The Deepest Problem in Prediction Markets Is Not Liquidity or Regulation It Is Alignment and Nobody Is Solving It

Every market that has ever existed eventually arrives at the same foundational question. Who is protecting whom, and from what, and at whose expense. Prediction markets have arrived at that question faster than most, and the answer the industry is currently producing is not encouraging.

The document I've been studying lays out a framework for understanding prediction markets that cuts through the promotional language around democratized forecasting and crowd wisdom and gets to the structural reality underneath. That reality involves three parties whose incentives are frequently misaligned, a power imbalance that systematically favors operators over participants, a regulatory environment that has evolved from protection into extraction, and a technological alternative that promises to solve the alignment problem but has not yet proven it can do so at scale.

Understanding this framework is not pessimism about prediction markets. It is the precondition for building ones that actually work.

The Three Parties and Why Their Alignment Determines Everything

Prediction markets involve three constituencies whose cooperation is necessary and whose conflict is structural. Risk takers are the participants who trade on event outcomes, whether for profit, hedging, or recreation. Middlemen are the operators who run the markets, set prices, manage liquidity, and handle settlement. Enforcers are the regulators and arbiters who determine who can operate in the space and how they must conduct their business.

In the minimalist sense, prediction markets do not require all three. Two parties who agree on terms can settle a wager independently on any prediction without touching any third-party system. But individual settlement at scale is impractical. Trade needs to happen quickly, reliably, and in sufficient volume to produce meaningful price discovery. That requires infrastructure managed by a middleman, and the middleman's involvement introduces the conflict of interest that sits at the center of every problem prediction markets currently face.

The risk taker's position in this structure is weaker than most participants recognize. Middlemen control bid and ask prices, decide which markets to offer, choose who they want to do business with, and set initial prices on outcomes. Bettors, by contrast, take prices as given. They negotiate from a position of structural disadvantage that is not incidental to the market design but is embedded in it. Understanding this power imbalance is the first step toward evaluating whether any specific prediction market is actually serving its participants or extracting from them.

Soft Books and Sharp Books: The Distinction That Explains Most of What Goes Wrong

The power imbalance between operators and participants gives rise to two fundamentally different operator strategies, and understanding the difference explains why most prediction market participants consistently underperform their theoretical edge.

Soft book operators extract money from their customer base through a set of tactics that are individually deniable but collectively constitute a systematic predation strategy. They bet against their own customers while limiting winners' upside to subsidize their bottom line. They induce customers to place additional wagers at heavily unfavorable odds through promotional structures designed to look like generosity. They discriminate against consistently profitable traders by reducing limits or closing accounts entirely. They exploit uninformed retail flow to subsidize a small pool of sophisticated participants who understand the game well enough to make money despite the structural disadvantage.

The soft book model is economically rational from the operator's perspective in any environment where uninformed retail participants continue entering the market faster than informed participants can exit it. As long as that condition holds, there is no competitive pressure to improve terms. The operator can maintain predatory practices indefinitely while marketing the product as fair, transparent, and participant-aligned.

Sharp book operators pursue the opposite strategy. They set prices that are more informative than their own intuitive estimates, place significant wagers themselves to spread risk broadly, constrain their margins to improve market efficiency, and actively work to aggregate the information held by their customer base into accurate probability estimates. The distinction between a truly sharp book and a proper market maker becomes difficult to draw when the sharp book model is operating at its best.

The problem is that the sharp book model is almost non-existent in practice. Left to their own devices, operators have inherent incentives to pay lip service to efficiency while pursuing soft book extraction. The competitive pressure that would force operators toward the sharp book model requires either the presence of other sharp books creating genuine market pressure, consumer protection mechanisms that penalize predatory behavior, or participants sophisticated enough to identify and avoid exploitative operators. None of these conditions reliably holds in current prediction markets.

The Market Maker Distinction That Changes Everything

One of the most important conceptual clarifications the framework makes is the distinction between a bookmaker and a market maker, because conflating them obscures where the structural problem actually lives.

A bookmaker sets the initial prices on outcomes, chooses who to do business with, and can change the prices offered to different participants based on who they are and what they do. The bookmaker's profit comes from the margin between the prices they set and the true probabilities, and from the informational advantage they hold over most of their customers.

