Skip to main content

​
Prerequisites

  • You have connected your exchange account to Gigabrain in integrations.
  • You have funded your account and have api credits available.
  • For getting notifications you need to connect your telegram account to Gigabrain in integrations.

​
Getting Started

​
Step 1: Navigate to Agents

  1. Visit Agents page
  2. Click Create New Agent to begin

​
Step 2: Configure your agent

1

Define strategy

Write your agent’s strategy in natural language in the input box. Be specific about:
  • Entry conditions (e.g., confidence thresholds, market state requirements)
  • Exit strategy (e.g., invalidation points, confidence drops)
  • Risk management rules (e.g., position sizing, stop losses)
Example:
Analyze the 4H BTC chart, microstructure data, and whale positioning
every 4 hours. Place a trade with 10% of account margin when:
- Confidence exceeds 75%
- Market State is Risk-On
- Funding rate is neutral or favorable

Re-analyze open positions every 30 minutes. Close position if:
- Invalidation point is hit
- Confidence drops below 50%
- Macro Analyst flags high-impact geopolitical event
2

Confirm your strategy

Review the generated strategy details:
  • Analysis Frequency: How often your agent will analyze market conditions (Real-time, Hourly, or Daily)
  • Position Sizing: Fixed amount or percentage of portfolio
  • Leverage: Leverage multiplier for trades
  • Confidence Threshold: Minimum confidence required to execute
  • Specialists: Which analysts will be consulted
Make any necessary adjustments and click Confirm to proceed.

​
Monitoring Your Agent

​
Reasoning tab

Track your agent’s key metrics:
  • Confidence: Mean confidence level of executed trades
  • Market State: Current market state
  • Specialists: Specialists consulted
  • Instructions: Instructions given to the agent

​
Learnings Tab

Your agent’s adaptive learning system tracks:
  • Outcome Monitoring: How predictions matched actual price action
  • Strategy Refinements: Weight adjustments to different variables
  • Pattern Recognition: Recurring market conditions that led to wins/losses

​
Changelog

Audit trail of all agent modifications:
  • Configuration changes
  • Instruction updates
  • Performance milestones