What Five New Academic Papers Say About Prediction Markets
A closer look at 'Are Betting Markets Better than Polling,' 'Exchanges are Using Federal Derivatives Law to Provide Gambling Products to Retail Traders' and more
Prediction markets are having a moment in academia.
There have been at least five academic papers published since the start of May that are about prediction markets and exchanges or tackle issues around them. And those are on top of the Journal on Prediction Markets, which is working on a US-based issue now.
Below is an overview of all these recent papers, plus links if you want to read them. (All passages are direct citations of the paper). I’ve done my best to pick out the parts that might be interesting to this audience.
Gambling product to retail traders
Published: June
Author: Ilya Beylin, Seton Hall
Abstract: A number of technological, legal, governance, cultural, industry and other changes have enabled and encouraged derivatives exchanges to design products attractive to retail traders. Combining approaches from mental health, finance and law, this work identifies the emergence of derivatives instruments that include features traditionally found in gambling products (e.g., short timelines with concrete resolutions, greater emphasis on maximum upside than expected value, small stakes, and importantly, consistent histories of losses among retail traders). The regulation of these derivative-exchange traded products under the Commodity Exchange Act (CEA) poses two related issues. First, derivatives exchanges exploit preemption under the CEA to offer gambling across the United States without regard to state law (including state law restrictions on gambling and state gambling taxes). This privileges derivatives exchanges over traditional gambling venues and disables state law protections. Second, having been formulated to govern markets serving institutional risk transfer functions, the CEA largely neglects the risk of exchanges offering unhealthy products for retail traders to over-consume (e.g., due to addiction and other mental health issues). This Article provides an analysis for identifying derivatives products that should be regulated as gambling instruments and cautiously proposes amendments to the CEA to address the risk from these products to retail traders.
Key passages:
The role of derivatives exchanges in product development, the structure of exchange-traded derivatives markets, and the absence of adequate consumer protection at the exchange level all point to a regulatory gap. Without regulatory changes, derivatives exchanges are likely to continue expanding offerings that enable retail traders to engage in gambling outside the strictures of state law.
Derivatives exchanges can and do innovate new product offerings. As for-profit entities, exchanges are predisposed to develop profitable products. Shareholders interested in financial returns from their investments will exert pressure on exchange boards to increase profitability.
It is important to keep in mind that the binary nature of event contracts is not only what generally makes them poor hedging instruments but also what makes them appealing to a retail audience. Retail and other small traders face challenges trading traditional derivatives products because of the associated margin obligations. For example, as the asset underlying a futures contract increases in price, the short position on that contract must post additional collateral within a couple of days. This creates liquidity risk as well as operational demands. In other words, the trader must have access to collateral (i.e., liquidity) to cover the increase in exposure and the trader must be able to respond to the request for collateral through posting it. For many small businesses, these are genuine challenges. This is perforce the case for retail traders. The design of event contracts avoids these challenges because – as noted above – event contracts are prepaid at the outset. When the bundle is purchased in the primary market transaction, the purchaser prepays the amount that is payable under the contracts obviating the need for further collateral. On this basis, the CFTC exempted event contracts from a variety of requirements related to clearing and the support of financial obligations.
Exchanges have designed event contracts to appeal to retail traders through more than dispensing with ongoing collateralization requirements. Event contracts are sold and traded in miniscule sizes. For example, leading U.S. event contract platform, Kalshi Exchange, typically denominates contracts in subdollar sizes. A number of contracts on Kalshi show total open interest (i.e., positions across all market participants) of less than six figures. Volume information from competing Polymarket, which operates an offshore exchange for event contracts is consistent, showing that traders’ positions generally do not exceed $100. This information on position sizes indicates that event contracts are not being used to transfer risk. The positions are too small to manage appreciable risk and cannot justify investment in risk management analysis. Kalshi explaining that it develops its products based on newspaper headlines supports that these contracts are being designed to appeal as entertainment rather than risk management devices.
Event contracts are designed to be used for gambling. As discussed above, they enable financially trivial positions consistent with an expenditure on a pack of gum. Many of them track attention-grabbing headlines and appeal to amateur interests in prognostication. Although some event contracts stay outstanding for long periods of time, many others settle swiftly. Some contracts are absurdly short term, such as contracts that Kalshi offers on whether the temperature within a given city will exceed a specified threshold “today” (another contract of questionable if any hedging utility). All event contracts have defined termination dates. That event contracts are being used for gambling follows not only from conscious design decisions on the part of exchanges that offer them, but also from financial theory. A successful market requires liquidity from uninformed traders.
