The felix.trade blog
Research the market.
Build better agents.
Practical writing on market research, backtesting, execution, risk controls, and the systems around autonomous trading.
Latest
When Should an AI Trading Agent Pause After Repeated Errors?
Repeated execution errors can signal deeper problems. This guide explains when to pause an AI trading agent, how to use owner‑authorized limits, and what operational safeguards to apply.
Archive
All writing
Every approved article is listed here, newest first.
How to Allocate Capital to an Experimental Trading Agent
Learn practical steps for allocating capital to an experimental trading agent, including owner‑authorized limits, emergency stops, and data integrity checks.
What the Bid‑Ask Spread Reveals to a Trading Agent
The bid‑ask spread is a key market signal that tells a trading agent about liquidity, price efficiency, and execution risk. This article explains its meaning, practical uses, and the uncertainties involved.
Choosing the Right Interval for an AI Trading Backtest
Selecting the proper interval for an AI trading backtest is crucial for realistic results. This guide explains the trade‑offs, practical steps, and common pitfalls to help you design robust experiments.
Why Consecutive Agent Errors Should Trigger a Pause
Back‑to‑back errors from an automated trading agent can cascade into larger losses. A well‑designed pause mechanism limits exposure, ensures reliable state, and creates a safe window for investigation.
Designing an Effective Activity Timeline for Trading Agents
A well‑designed activity timeline helps users understand what an automated trading agent did, when it acted, and why. This guide covers essential data, visual cues, and safety controls for building transparent timelines.
Owner Authority vs Agent Authority: Understanding the Core Differences
Owner authority controls the wallet and sets policy, while agent authority is limited to specific actions like placing trades. This article explains their distinct roles, limits, and the safeguards needed for secure automated trading.
Why Positions Must Be Reconciled After Every Agent Order
Reconciliation after each agent order ensures the system’s view of positions matches reality, helping detect failures, enforce limits, and keep risk controls effective.
Understanding Daily Notional Limits for Trading Agents
Daily notional limits cap the total value of trades an agent can execute in a day, helping owners manage exposure while preserving algorithmic freedom. This article explains the mechanics, configuration, and practical considerations.
How to run an AI trading agent with real-money controls
Real money does not mean unrestricted authority. Keep custody, scope the key, bind the caps, preview larger orders, reconcile ambiguous outcomes, and maintain an emergency stop.
MCP and direct API integration paths for AI agents
Compare the live verified MCP and direct API paths, with owner authorization and runtime availability governing use.
One API for stocks, crypto, perps, options, and prediction markets
Five market types, one normalized interface, dollar sizing, and runtime status for current availability.
Non-custodial trading for AI agents: trade authority without withdrawal authority
Non-custodial rails separate owner custody from agent permissions: a trade-scoped key cannot choose an arbitrary withdrawal destination, while owner-key, module, venue, software, operational, and market risks remain.
MCP trading architecture for Claude, Cursor, and Codex
After installation and owner authorization, MCP can expose Felix tools in supported AI clients. Venue truth determines whether an order is accepted or filled.
How to give an AI agent a trading account
Give an AI agent a scoped account, owner-signed limits, and one normalized interface. Runtime status reports which market capabilities are currently enabled.
What is agentic trading?
Agentic trading uses software agents to research, decide, and, after separate owner authorization, submit real-money actions inside explicit controls.