What it does not do

  • It does not give advice. A score is an opinion computed from data.
  • It does not hold money. It is not a broker; it places orders through yours.
  • It does not guarantee data. Prices, fundamentals and calendars come from third parties that have no obligation to be right or on time.
  • It does not claim profitability. See the disclaimer, and the FAQ.

Clone this wiki locally