Superior-Trade/trading-terminal

What it claims

The terminal is a Next.js-style web app on port 3200. You can run it yourself or use the hosted build at terminal.superior.trade. The hosted build is described as "this repository with a login in front of it." Both talk to the same Superior Trade API and trade the same accounts, so self-hosting changes who runs the process, not what backend executes trades. The page says the interesting part is the gate between "the model wrote it" and "it is trading your funds." Every validator rule exists "because a strategy once reached production without it."

How it works

The agent's input is more than the user's prompt. It reads:

• the user's drawings,
• the visible chart range,
• the plotted indicators,
• a screenshot of the chart.

On "deploy" it writes a Freqtrade strategy. The page's example populate_entry_trend combines a volume > 0 guard ("no signals on dead candles") with Bollinger lower band, RSI < 40 and EMA-50 conditions.

The strategy is parsed and checked before submission. Rejections are not shown to the user. They go back to the model, which rewrites the strategy until it passes. The README does not list the rules; they are in docs/strategy-pipeline.md.

flowchart LR
  A[Chart + drawings + screenshot] --> B[Agent plan]
  B --> C[Generate Freqtrade strategy]
  C --> D{Validator}
  D -- fails --> E[Complaints to model] --> C
  D -- passes --> F[Submit to Superior Trade API]

Once a strategy is running, the terminal shows live PnL, an optional auto-stop timer for time-boxed edges, and a notification inbox.

How to use it

git clone https://github.com/Superior-Trade/trading-terminal.git
cd trading-terminal
npm install
cp .env.example .env.local     # add the two keys below
npm run dev                    # → http://localhost:3200

• .env.example has 38 variables. Only two lack working defaults. The README excerpt does not name them.
• An embedded Postgres writes to ./.data/ and migrates itself on first boot. No database setup and no login screen.
• Use testnet with NEXT_PUBLIC_HL_NETWORK=testnet.

Charts. A fresh clone runs the bundled Lightweight Charts (Apache-2.0). This has been the default since 2026-08-21. It supports these drawing tools, and the agent reads them:

• brush,
• trendlines,
• rays,
• horizontal and vertical lines,
• rectangles,
• fib retracements.

It lacks text notes and indicator studies. Those need TradingView Advanced Charts, which is free but not redistributable. You request it at tradingview.com/advanced-charts ("a day or two") and then run npm run setup:charts.

Agents without the UI: give the agent "Read https://superior.trade/SKILL.md and register." If a coding agent is working on the repo itself, point it at AGENTS.md.

Testing:

• npm test runs unit tests.
• npm run test:e2e runs 19 checks against a running server with nothing mocked. It makes real model calls, uses market data and calls the Superior Trade API. It stops before the two irreversible actions, deploy and withdraw.
• evals/ holds the agent evaluation suites that prompt changes should be measured against.

Limits

• It places real orders with real funds. The validator catches "the mistakes we know about, not the ones we don't." The authors say to read the generated code before deploying.
• Self-hosting means:
• no authentication (see SECURITY.md),
• withdrawals that "stop one hop short" (see docs/withdrawals.md),
• a dependency on the Superior Trade API even though your keys stay local.
• Without TradingView approval, the self-hosted chart has no indicator studies. The agent's indicator-aware reading depends on those being plotted.
• The authors state that neither backtests nor the agent's reasoning are predictions.
• Merged contributions ship to the hosted build. Changes to money-moving code are expected to come with tests.