Prediction Markets: Truth Engine or Casino? A Scoring Framework

Plus a rant about gambling and a few hot takes

Disclosure: I run Kalshinomics.com, which may earn Kalshi referral fees. I may trade event contracts on Kalshi and securities on other platforms. Readers should consider this relationship when evaluating my analysis. For educational purposes only, not investment advice.

As prediction markets have grown in volume and funding, there’s a robust debate: are they a new “truth engine,” another step towards the casino-ification of everything, or structurally broken, as in Agustin Lebron’s Predicting Our Own Demise.

I’ve been around both the trading and forecasting / prediction space for a while, so I’ll share a nuanced opinion that makes no one happy.

And let’s get one important point out of the way. A lot of ink has been spilled on the question “Are prediction markets gambling or not?” Of course they’re gambling. I have a very wide definition - distinct from the legal one - which is “taking risk under uncertainty.” And I get why the people running these companies don’t use the term gambling. There’s a parallel set of legal challenges that rests on whether these are gaming or not. I’m going to mostly steer clear of the legal questions and focus on their place in the spectrum of risky decisions.

Slot machines, sports betting, sports “predicting,” even investing in index funds or (gasp) choosing a college I would define as forms of gambling. Hang with me for a second - these are not all close to the same thing! But in each case you’re risking something - often your money, sometimes your time, and there’s some uncertainty about the outcome. You can expend effort to reduce or better understand this uncertainty (researching stocks, understanding casino odds, learning card counting, visiting a college). But even then some uncertainty remains. The tools to understand risk and make decisions are the same across many of these questions, while the underlying activity, the expected value, and the societal costs vary dramatically.

I loved this framework from Matt Glassman that ranks how destructive different types of gambling are. He ranks types of gambling (using a more traditional definition based on games of chance) from least to most social cost, on one end a family poker tournament, on the other roulette on your phone. I am also concerned about the negative effects of loosened access to gambling-like products. Prediction markets can be on the better end of this spectrum, but it depends on how they’re designed and marketed.

Breaking down Prediction Market Contracts

Prediction markets are gambling, but some contracts are socially valuable tools for risk allocation and information, and others lean casino. Which contracts are worth having?

Some market prices are useful for understanding the world, others may be novel or fun for participants but carry no social value. If there’s no liquidity or interest in trading, the market won’t be efficient, therefore no value to the price signal.

Complex or poorly worded contract definitions make it difficult for market-makers to provide liquidity and confuse customers. Narrow contracts may only be of use to experts, not the general public. Not everything needs to be useful! I’ve spent many dollars and hours playing video games. But exchanges can choose how to design and market various types of contracts. Those decisions will naturally prioritize the economics of the business - they need to make a profit. But for long-run success as well as (hopefully) better regulatory treatment, designing high quality contracts that provide a public benefit should be encouraged.

Recently Kalshi announced partnerships with CNBC and CNN to highlight probabilities in news. Are we turning the news into a casino? I actually think this is one of the best use cases for prediction markets. Seeing a news story that matters politically or economically, then directly how it changes a relevant probability, is an excellent way to communicate news to the public. Most people will not (and likely should not) be trading on prediction markets. The price signal provides value to everyone else. This is where the ‘truth engine’ claim starts to carry some merit.

Here is the Dec Fed market on Kalshi today:

And in comparison here is the CME’s FedWatch tool:

So yes with this tool it backs out probabilities, but if you’re looking at prices here’s what you’d be trading:

Far less easy to understand than the prediction market! You don’t need to decode the futures contract conventions.

High quality markets can improve decisions, help allocate and hedge risk, and provide context. Low quality markets function more similarly to entertainment. This is not unique to prediction markets. ETFs overall have been incredible for the investing public in lower costs and tax-efficiency over mutual funds. But for every VOO, SPY, or VXUS ETF (low-cost, diversified ETFs) how many “triple-levered super dividend” products list and trade huge volumes? Capitalism produces a lot of everything, including junk.

Below I have worked out a highly opinionated framework for ranking quality of contract types on prediction markets. When I talk about how I’m pro prediction markets, my interest is on the more useful end of the spectrum. Different users will have different interests and value judgements but there are some principles that should guide us.

On Sports

A huge % of prediction market volume is currently related to sports. These exchanges are able to effectively offer sports betting nationally under the CFTC. This is currently being litigated in a variety of venues, but is responsible for a huge chunk of current valuations. There are fundamental differences between sportsbooks and exchanges that I believe favor exchanges, but that could be a separate essay.

I’m one of the few people who doesn’t watch much sports. Outside the Super Bowl I watch maybe one game a year. But the demand for wagering on sports is undeniable. Is the new availability of sports via prediction markets good or bad? On the positive side, people are going to continue betting on them, and I’m hopeful that competition between prediction markets and sportsbooks will drive transaction costs closer to zero. Right now it’s not clear that odds are significantly better. But medium term we might see sub-penny spreads with tiny fees on major events. If bettors wager the same amounts, this is a lower total cost for the same entertainment.

