Prediction Markets — Forecasting, AI, and the Future of Information In this episode of Investment Wars, Joe Halpern sits down with Zvi Mowshowitz, a deep thinker with hands-on experience in professional trading, prediction market design, and AI, for a wide-ranging conversation on what prediction markets are, what they're actually good for, and where they're headed. Prediction markets get a lot of hype. But are they genuinely useful economic tools or just sophisticated entertainment? This episode cuts through the noise, exploring how well-designed questions create real informational value, where prediction markets outperform traditional hedges, and why AI may be the next great participant, not the replacement, for these markets. Key Topics Covered What Makes a Good Prediction Market Question: Why specificity, measurability, and verifiability are everything, and why questions like "Is Jesus coming back?" generate nothing but risk-free arbitrage, while questions like "How many ships will pass through the Strait of Hormuz by [date]?" can be genuinely valuable.Real-World Hedging Use Cases: How a shipping company like Maersk could use a well-structured prediction market as a direct, targeted hedge, perhaps more efficient than buying oil futures, which measure far too many variables at once.The Insider Trading Debate: Why insider trading, so clearly harmful in equity markets, is far more nuanced in prediction markets and why thinkers like Robin Hanson argue it should be permitted, since insiders reveal information the market needs.Super Forecasters vs. AI vs. Prediction Markets: Zvi defines super forecasting, explains where super forecasters excel (most near-term political and geopolitical questions) and where they reliably fail (long-range predictions, fat-tail events, and AI capabilities). He then explains how well-prompted AI is now outperforming all but the very top human forecasters and what that means for prediction markets going forward.The Centaur Model of Decision-Making: Drawing on the arc of chess, from human dominance, to AI dominance, to a brief window of human-AI centaurs outperforming both, Zvi explains why we're in the centaur phase of forecasting right now. The winning approach: challenge the AI, push back on its logic, and let the best reasoning win.World Models and Decision Systems: Why building a causal, gear-level model of how the world works, rather than relying on vibes or headlines, is the foundation of better investing and better decisions. Not just for professionals. For everyone.The Real Costs of Participating in Prediction Markets: Beyond bid-ask spreads and platform fees, Zvi makes the case that mental transaction costs, time, focus, and the risk of addiction, are the most underappreciated barriers to participation. His framework: trade when the market justifies those costs, and not a moment before.Why Markets Are Still the Best Aggregators: Even as AI improves, prediction markets may become more powerful, not less, because they synthesize inputs from humans and AIs alike, including sources that hold private information they'd only reveal in exchange for profit.Subscribe to Investment Wars on YouTube, Apple Podcasts, and Spotify to ensure you never miss an episode. Hear more from Zvi Moshshowitz, subscribe to Don’t Worry About the Vase on SubStack. If you are a financial advisor interested in partnering with Obsidian CIO or would like to suggest a future guest, reach out at ocio@obsidiancio.com. --- IMPORTANT DISCLOSURE: Obsidian CIO sponsors the podcast to further education and critical thinking about the factors that affect markets and investing. The podcast does not provide investment advice. Investment advice is offered only to clients of Obsidian CIO who have entered into an advisory agreement and with whom Obsidian CIO has identified individual objectives, risk tolerance, and other investment needs.
