The Price Is the Intervention
On reflexivity, the $40 billion prediction market moment, and why existing coverage misses the point.
On Super Bowl Sunday, more than $150 million has been traded on a single prediction market contract: who wins. On Kalshi alone, 91 percent of all trading volume is now sports. The platform processed $543 million in a single day during conference championships. Robinhood routes more than half of Kalshi's volume. The combined valuation of Kalshi and Polymarket exceeds $20 billion. The industry did $40 billion in volume in 2025, up 400 percent year-over-year.
These numbers describe a market that has arrived. But they do not describe what the market is.
The standard narrative — the one you will read in every venture capital outlook, every crypto fund deck, every podcast with an economist — goes like this: prediction markets aggregate dispersed information. They surface the wisdom of crowds. They produce probability estimates that are more accurate than polls, pundits, and models. The intellectual lineage runs from Hayek through Tetlock through Vitalik. Markets are truth machines. Build more of them.
This is half right. The half it gets wrong is the half that matters.
When Polymarket showed Trump at 64 percent on the morning of the first 2024 debate, CNN cited it on air. Campaign strategists recalculated. Donors shifted money. The prediction market price was not just reflecting reality. It was shaping it. The number became an input to the event it claimed to measure.
George Soros spent four decades articulating this idea. He called it reflexivity: markets are not passive mirrors of reality. They are active participants in the reality they describe. Beliefs alter fundamentals. Fundamentals alter beliefs. The loop never stops.
“Financial markets cannot possibly discount the future correctly because they do not merely discount the future; they help to shape it.”
George Soros, The Alchemy of Finance
Soros applied this to currency markets and sovereign debt. Prediction markets make the loop visible. In an equity market, the feedback between price and fundamentals is obscured by thousands of variables. In a prediction market on a binary outcome — will this candidate win, will this bill pass, will this drug trial succeed — the loop is exposed. You can watch it happen in real time. A market price on whether a candidate wins does not just aggregate opinion about the election. It changes the election.
The Hayekian framework — the one a16z publishes, the one economists teach, the one prediction market proponents recite — treats prices as outputs. Information goes in, an efficient price comes out. Reflexivity says prices are also inputs. They re-enter the system they claim to measure. The distinction matters because it means prediction markets are not converging on truth. They are generating feedback loops that can amplify, distort, and self-fulfill.
This is not a theoretical distinction. It is an operational one. If you trade prediction markets as though prices are passive signals, you will be surprised by how often the signal changes what it signals about. If you build infrastructure for these markets without understanding the feedback loop, you will build the wrong things.
The prediction markets industry in early 2026 is both massive and misunderstood. Consider the structural picture:
Ninety-one percent of Kalshi's volume is sports. More than half comes through Robinhood. The “information finance” narrative — the one used to justify $11 billion valuations — is intellectually compelling but commercially marginal. The actual product-market fit is a sportsbook that is federally regulated as a derivatives exchange, allowing it to operate in states where sports betting is illegal.
This is not a criticism. It is a description of the growth engine. And it creates a specific reflexive dynamic: the more prediction markets succeed as sports betting alternatives, the more state regulators push back. Massachusetts has issued an injunction. New York's attorney general calls these platforms bets “masquerading” as event contracts. Nevada's Gaming Control Board filed a civil complaint against Polymarket. Eight states have sent cease-and-desist letters. The NFL has banned prediction market companies from advertising on its broadcasts — alongside tobacco, pornography, and firearms.
The regulatory response is itself reflexive. Prediction markets grow by being used as sportsbooks. This growth attracts regulatory attention. Regulatory attention threatens the federal preemption argument that makes the growth possible. The market price of the industry's future is an input to the industry's future.
Meanwhile, on the other end of the market, something genuinely important is being obscured by the sports volume. Polymarket's trading volume grew 5x post-election, countering the narrative that these markets are seasonal novelties. Kalshi lists over 3,500 active markets spanning economics, geopolitics, weather, and culture. Google has integrated prediction market data into search results. CNN and the Wall Street Journal display these prices as authoritative forecasts.
This is where the reflexivity gets interesting. When CNN puts a Polymarket price on screen as though it is a fact about the world, it becomes one. The audience internalizes the probability. It changes their beliefs, their behavior, their votes, their capital allocation. The prediction market price, displayed as a neutral measurement, functions as an intervention in the event it measures.
And no one is writing about this.
The existing landscape of prediction market content falls into three categories, none of which addresses the feedback loop.
Trade newsletters cover what happened. Dustin Gouker's Event Horizon is the most comprehensive — 236 editions in 2025, covering every regulatory filing, partnership announcement, and volume record. It is the wire service for prediction markets. But it does not have a thesis. It does not analyze specific market dynamics. It does not write from the perspective of someone who trades these markets. It tells you what happened without telling you what it means.
Venture capital content explains why you should build. a16z has published the most sophisticated analysis, citing Hayek and Tetlock and mechanism design. But they write for builders and investors, not participants. Their framework is Hayekian — markets aggregate information efficiently — and it conspicuously never mentions Soros or reflexivity. They are structurally bullish on crypto and structurally incapable of acknowledging when blockchain technology is irrelevant to the product. Their audience is the supply side. The demand side — people who actually trade, watch, and use these markets — is unaddressed.
Crypto-native educational content teaches the mechanics. DeFi Education and similar publications explain how prediction market contracts work, which platforms to use, and where the alpha might be. This is useful and well-executed. But it treats prediction markets as a crypto subcategory rather than a new financial primitive. The audience is people who already hold USDC. The 91 percent of Kalshi volume flowing through Robinhood comes from a different population entirely.
What does not exist: analysis of prediction markets as reflexive systems. Writing that treats the feedback loop between prices and outcomes as the central feature, not an edge case. Original quantitative work — cross-platform price comparisons, mispricing tracking, accuracy backtests, market microstructure analysis — using the data these markets produce in real time. A practitioner's perspective that is intellectually serious without being academic. Content that bridges the insider community and the broader audience that is about to discover these markets exist.
This is what Reflexivity is for.
This newsletter will cover prediction markets through the reflexivity lens. Every piece will ask: how does this market change what it measures? We will lead with data — specific contracts, specific prices, specific mispricings, specific movements — and use that data to develop arguments about where these markets are headed and what they mean.
We will write about the structural mispricings that persist across platforms because Polymarket, Kalshi, and traditional sportsbooks do not share infrastructure or user populations. We will write about the regulatory feedback loop that is currently the single largest risk to the industry. We will write about market integrity — insider trading on Super Bowl ad contracts is not a hypothetical; the Spotify contract spiked from $0.35 to $0.69 two weeks before the game on no public information. We will write about what prediction market prices mean when they appear on CNN, and what it means for the events they describe.
We will not pretend to have answers to every question. Some problems in this space are genuinely hard and genuinely unsolved: how to build credibility infrastructure when positions are anonymous, how to measure when a prediction market price changes the outcome it forecasts, how to build media that moves at the speed of markets. We will write about these problems honestly.
The prediction market infrastructure layer does not exist yet. We are building it — through The Odds Desk (the cross-market terminal), TheMO (the media layer), and Reality Managers (the community in San Francisco). This newsletter is the connective tissue. It is where the thesis meets the market.