Forkast, an agentic prediction market platform, announced the launch of ATLAS, a purpose-built agent harness that enables users to deploy autonomous trading agents across live markets with just a few clicks. The system, whose name stands for Adaptive Trading & Learning Agent System, is designed for esports, crypto, and major sports events, and requires no coding or trading desk experience.
ATLAS agents feature persistent memory, tool use, and a self-reflection loop that continuously improves decision-making. Users configure their agent’s market focus, risk appetite, and trading behavior, and the agent operates 24/7, acting on high-conviction opportunities and reporting back via natural language chat for strategy refinement. Under the hood, it is a full agent harness, not a simple LLM wrapper, with an architecture built to support multi-agent orchestration as the product evolves.
The launch comes amid explosive growth in prediction markets: monthly trading volume skyrocketed from under $5 billion in September 2025 to roughly $24 billion in April 2026, attracting hedge funds and federally regulated platforms. Yet retail traders, who drove that growth, remain stuck with outdated interfaces. Forkast founder Chris Gonsalves argued, “The 2027 trader doesn’t click. They orchestrate. Retail doesn’t need to beat Citadel to benefit from this—they need to beat the version of themselves that was click trading at 1am on gut instinct.”
Forkast is particularly focused on esports, which it sees as the fastest-growing and most underbuilt vertical in prediction markets. ATLAS integrates high-frequency data from GRID on player performance, match state, and odds movement, giving agents a real-time edge. Alongside the launch, Forkast introduced the TVT Trading League, where users deploy agents to compete for real USDC prizes, as well as esports player prop markets and the Battle Bar, a visualization that animates live market odds between 0% and 100%.
