01 · Signal Engine
The backtested desk
Scheduled drops · BTC & ETH
A vision model reads 4H, 1H, 30M and 15M charts under a fixed playbook and returns a structured signal.
Deterministic gates reject anything malformed before you ever see it.
- Entry zone, stop and two targets on every signal
- A candle-replay tracker follows each signal to its conclusion
- Track record with win rate, profit factor, drawdown and R-multiples
- Position-size calculator from your own risk settings
02 · Analysis Studio
Six models, one chart
Any spot pair · 1W → 15M
Pick a pair, timeframes and a mode. Every model receives the identical context — charts with the structure
drawn on, plus an exact numeric digest — and answers in the same report format.
- Up to six models side by side, with a consensus bar
- Modes: full technical, support bounce, trendlines & breakouts, smart money, swing plan, levels & risk
- Trade-idea risk/reward recomputed from the geometry, not taken on trust
- Cost and latency shown for every model
03 · Breakout Monitor
Only charts that bounce
Every 15 minutes · 1D, 4H, 1H
Scans the top 80 spot pairs for support-trendline structures — and filters out what shouldn't be traded:
assets in a continual downtrend since listing, and lines that don't produce real bounces.
- Six alerts: breakout, retest hold, failed breakout, support test, bounce, support lost
- Breakouts need volume at 1.5× the 20-bar average
- Desktop and sound alerts; one click sends the chart to the Studio
- Runs on closed candles and spends no model tokens
04 · Model Hub
Any model you like
15+ providers · local models
Claude, ChatGPT, Gemini, Grok, DeepSeek, Kimi, Mistral, Qwen, GLM, the OpenRouter catalogue, Ollama and
LM Studio — or any OpenAI-compatible endpoint.
- Fetch the live list of models your key can access
- Every call metered for tokens and estimated cost
- Keys stay in a local file — never in the database, never sent to the browser
- New models work without code changes