One may argue that these contracts serve to generate predictive information through prices, and this information itself is a public benefit. That is a fair argument, but it raises several questions. First, whether the social value of that information is sufficient to justify gambling outside of state gambling law. After all, wagers on sports generate information (e.g., which team is likelier to win) and yet that has not in the past been seen as sufficient to move sports wagers outside of state gambling regulation.
Critically, exchange regulation under the CEA does not impose customer protection requirements that meaningfully protect derivatives traders from financially unhealthy product offerings. A derivatives exchange that provides access to retail traders must satisfy certain “Core Principles.” The Core Principles focus on the integrity of market function (e.g., preventing manipulation, front running, and other practices that lead to some traders receiving inferior prices) and financial support for executed transactions (e.g., appropriate segregation of customer collateral). These requirements neglect the protection of customers from products that may be appealing but counterproductive. The gap makes sense historically, because markets subject to the CEA traditionally sought to provide risk management tools for businesses.
This Article does not take aim at the old problem of some retail traders misusing traditional financial instruments. Rather, this Article points to a distinct problem; namely, derivatives exchanges exploiting their access to retail traders and exemption from state gambling regulation to develop products that increasingly resemble traditional gambling instruments. As repeat commercial players facing decreasing costs in marketing gambling instruments to the retail public, exchanges have motive and opportunity to develop products that cater to gambling predilections rather than advancing investment or risk management goals. Indeed, in event contracts, exchanges provide the retail market with quintessential wagering instruments that cannot be used for investment and in many cases do not serve risk management goals.
The CEA has unintentionally become a backdoor through which wagering instruments can be made available to Americans outside of the safeguards of state gambling regulation. This is concerning because modern technology enables exchanges to reach retail consumers at all hours and in all places through smart phones and other computing devices. Exchanges offer gambling products from sunup past sundown in living rooms, bedrooms and all other rooms across the US. The CFTC has not acted to curb the availability of these products, and neither has Congress. These exchanges pose a serious risk to those with weak impulse control, and the CEA provides the exchanges with a competitive advantage over bookies and other traditional gambling venues operating under state law.
Betting on Everything
Title: Betting on Everything
Published: May
Author: Karl M.F. Lockhart, DePaul
Abstract: Investing and gambling occupy two different regulatory worlds, as different as New York and Las Vegas. But how and where to draw the line between these activities has flummoxed lawmakers for centuries. This paper seeks to add a new chapter to that debate in light of three recent developments: sports gambling’s legalization and rapid growth, the increased prominence of retail investors (and the risky investment products they are purchasing), and the rise and widespread acceptance of event contracts—including those related to elections that were first available during the 2024 election cycle.
Part I of this paper reviews the Commodity Futures Trading Commission’s rationales for recently attempting to ban election and sports-related event contracts as gambling. Part II applies those same rationales—a lack of hedging and pricing utility, opaque information sources, market manipulation concerns, and participant harm—to derivatives and securities investment products as well as gambling activities. It finds that those criteria fail to distinguish many recently popularized investment products from some gambling products, in particular sports gambling. Part III then searches for a workable principle to distinguish investing from gambling by aggregating distinctions proposed by jurists and scholars. Of these five potential heuristics—information-gathering potential, risk profile, time horizon, the existence of an underlying asset, and extant versus created risk—only the last principle sufficiently separates the two activities. Part IV concludes by framing this heuristic’s normative implications, describing an improved understanding of today’s market structures, and sketching two preliminary prescriptions for rethinking gambling and investing regulation.
Key passages:
Ultimately then, we are all bookies. None of us is doing God’s work. Whether you are a jersey-wearing NFL fan betting on the Super Bowl, or a suit (or vest-wearing) private equity associate making a recommendation on a company to buy, the core skill is the same.
What this also means is that all markets are becoming more and more similar. They all involve a spectrum of participants, but these participants can be broadly grouped into two categories: those with sophisticated tools and models, and those without. Those with the best data and algorithms will make better predictions, and therefore more money, than those with weaker tools. But for the most part, everyone with models and machines will beat everyone without these tools, every time, in any market, anywhere in the world.
For example, a retail markets participant might open three apps on his phone: Robinhood to buy NVIDIA options, Kalshi to purchase an event contract on who will be the next Pope, and FanDuel to wager on the NBA Finals. What he probably does not know is that the same quantitative hedge fund, Susquehanna, could be using its vastly superior data and algorithms to calculate that all three bets he’s made are almost certainly incorrect, and take the other side of the trade in all three markets.