A market maker is a fundamentally different animal. A market maker provides liquidity by standing ready to buy and sell contracts, earning a thin spread between bid and ask prices without the line-setting privileges of a bookmaker. The market maker cannot independently change the numerator and denominator of the price fraction or discriminate between participants based on their profitability history. The only reward is the spread, and competitive pressure on that spread from other market makers drives it toward the minimum consistent with the risk of holding inventory.

Platforms like Polymarket have made the market maker model accessible to individual participants through liquidity provider reward programs that subsidize the spread with explicit USDC incentives. This is a structural innovation because it aligns the operator's interest with market depth rather than with extracting margin from individual participants. When the operator profits from volume and liquidity rather than from individual trades, the incentive structure shifts toward creating conditions where participants want to trade rather than conditions where participants are exploited until they stop.

The LP reward model is not perfect and introduces its own complications, including the adverse selection problem where sophisticated market makers cream-skim the most profitable order flow while leaving liquidity providers with the worst of the remaining flow. But it represents a genuine structural improvement over the bookmaker model because it separates the operator's revenue from individual participant outcomes in a way that reduces the most obvious conflict of interest.

Where Middlemen Have Failed

The framework's treatment of middleman failures is worth examining carefully because the specific examples reveal a pattern that goes beyond individual misconduct into structural failure.

The FTX collapse is the most dramatic example, where user balances were converted into personal assets by the operator. The harm here was catastrophic and unambiguous. But the framework also points to subtler failures at regulated entities like FanDuel and BetMGM, where technical glitches disproportionately reduced the value of winning bets. The regulatory environment around sportsbooks has consistently failed to hold operators accountable for these outcomes in ways that adequately compensate affected participants.

The common thread across these failures is that the middleman's primary duty, removing frictions like solvency and settlement risk so that participants can trade with confidence, is not backed by enforcement mechanisms strong enough to make failure costly. The economic structure of prediction market operation means that operators can extract rents from participants in exchange for a promise of fair settlement, and then fail to deliver on that promise without facing consequences proportional to the harm caused.

This is not an argument that all operators are bad actors. It is an argument that the structure of the market makes bad behavior less costly than good behavior in many circumstances, and that without robust enforcement, this structural reality will reliably produce outcomes where some operators exploit that imbalance.

The Regulatory Failure and Why It Is Worse Than Simple Incompetence

The most damning part of the framework is its analysis of regulatory failure, because it argues that the enforcement mechanism that should constrain operator predation has itself become a source of extraction.

Until recently, US gaming regulators operated with a genuine orientation toward participant protection and market integrity. The framework argues this changed in 2018, when regulators began judging success by how fast they could drive market growth and started extracting rents based on revenue growth rates rather than participant welfare. The result is an enforcement environment where regulatory decisions favor the maximization of the industry's revenue rather than the protection of its participants.

This is a specific and serious accusation. It means that the natural check on operator predation, the threat of regulatory action for misconduct, has been compromised by regulators whose revenue model aligns with operator growth rather than participant protection. The circular structure of this failure, where the enforcement mechanism designed to constrain operators has been captured by operator incentives, is what the framework describes as unprecedented in the history of finance.

The tension between state gaming regulators and the CFTC adds another layer of complexity. State-level gaming authorities and the federal CFTC both claim jurisdiction over prediction market activity, with the CFTC treating event contracts as derivatives subject to federal oversight and state gaming boards treating the same activity as gambling subject to state regulation. This jurisdictional conflict has produced regulatory uncertainty that benefits operators in some ways, by creating gaps in enforcement, and harms the ecosystem overall by preventing the development of clear, consistent standards.

The Blockchain Alternative and Its Honest Limitations

The framework defends decentralized blockchain-based prediction markets as a potential solution to the alignment problem, and the logic is worth examining carefully rather than accepting or rejecting wholesale.

