trading

Recommended several bloggers focusing on AI, US stocks, cryptocurrencies, quantitative trading, etc., and briefly introduced their characteristics and achievements.

A deterministic guardrail tool is needed between an AI agent's decision and its irreversible actions such as trades or sending funds, to ensure safety.

A thread introducing Loop Engineering as a solution to the common problem of quant strategies that backtest perfectly but fail in live trading, emphasizing the need for iterative optimization.

Goldman Sachs open-sourced a quantitative trading framework used internally, including Claude AI skills, enabling users to build trading strategies.

TradingAgents is a multi-agent trading framework that uses multiple AIs working together to automate crypto and stock trading, supporting real-time data ingestion, automatic strategy generation, and continuous optimization. It has gained nearly 90k stars on GitHub.

Coinbase has launched a feature allowing AI agents to trade and make payments using its platform, expanding the capabilities of autonomous agents in the crypto space.

This paper reviews and audits execution realism in LLM-based trading research, proposing clearer reporting standards for reproducibility and evaluation comparability.

An open-source framework simulates a real trading floor with multiple AI agents performing fundamental, sentiment, technical, and risk analysis, compatible with major LLMs like GPT-5, Claude 4, and Gemini 3.

A user reports on giving Julius AI $1K on Robinhood to manage a portfolio, noting that despite providing tools, the AI still requires extensive context. The AI made one trade (AMD).

A curated list of 10 open-source projects that provide powerful AI and developer capabilities, often replacing costly commercial software—including AI trading agents, multi-model chat interfaces, video generation, voice cloning, and more.

Limitless released a full developer stack designed for coding agents to operate prediction market trading, aiming to replace manual wiring with autonomous agent execution.

An analysis of how autonomous AI agents are beginning to trade, negotiate, and manage crypto assets 24/7, and why privacy technologies like zero-knowledge proofs are critical to prevent front-running and surveillance in this emerging agentic trading market.

Introducing Co-Invest, the first way to trade directly through ChatGPT and Claude, enabling AI-powered trading.

A developer built a local web GUI for the TradingAgents multi-agent LLM stock analysis framework, supporting various LLM providers and adding features like live pipeline visualization, a report reader, and multi-session chat.

A Chinese developer recorded a Claude Code proxy setup tutorial, accidentally revealing his Polymarket wallet running with a profit of $868,862. The video went viral; he deleted it, but the wallet is still running.

A builder reflects on their AI trading agent Pip, which uses 17 hard gates instead of structured context files like CLAUDE.md, and questions whether the workspace is for the agent or the builder.

A curated thread listing 50 GitHub repositories that promise to reduce AI subscription costs by up to 80% and generate automated income through job applications, video content, social media bots, and algorithmic trading.

This tweet introduces three financial/engineering tools open-sourced by top quantitative institutions such as Jane Street, Goldman Sachs, and J.P. Morgan: magic-trace (high-precision process tracing), gs-quant (Python package for derivatives pricing and risk management), and Perspective (real-time data visualization tool), helping quant enthusiasts gain institutional-level technical capabilities for free.

This paper presents a systematic survey and evidence map of 77 studies on LLM-based trading agents, finding that architectural experimentation is expanding rapidly but evaluation protocols, execution semantics, and reproducibility remain critical bottlenecks.

A tweet reports that Renaissance Technologies' entire trading structure has been leaked on GitHub under an Apache 2.0 license, amassing 76,800 stars.