It is this aspect of the markets that makes them feel like a casino. Across all three markets, the average retail market participant will have a negative expected value outcome. Without access to reams of data, sophisticated algorithms, and the know-how to put the two together, we are gamblers, wagering against the house—hedge funds, banks, and corporations with immense resources. Dumb money versus smart money; we are bound to lose by a country mile. Faced with this farce, the rational response for retail markets participants is to treat everything like gambling.
If everyone is betting on everything, the casino needs new rules. If markets really are converging and structured as just described, then this trend has two significant implications for regulation: one deregulatory, and one calling for increased regulation.
For the most part, regulators in the United States have sought to protect retail investors by making sure that the markets are working fairly, not by limiting investors’ participation. But if they are frequently trading against parties with tools and capacities that far outstrip their own—and are therefore likely to lose—then perhaps regulators should set caps on how much retail investors can invest in certain types of financial products. For our own good, the law might place limits on how much we can wager on all the things we bet on: individual stocks, esoteric derivatives, who will win the Presidency or the Super Bowl. Accepting this implication means calling for stronger regulation.
Care for others, and for ourselves, sometimes means imposing limits. The real world involves real risk, risk that can ruin us. Life is not a game. The odds are not known. In that, there is uncertainty. But that uncertainty is also freedom —and the beauty of possibility.
Betting markets vs. polls
Title: Are Betting Markets Better than Polling in Predicting Political Elections?
Published: June
Authors: Laurie E. Cutting , Sarah S. Hughes-Berheim , Paul M. Johnson, Hiba Baroud, and Brett Goldstein, Vanderbilt
Abstract: Political elections are one of the most significant aspects of what constitutes the fabric of the United States. In recent history, typical polling estimates have largely lacked precision in predicting election outcomes, which has not only caused uncertainty for American voters, but has also impacted campaign strategies, spending, and fundraising efforts. One intriguing aspect of traditional polling is the types of questions that are asked – the questions largely focus on asking individuals who they intend to vote for. However, they don’t always probe who voters think will win – regardless of who they want to win. In contrast, online betting markets allow individuals to wager money on who they expect to win, which may capture who individuals think will win in an especially salient manner. The current study used both descriptive and predictive analytics to determine whether data from Polymarket, the world’s largest online betting market, provided insights that differed from traditional presidential polling. Overall, findings suggest that Polymarket was superior to polling in predicting the outcome of the 2024 presidential election, particularly in swing states. Results are in alignment with research on “Wisdom of Crowds” theory, which suggests a large group of people are often accurate in predicting outcomes, even if they are not necessarily experts or closely aligned with the issue at hand. Overall, our results suggest that betting markets, such as Polymarket, could be employed to predict presidential elections and/or other real-world events. However, future investigations are needed to fully unpack and understand the current study’s intriguing results, including alignment with Wisdom of Crowds theory and portability to other events.
Key passages:
Forecasts from the Polymarket data appear to be less precise than the polling data. The 95% predictive interval bands from the Polymarket model predictions fan out through time while those from the polling data remain consistent. This finding is due to the underlying dynamics of the two data sources that the BSTS models captured. The Polymarket data exhibit more of a random-walk like time series characteristic of financial markets (i.e., has a higher degree of auto-regression).
However, even with the increased uncertainty due to the random-walk-like structure of the Polymarket data, overall predictions are more accurate than those of the polling data. The mean posterior prediction across time mostly favors Trump, versus that of the polling data always favoring Harris. Moreover, most posterior predictions from the Polymarket data contain the correct prediction …. Conversely, almost none of the predictions from the polling data favor Trump (i.e., almost all polling uncertainty prediction intervals fall below the 50% probability line). As election day draws nearer, the forecasts from the Polymarket models become more precise (i.e., they extrapolate the random-walk over a shorter duration). By mid-October, almost all predictions from the Polymarket data have Trump winning while at the same time, almost all predictions from the polling data favor Harris.
Lastly, the Polymarket forecasts respond much more dynamically to events leading up to the election than those from the polling data. After the first assassination attempt in Pennsylvania, Trump’s odds of winning are intuitively predicted to be higher in the Polymarket forecasts. However, those of the polling data remain unchanged. Analogous trends occur when Harris becomes the nominee and performs well during the second Presidential debate. Here, we see the Polymarket predictions reflect a sentiment shift toward Harris, but predictions for the polling data remain largely unchanged.
…It is not only unclear how many individuals were betting on Polymarket, but also the demographic characteristics of those individuals is unknown. This issue is especially relevant regarding the singular individual who was confirmed to be placing large bets across multiple accounts (Osipovich, 2024). Beyond this specific limitation, Polymarket was also not supposed to be open to bettors within the United States (CFTC, 2022). Thus, the sample should have been very unique from that of the polling data, which was done with American citizens; however, sources indicate Americans were capitalizing on the anonymity of the website and betting via VPNs (Beyoud and Natarajan, 2024; Sundar, 2024). If true, and a large sample of Americans were participating in Polymarket, the representativeness of the sample is likely still limited due to the fact that the platform is based in cryptocurrency, since only about 17-20% of Americans have cryptocurrency (Matos, 2025; Murphy, 2025).
