Michael, a software engineer and data analyst, presents a method to fully automate crypto and stock portfolios using AI, specifically by connecting Claude (AI assistant) with Signum (data and automation platform) and HyperLiquid (exchange). The video demonstrates a real-money setup that scans the market daily, executes trades based on a pre-defined trading prompt, and manages risk automatically.
🔍 The Main Problem/Opportunity
- TradingView automation is limited – you must manually pick which coins to watch, then set up price alerts. You can catch pumps, but you never know which coin will matter next.
- AI removes the guesswork – Claude uses Signum’s Trend Radar to scan the entire market and identify both long and short opportunities before the move happens. This works in bull and bear markets alike.
- The opportunity is a fully autonomous system that decides which coin to trade, when, and with how much money – all without human intervention.
âś… Key Benefits Over TradingView
| Aspect | TradingView Automation | AI (Claude + Signum) Automation |
|---|---|---|
| Coin selection | You pre-select symbols | AI scans and picks dynamically |
| Position sizing | Fixed or alert-based | AI decides % of portfolio per trade |
| Market direction | Usually long-only unless manually set | AI goes long and short automatically |
| Setup complexity | Requires strategy coding & alert management | One‑time prompt copy‑paste, daily execution |
| Adaptability | Static alerts | AI re-evaluates daily with fresh data |
The AI “levels up” from when to buy/sell to which coin to even trade – a fundamental shift in automation capability.
🛠️ Step‑by‑Step Setup (Claude + Signum + HyperLiquid)
- Create a Signum account (free at signum.money) → click “Create Bot” → select “Futures” → choose “HyperLiquid”.
- Connect HyperLiquid – fund your account first, then generate an API key (wallet address + private key). Paste those into Signum’s bot creation fields. Choose any trading pair (the AI will override it).
- Enable MCP (Model Context Protocol) – in Signum, go to the MCP tab → copy the Remote MCP Server URL → In Claude (desktop app or web), add a custom connector named “Signum” with that URL → connect and allow permissions.
- Test the connection – open a new chat in Claude with the Signum connector active. Ask “Please summarize my Signum account for me and show me how my funds are distributed in a diagram.” Claude will fetch bots, assets, and even draw a pie chart.
- Create a routine (daily automation) – in Signum, open the bot’s MCP tab again → click the Routines URL → in Claude Code, click “New Routine” → name it after the bot.
- Copy the trading prompt – from Signum (e.g., “Trend Radar Gaussian Channel Crypto Long and Short Version 2”) → paste it into the routine’s instructions. The prompt already contains the bot ID.
- Schedule – choose daily execution at half past midnight UTC (ensures the Trend Radar data has updated). Select model Sonnet 4.6, effort “medium”, and keep only the Signum connector to save tokens.
- Create – the routine is now fully automatic. Every day at the set time, Claude will execute the prompt: scan the market, decide trades, place orders on HyperLiquid, and send a summary email.
đź“Š Performance Results (Shown in the Video)
- Gaussian Channel Trend Radar (crypto, long+short)
- Net profit: 7,492%
- Max drawdown: 35.6%
- Trades: 2,689 (since 1 Jan 2018)
- Equity curve shown – impressive growth with controlled risk.
- Stock prompt (beats S&P 500) – 506% profit in 609 trades since 2018, with very low drawdown. For comparison, buy‑and‑hold S&P 500 returned only 178% with 18.82% max drawdown over the same period.
- Higher‑portfolio prompts – one achieves 12,226% net profit; another trades crypto’s riskiest segment at 8,417% since 2018 with extremely low drawdown.
All KPIs were verified through “massive simulations” before real money was deployed.
đź’ˇ Conclusion & Call to Action
The AI system is live and profitable. The video demonstrates a real execution with 200 USDC: Claude opened a short position and immediately showed profit. The creator recommends starting with minimum 200 USDC for the given prompt to avoid dust orders.
For access to the full suite of prompts (including the 12,226% and S&P‑beating stock prompt), join the Auto Trading Master Class. It covers risk management, portfolio automation, and the complete system. The creator notes that for his personal large‑coin holdings (Bitcoin, Ethereum, Solana) he does not use AI – he uses his own manual strategy. He invites questions in the Telegram group.
Final takeaway: Manual TradingView automation is outdated. AI‑driven automation (Claude + Signum + HyperLiquid) is now easy to set up in ~10 minutes and delivers verified, superior performance across both bull and bear markets.