a16z's key investment: Kalshi's weekly trading volume approaches $3 billion, shifting from a "prediction game" to financial infrastructure, as the market begins pricing in "uncertainty."

In the traditional financial system, “price” usually only belongs to assets.

Stocks, interest rates, commodities—they can be traded because there is a unified measurement method and a consensus pricing mechanism. In contrast, those variables that truly influence market volatility—policy directions, macroeconomic data, political events—remain in a more primitive state: discussed, forecasted, but rarely directly priced.

These types of variables have always existed, but lack standardized expression. The emergence of Kalshi fundamentally changes this. It doesn’t create new information but provides a tradable price system for “the event itself.”

In recent research conferences, a noteworthy data point is: the weekly trading volume of sports-related markets has approached $3 billion, but its proportion of total trading volume is declining. In other words, the most visible part is growing, but the underlying structure is changing.

Meanwhile, institutions including a16z are increasingly paying attention to this track. This isn’t because prediction markets are “becoming hotter,” but because they are beginning to exhibit infrastructural features. Prediction markets are shifting from marginal products to a foundational infrastructure for “pricing uncertainty.”

01 Wall Street’s Focus: From “Discussable” to “Pricable”

The operation of financial markets relies on a premise: there must be tradable benchmark prices.

• S&P 500 is the core anchor of the stock market

• Interest rate curves define the cost of capital

• Commodity futures provide forward expectations of supply and demand

But in many key decisions, the variables that truly influence outcomes are not within these assets, especially “event-based variables,” which lack standardized long-term pricing methods. For example:

2. Whether a policy is implemented

3. Whether inflation data exceeds expectations

4. Whether regulatory changes occur

These factors influence the market but cannot be directly traded. The past solution was to indirectly express them through “related assets” (e.g., hedging election risk with stock indices). The problem is that this approach implicitly involves two layers of risk assumptions:

The second layer is often more uncontrollable. The core value of prediction markets is to eliminate this structural bias: turn “the event itself” into a tradable object. When the market prices the probability of “a policy passing” at 40%, this number is no longer just an opinion but a variable that can be traded, hedged, and modeled.

02 The Misunderstood Starting Point: Why “Sports” is Not the Focus, Just an Entry

The earliest large-scale prediction markets emerged from sports and elections, which is a natural outcome:

• Clear event boundaries

• Discrete outcomes

• Low user participation barriers

These scenarios are naturally suitable for early market launches but also lead to a misconception: people treat “the most visible demand” as “the entire demand.” But data from Kalshi shows that the structure is reversing:

This highlights a key issue: High-traffic scenarios do not necessarily equate to high-value scenarios.

Sports markets are more like “cold-start mechanisms,” providing users and liquidity; but the variables with true financial attributes are those that institutions can use for hedging and pricing. Participants from Goldman Sachs and Tradeweb mentioned in conferences that macro events (like CPI, interest rate paths) are becoming the most promising categories for prediction markets.

These variables share a common feature: they are not assets themselves but determine asset prices.

03 The Path of Institutional Adoption: From “Reference Indicators” to “Trading Tools”

Despite increasing discussion, prediction markets are still in the early stages of institutionalization. According to Kalshi’s classification, the adoption path can be divided into three stages:

Currently, most institutions remain in the first two stages. A key constraint comes from the trading structure itself: Current prediction markets require 100% margin to establish positions.

For institutions relying on leverage and capital efficiency, this implies high opportunity costs. This is why Kalshi is working with the CFTC to promote the introduction of margin mechanisms. Once this constraint is lifted, the growth at the trading layer could undergo a structural shift.

04 From Asset Pricing to “Probability Pricing”: An Extension of the Financial System

Viewing prediction markets within the broader history of finance, they are not an isolated innovation but rather an extension of the pricing system.

• Traditional markets price: assets, cash flows, risk premiums.

• Prediction markets price: events, probabilities, expected paths.

The key difference is: the former is result-oriented, the latter is process-oriented. This shift brings an important change: information begins to be expressed in “prices” rather than remaining in analysis and narratives. For example, when the market assigns a 60% probability to a policy passing, this number can be embedded into quantitative models, used for hedging, or as a decision input. This approach is closer to how financial systems utilize data compared to traditional expert judgment or polls.

05 Cross-Over with Agents / AI: From “Prediction Tools” to “Decision Input Layer”

Another significance of prediction markets is their potential integration with AI systems. Currently, most agents face a common problem: they can generate conclusions but struggle to quantify uncertainty.

Prediction markets offer a different path:

2. Constrain forecasts with real capital

3. Aggregate information through market mechanisms

4. Express probabilities via prices

As agents begin to participate in financial decision-making, risk management, or strategy generation, these “probability prices” will become critical inputs.

06 The Endgame Is Not Complex: Becoming a “Default” Infrastructure

A recurring point in the conference was: When it becomes boring, it’s truly successful.

This is not a devaluation but a typical path for financial infrastructure:

• Options markets in the 1970s faced similar controversy.

• ETFs were initially seen as marginal tools.

But once they become standard components, they are no longer questioned. Prediction markets may be entering a similar phase: from academic experiments to tools for elections and sports, then to macro and institutional applications, ultimately becoming a “default” layer of pricing. When that happens, it will no longer be called “prediction markets” but simply part of the financial system.

07 When “Uncertainty” Is Incorporated into the Pricing System

Returning to the initial question, the core of this change is not about trading volume or user scale, but a more fundamental shift: uncertainty is beginning to be standardized and expressed.

When events can be priced and probabilities traded, the future is no longer just a subject of discussion but becomes a variable that can be involved in calculations and allocations. In this process, prediction markets are not just a new product but a layer of new financial language. Once this language is widely adopted, it will change not only trading methods but the entire decision-making framework.