The core argument for decentralized prediction markets is that they can replace discretionary enforcement authority with decentralized consensus protocols that are not subject to the capture dynamics that have compromised regulatory enforcement. If the rules governing a prediction market are encoded in smart contracts that execute automatically based on verified outcomes, the operator cannot selectively apply those rules in ways that favor their own interests. The contract either executes as written or it doesn't, and no regulator with misaligned incentives can intervene in the execution.

Polymarket's architecture is the primary example. Its use of the UMA Optimistic Oracle for outcome resolution creates a mechanism where any user with sufficient capital can dispute a proposed resolution, and a community of voters makes the final determination. The system is not perfectly neutral, large holders of the protocol's governance token have disproportionate influence over disputed resolutions, but it is considerably more transparent and auditable than discretionary regulatory decisions made behind closed doors.

The honest limitation of this alternative is that decentralization does not fully solve the problem. It shifts the alignment problem from human intermediaries to protocol design, and protocol design has its own failure modes. Code bugs, governance attacks, oracle manipulation, and the concentration of governance power among large token holders all represent ways in which decentralized systems can fail to serve participant interests even without the traditional forms of operator misconduct.

Polymarket's own history includes examples of contested market resolutions where the community voted in ways that many participants considered incorrect, and where the UMA dispute mechanism produced outcomes that were technically consistent with the protocol's rules but arguably inconsistent with the spirit of the market. The decentralized alternative is more resistant to some forms of predation but is not immune to all of them.

What a Functioning Ecosystem Actually Requires

The framework ends with a clear statement of what the ecosystem requires to function well, and the statement is specific enough to be useful as an evaluation framework for anyone deciding where to trade or how to think about prediction market development.

Risk takers need access to well-defined event contracts with clear resolution criteria and enforcement mechanisms that reflect the contract's true terms rather than the operator's interpretation of those terms. The quality of resolution is the foundational trust requirement. Participants can tolerate unfavorable odds if they trust that winning bets will be paid. They cannot tolerate favorable odds if they suspect that winning bets will be contested, limited, or voided.

Middlemen need frictionless access to financial flows and regulatory environments that permit them to operate without onerous requirements that drive activity offshore or underground. This is a legitimate requirement, and excessive regulation of prediction market operators can harm participants by limiting the liquidity and product variety available to them. The goal is appropriate regulation, not maximum regulation.

Enforcers need a revenue model that aligns with participant protection rather than with market growth for its own sake. This is the requirement that the current regulatory environment most clearly fails to meet, and fixing it requires structural changes to how regulatory agencies are funded and evaluated rather than simply replacing the individuals who run them.

The critical alignment condition is this: operators' incentives must be calibrated to uphold participant interests rather than subvert them. This calibration can come from competitive pressure, from effective regulation, from participant sophistication, or from technological structures like smart contracts that make cheating impossible rather than merely costly. In practice, producing a well-functioning prediction market ecosystem likely requires all four in combination, because any single mechanism for alignment has failure modes that the others can compensate for.

The Honest Assessment

Prediction markets are genuinely useful instruments when they function as they should. The aggregation of dispersed private information into publicly observable prices, the ability to create markets for events that traditional financial instruments cannot capture, and the potential for participants to profit from genuine information advantages all represent real economic value. The accuracy of prediction market prices on well-covered events, consistently outperforming polling and expert forecasting on measurable questions, demonstrates that the mechanism works when the conditions are right.

The problem is that the conditions are not reliably right. The power imbalance between operators and participants, the dominance of the soft book model over the sharp book alternative, the failure of regulatory enforcement to protect participants consistently, and the governance challenges that remain unresolved even in decentralized alternatives mean that the gap between what prediction markets can be and what they currently are is substantial.

Closing that gap requires being honest about the structural problems rather than marketing around them. The three parties that make or break a prediction market need to operate in alignment with each other, and that alignment does not emerge automatically from good intentions or from the presence of technology. It requires deliberate design, competitive pressure, and enforcement mechanisms that are actually aligned with the participants they are supposed to protect.

The prediction market ecosystem's future depends on whether the industry can solve this alignment problem before regulatory overreach or operator predation destroys the trust that makes these markets useful in the first place. The framework for thinking about that problem is now clear. What remains is the harder work of actually implementing solutions that hold up under the pressure of real money and real conflicts of interest.