On the other hand, if users increase their wagers drastically we may end up with the same costs and higher variance. Or the increased availability might increase the number of problem gamblers. While the financial industry focuses more on disclosure requirements, the gaming industry has specific requirements for resources on problem gambling. Kalshi does have a "responsible risk management” page, but it could be more prominent.

My scoring framework

1. Wide public interest (25%)

Needs to matter to enough people for trades to happen.

Examples: Politics. Affects everyone within a country and foreign elections often matter too. Systems are designed so elections are often close.

Sports have a long history of wagering and a huge audience.

2. Social usefulness of probability (20%)

Does having a series of prices for this contract provide value to the public in any significant way?

3. Breadth of information surface (10%)

Can outsiders and non-experts add valuable information? Is that information dispersed widely across fields or a small subset of experts can price perfectly and no one else is close?

Example: Sports has opportunities for both statistical methods and finely-grained current information, subjective measurements of player performance, breaking news, creates a broad array of market participants

4. Resistance to manipulation (10%)

Could one self-interested actor or insider easily decide the market?

Are there negative externalities to the market existing in the first place? (assassination markets)

5. Clear resolution criteria (10%)

Does the title of the contract match the details in the fine print? Can a contract be written in a way that leaves no room for ambiguity in how it will resolve under different states of the world? Example: “Will AGI be achieved by 2030?” -> according to who? Many possible definitions leave the title fuzzy. Huge differences are possible across exchanges for the contract with the same name

6. Hedge utility (10%)

Is there interest in hedging this risk? Would groups with exposure to the risk have utility in having a market to lay off some of their risk? Example: Insurance is available for certain natural disasters, but probabilities are not publicly disseminated / tradeable.

7. Market uniqueness (5%)

Does another product already exist for this type of event? Example: fed funds futures already tell us about fed probabilities.

8. Right level of difficulty / randomness (5%)

Could you learn something about the subject by studying, performing experiments, interviewing experts? Or is the answer so difficult or so random that work is futile? Example: NFL coin toss bets are pure gambling on chance.

9. Useful path volatility (5%)

Are future events likely to change the probability of the outcome? A time series of market prices along with news headlines tells you about the relative importance of events in real time. Example: Probabilities around the Trump-Biden debate in (Jun 27th, 2024)

I picked several categories and subjectively ranked each out of 5 points, then adjusted by the category weight for a total score out of 100.

*A Mention Market is a bet on whether a public figure will use a specific word in a speech or debate.

Let’s break down a few:

Hurricane Landfall

Strengths: Societally important, can directly impact millions both physically and financially. Important to get odds on where it will hit - this is offset by the existing strength of weather reporting / other alerts.

So-so: Information surface - mostly technical meteorological data, requires specialized skills. Informed traders are a limited subset.

Weaknesses: Existing retail insurance products already exist, reinsurance is available for institutional players. But no way for retail to go short this risk outside prediction markets, or easily view probabilities.

In my framework gets a 79.5/100, but unfortunately, even though some markets do exist in this category, so far volumes have been very small.

Mention Markets

Strengths: Unique product, some people find them “fun.” (I am great at parties) Some interesting data analysis possible combining NLP on historic speeches with context of current events.

Weaknesses: In most cases lack of any hedge utility. Extremely vulnerable to manipulation, in a way where a clever manipulator would be nearly impossible to catch.

At the end of their Q3 earnings call Coinbase CEO Brian Armstrong deliberately rattled off a list of words traded on the mention market. This is a joke of a market, manipulated for sport.

Additionally (though I’m not a lawyer), there is a related CFTC rule: [§ 38.200 Core Principle 3] :

“The board of trade shall list on the contract market only contracts that are not readily susceptible to manipulation.”

These contracts should not be listed.

Other Comments:

Novelty Markets: The Zelenskyy suit market which generated controversy summer 2025 over the “what is a suit” debate. This category encompasses other silly markets.

IPO Markets: An important weakness is the difference of the information surface to insiders vs outsiders. Many insiders know 100% one way or the other, while for outsiders the surface is limited to rumors/leaks.

My Verdict

I doubt anyone is going to completely agree, and I welcome reasonable criticism. Like with most things, it’s not black or white, everyone has their own incentives, some people want entertainment, traders and businesses want to make money. Just like the stock market can both offer a casino-like experience but valuable signals on events and corporations, prediction markets offer a bit of both.

My hope is - competition will push the (currently high) transaction costs down, and while the general success of these platforms may continue to rest on sports, the other more useful markets will develop enough of a following that they achieve “the wisdom of crowds,” and improve the way news and statistics are communicated.