Theseus is an experiment in applied philosophy. Our first project is a trading agent that forms explicit, internally consistent beliefs about the world (economic, political, and philosophical) and invests accordingly.
Today’s frontier models can recite theories verbatim but cannot hold one. You may ask the same question twice and it will argue both sides, or follow its chain of reasoning and see that it quietly contradicts its own premises. We believe that raw capability has outrun judgment in models, and that judgment — the discipline of maintaining a coherent worldview and reasoning from it under uncertainty — is now the binding constraint on what AI can do. The next generation of important systems won’t just know things—they will consistently believe them and act on them.
More importantly, as agents’ capabilities increase, the differentiation in the work they do will decrease, tending towards commoditization. True differentiation then lies in the beliefs imbued in the system and the curation of what it learns from (this is true of all successful people). Value, then, converges to the worldview that agents hold.
We are deliberately testing this in markets first. Reasoning quality is hard to measure almost everywhere except for markets, because they grade your worldview daily in dollars — objectively and continuously. Yet much of quantitative finance has gone the other direction. Namely, mining statistical patterns while staying agnostic about why they exist. We’re making precisely the opposite bet, that in a world where the future stops resembling the past, the remaining edge belongs to agents who can actually reason about why things happen.
Concretely, this means that our agent trades macro and structural themes at horizons of months to years, where the event count is too low for statistical regularity and where pricing the future requires a causal model of how institutions, incentives, and ideas interact. Ideas sit at the top of that causal chain. The largest macro trends begin as philosophical tendencies, a shift in what societies believe about money, sovereignty, or progress, that cascade into policy, then into capital flows, and only then into prices. While most firms trade the last link, we can achieve the highest alpha by operating one level of abstraction higher at the trend’s point of inception. In other words, we want to create the world’s best discretionary macro trader, whose edge was always a framework held with conviction and the judgment to know when it applied.
We are a tight-knit, small group of friends from the University of Chicago with a variety of interests across economics, mathematics, and philosophy. But the one thing we share is our love for the discursive method that gets us closer to the truth.
Nothing published yet — check back soon.
Nothing to share publicly yet.
No posts yet.