Blog
Prediction market analysis, strategies, and research. 24 articles.
Compute ROI in Agent Economies: A Framework and Early Data
Most AI systems measure cost per token. We measure dollars of compute per dollars of information discovered. Prediction markets make this possible.
Which Prediction Market Contracts Are Institutional-Grade
We Gave 3 AI Agents a Trading Terminal and One of Them Crashed the Market
A market maker, a momentum trader, and a mean-reversion bot — all autonomous Claude agents. 98 trades in 8 minutes, a live reference price oracle, and a $45 billion flash crash caused by a missing price collar. Here is the full session.
We Locked 3 AI Agents in Docker Containers and Told Them to Hack Each Other
Three Claude agents. Twelve OWASP vulnerabilities. One exchange with a million-credit vault. In under 10 minutes, they independently discovered the same critical exploit, raced to patch before being breached, and one of them looted the treasury. Here is what happened.
How We Bet on Peru's Presidential Election with an AI Agent
35 candidates. 40% undecided voters. 48 hours to go. We used prediction market tools to find an 11-cent mispricing window, designed a two-phase Taker+Maker strategy, and deployed $1,000 — all with an AI agent doing the legwork.
The Shape of a Prediction Market Yield Curve
The first time I plotted implied yield against tau across an event family, the curve had the same steep contango shape as a freshly-issued credit-risky bond ladder. Notes from the morning that happened.
Why I Built the Indicator Stack
A personal account of the frustration that produced IY, CRI, EE, LAS, and CVR — and the rule I locked in early that made the whole thing tractable: pure compute first, language model never.
When the Orderbook Is Empty, You Have Information
An empty orderbook is not missing data. It is one of two specific stories — and the second story is the most reliable maker setup I have found on Polymarket.
The Day I Stopped Trusting Raw Probabilities
A specific morning, two Kalshi Fed-decision contracts at almost the same mid, and the realization that the cents on the screen had been hiding the trade from me for months.
Prediction Markets Need Fixed-Income Language
Yield, spread, duration, convexity. The vocabulary of bond desks already exists, and prediction markets are mathematically the same instrument. Here is the dictionary that bridges them.
The Prediction Market Data Stack: From Raw Prices to Actionable Intelligence
The six layers of the prediction market data stack — from raw exchange ticks to executed trades — and how to build or buy each one.
5 Ways to Connect Your AI Agent to Prediction Markets in 2026
Five integration approaches for connecting AI agents to prediction markets — from MCP servers to custom scrapers — with code examples and honest trade-offs.
SimpleFunctions vs Oddpool vs Raw Kalshi API — Which Prediction Market Tool Should You Use?
A practical comparison of three approaches to prediction market tooling: agentic reasoning (SimpleFunctions), data aggregation (Oddpool), and direct exchange access (Kalshi/Polymarket APIs).
Orderbooks Are Fossilized Beliefs
Every resting limit order on a prediction market is a belief someone held strongly enough to lock up capital. The orderbook is not just a price discovery mechanism — it's a geological record of conviction, frozen at the prices where people decided to take a stand.
Three Data Sources That Tell You What the World Thinks, What the World Is Doing, and What the World Is Feeling
Prediction markets are belief. Traditional markets are action. Social media is sentiment. Each alone is incomplete. Together, they form the most complete real-time picture of the world available to any agent.
The Most Important Number in a Prediction Market Isn't the Price — It's the Delta
A price tells you what the market believes. A delta tells you that the market just changed its mind. One is a snapshot. The other is the signal.
How to Read the World Through Prediction Market Prices
A practical guide to translating prediction market prices into world state. What prices mean, what price changes mean, and how to build a real-time world model from market data.
News Tells You What Happened. Prediction Markets Tell You What's Happening.
Headlines are past tense. Prices are present tense. If your agent reads news to understand the world, it's always one step behind.
Prediction Markets Are the Best Real-Time Sensor for World Events
Prices move before headlines. If you want to know what is happening in the world right now, prediction market prices are faster, more honest, and more calibrated than any other public signal.
Abelian and Non-Abelian Groups: Stackable Risk vs Non-Stackable Risk
When the order of events changes the outcome, every model that assumes otherwise is lying to you.
Group Actions and Orbits: Why the Same Event Has Different Value for Different Traders
The mathematics of symmetry explains why two rational traders can look at the same headline and reach opposite conclusions — and why both can be right about different things.
Congruence Classes and Signal: Modular Arithmetic as Attention Compression
The most powerful operation in number theory is also the most violent: division with remainder. What you throw away defines what you can see.
Why Your AI Agent Needs a Thesis, Not Just Data
Most AI trading agents make money for a week, then blow up. The problem isn't the model — it's the architecture. Here's why structured reasoning beats raw data every time.
How Causal Tree Decomposition Beats Vibes-Based Trading
You read the headline, formed a view, bought YES at 55 cents, and watched it bleed to 30. Here is why that keeps happening — and the structural fix that turns gut-feel gambling into systematic edge.