The economics of Kalshi
Title: Makers and Takers: The Economics of the Kalshi Prediction Market
Published: July
Authors: Constantin Bürgi, Wanying Deng, Karl Whelan; University College Dublin
Abstract: Since 2021, Kalshi has operated as the only licensed prediction market in the United States. We show that Kalshi’s contract prices are relatively accurate predictors of outcomes and their accuracy increases as markets come closer to closing. However, the prices also display a systematic favorite-longshot bias. Contracts with low prices win less than required for them to break even on average, while the opposite applies to contracts with high prices. We explain these results with a theoretical framework in which there are two types of participants: Makers who are relatively well-informed traders who make offers seeking a positive return but who may be too optimistic about their chances of winning and Takers who either accept or reject these offers. The framework predicts different patterns of favorite-longshot bias for Makers and Takers. We use data from over 300,000 Kalshi contracts to show these predictions are supported by the evidence.
Key passages:
In particular, investors who buy contracts costing less than 10c lose over 60 percent of their money. In contrast, there is statistically significant evidence that contracts with prices above 50c earn a small positive rate of return. These findings hold across a wide range of different categories of Kalshi’s markets, though there is some evidence that the bias in prices is getting smaller over time. Prior to fees being charged, the average rate of return on Kalshi contracts is minus 20%. Even though the contracts involve people swapping money so that the total return before fees on all money invested is zero by definition, the average losses incurred by those who buy cheap contracts are larger as a percentage of their investment than the average profit rate on more expensive contracts. Factoring in Kalshi’s fees, the average return across all contracts is minus 22%.
There is some evidence that Makers who buy contracts priced above 50c earn statistically significant positive returns but these returns are small relative to the huge loss rates for Takers who buy cheap contracts. On average, Makers who buy contracts costing 50c and over make a 2.6% rate of return before commission and a 1.9% rate of return including commission.
The design of Kalshi’s platform is likely to particularly encourage those who are highly overoptimistic to make bad financial decisions. A review of previous research on prediction markets would tell you that their prices tend to be relatively accurate summaries of the probability of winning. Someone who believes this would likely conclude that they would lose money on average once you factor in Kalshi’s fee. And indeed, on average this is correct. Kalshi earns money from commissions and its participants on net lose money. This illustrates a paradox of commercial prediction markets previously discussed by Whelan (2023): Those with the most accurate views are likely not to participate.
Sports betting as investing
The Sports-Betting Market: A Road to Sports Betting as Viable Investing
Published: May
Author: Tyler Gottlieb, Vanderbilt
Abstract: Since its legalization, sports betting has experienced rapid growth, both in terms of economic output and expansion into more states. The current system of sports betting in the United States requires individuals to place bets using a sportsbook. The sportsbook sets the lines on every bet. If an individual wins their bet, the sportsbook pays them money according to the listed odds. If an individual loses their bet, the sportsbook keeps the amount staked. Sportsbooks set lines in a way to ensure that on average, they make money on every bet. This system has been widely accepted, often based on the common understanding that “the house always wins.” When the house wins, the individual loses. Sports betting in its current state is riddled with pricing inefficiencies that can be best remedied by treating sports betting as a market in which individuals can buy and sell bets. By reimagining the way sports betting is regulated, a system can be created where individuals can place bets at fair prices and be rewarded with gains based on superior ability. By thinking of sports betting like a capital market, it becomes clear how flawed the current system is, and we can draw inspiration from other capital markets to create a vision of an efficient sports-betting marketplace that protects investors instead of exploiting them.
Key passages:
A market-based structure in this context means introducing short sellers to the market. This idea is similar to the structure of betting exchanges, in which bettors can both buy and sell bets. By allowing for buy- and sell-side pressure, prices will move closer to their fair value. Instead of the vig ensuring that the seller, who until this point has exclusively been the sportsbook, always profits, premiums will be much smaller and more analogous to transaction fees.
Because a market-based system would not impose limits on winners, institutional investors can play an increased role. By requiring an underlying state law violation for the Wire Act and moving to a system where bettors can both buy and sell bets without limits for winning, institutional investors may be able to return to sports betting. Institutional investors make up for a large proportion of short sales, meaning they are key in realizing the price and liquidity benefits of introducing short sellers to the market.