While human traders grapple with the psychological toll of a volatile market, a new class of autonomous actor is entering the fray.
Olas, the firm behind the agentic ecosystem, has launched Polystrat, a bot designed to trade on prediction markets without the baggage of human emotion. The launch represents a shift in how prediction markets are viewed as they transition from niche hobbies to the next frontier of global financial systems.
The rise of the truth machine
Prediction markets have experienced a meteoric rise, with monthly trading volumes surging from less than $100mn in early 2024 to more than $13bn by the close of 2025. Polymarket has emerged as the dominant force in this sector, recently cementing its cultural relevance through a partnership with the Golden Globes to provide real-time probability data during the live broadcast.
Olas argues that as the underlying open source models improve, these autonomous agents will eventually outperform their human counterparts. "They might not have insider knowledge, they might not feel excitement or despair, so instead it is all very numbers based," David Minarsch, CEO of Valory and a founding member of Olas, told Sandmark. The goal is to convert model improvements directly into a trading edge, using the generalist nature of AI to process vast quantities of real-time data that would overwhelm a human analyst.
Security in the agentic economy
The technical architecture of Polystrat is built on Pearl, an app store for AI agents operating onchain via Gnosis and Base. While the promise of "set and forget" trading is attractive, it brings significant risks. The industry is still grappling with whether AI agents can be trusted with wallets given the potential for hacks or malicious exploits. Minarsch noted that Polystrat uses self-custodial Safe accounts and operates within a constrained framework to prevent a single "hallucination" from causing widespread financial ruin.
Beyond the immediate mechanics of trading, the project is a bid for sovereignty. Minarsch warns that the window for decentralized AI is closing as large labs create a "vortex" around centralized infrastructure. By using public blockchains as a counterbalance, Olas hopes to ensure that the future of automated finance remains in the hands of users rather than a few Silicon Valley giants.
For now, the "truth machine" is running on code, and the humans are merely watching the scoreboard.
