Tarek Mansour believes prediction markets can become one of the internet’s most powerful forecasting tools. The Kalshi CEO now sits at the centre of a growing debate around artificial intelligence, automated trading, and real-time probability markets, while researchers already test AI agents on live event contracts using real money under real market conditions.
Prediction markets have attracted far more attention since Kalshi CEO Tarek Mansour started publicly framing them as information systems rather than gambling products. In several interviews during early 2026, Mansour argued that markets built around real financial exposure produce stronger forecasting signals than opinion polling or social media speculation.
Artificial intelligence is now pushing that argument even further.
Mansour’s broader point is that prediction markets become more powerful when they process information faster than traditional forecasting systems. AI strengthens that case because automated tools can scan news, polling data, earnings reports, sports updates, weather models, and social signals almost instantly. That does not mean AI will make prediction markets perfectly accurate, but it could change how quickly prices react to new information and how traders identify value before the rest of the market catches up.
Researchers already test autonomous AI agents on live prediction markets, while regulators struggle to decide whether these platforms belong inside financial regulation, betting law, or an entirely new category.
Tarek Mansour Wants Prediction Markets To Function Like Information Systems
Tarek Mansour spends a lot of time arguing that prediction markets should operate more like forecasting infrastructure than traditional betting products. The Kalshi co-founder repeatedly described markets as tools for “pricing the future,” particularly once traders react to real-world information using financial exposure instead of online opinion alone. That framing helped pull prediction markets into serious conversations around economics, politics, sport, inflation, and data modelling.
A growing number of firms build custom sports analytics platforms around real-time decision-making because modern data systems rely heavily on live information processing. Kalshi now operates inside that same environment. Event contracts react instantly to polling swings, injury reports, inflation figures, Federal Reserve commentary, and election news.
The platform handled $5.42 billion in taker volume during April 2026 alone, while institutional trading activity reportedly climbed 800% across six months.
Those figures matter because they show how quickly prediction markets are moving from niche consumer products into serious financial infrastructure. Kalshi said institutional trading volume rose 800% over six months, while annualized trading volume more than tripled to $178 billion, according to company statements around its latest funding round. That kind of growth makes AI-assisted trading more consequential, because automated systems are no longer experimenting inside thin, obscure markets. They are entering venues with real liquidity, institutional participation, and regulatory attention.
Artificial intelligence now gives traders the ability to process that information much faster than manual analysis ever could.
AI Agents Are Already Trading Real Money Markets
The theory stage already ended. Researchers are now testing autonomous AI systems directly on live prediction markets using real money. A recent paper examining AI agents trading autonomously on prediction markets followed several frontier models operating on prediction markets between January and March 2026, with $10,000 starting balances allocated to each system.
The study produced mixed results, which probably explains why prediction markets remain attractive to researchers. One model achieved a 71.4% settlement win rate on contracts, yet several systems still posted losses on Kalshi, ranging between negative 16.0% and negative 30.8%. Researchers also found that pricing efficiency created major problems for automated systems once liquidity tightened around major events.
Those findings line up with what prediction-market operators have argued publicly for months now: forecasting correctly is only part of the equation. Timing, pricing, contract value, and probability interpretation still separate profitable trading from bad trading.
That makes AI useful, but not automatically profitable. A model can read news faster than a human, summarise polling data instantly, or compare market prices across platforms, yet still overpay for a contract if the market has already priced in the same information. In prediction markets, speed only helps when it is paired with discipline around entry price, liquidity, fees, and settlement risk. Otherwise, automation can simply make bad trades faster.
Retail Traders Are Entering a Market Built Around Probabilities
Retail participation has accelerated quickly because prediction markets removed a lot of the complexity traditionally associated with derivatives trading.
Kalshi contracts usually trade between 1 cent and 99 cents, with the pricing reflecting implied probability around a specific outcome. A contract priced at 35 cents implies roughly a 35% probability of that event occurring. That structure makes prediction markets easier for newer traders to understand without requiring deep financial-market experience.
That simplicity can also be misleading. A low-priced contract may look cheap, but it can still be badly priced if the true probability is lower than the market suggests. Likewise, a contract that feels obvious may offer poor value if most traders have already moved the price. For retail users, prediction markets are easiest to understand when prices are treated as probabilities, not guarantees.
The onboarding process also became much more accessible during the last year. New-user incentives tied to event-contract trading now mirror the kind of onboarding systems already common across fintech and investing apps. For new users, the important part is not only the headline value of an offer, but also the funding requirement, eligibility rules, contract availability, and whether the platform fits the type of markets they actually want to trade. A Kalshi referral code: SBR currently gives new users $10 in bonus funds after completing $10 in trades, while referral rewards can reportedly reach $1,000.
Sports contracts accounted for roughly 89% of Kalshi revenue during 2025, though the platform also covers inflation, crypto pricing, weather events, elections, interest rates, and entertainment outcomes.
Prediction Markets Are Becoming a Regulatory Battleground
Regulators are paying much closer attention now that prediction markets process serious money and attract institutional trading activity. U.S. lawmakers recently questioned whether event contracts operate more like financial instruments than traditional betting products, especially once AI systems and automated trading models enter the picture.
The debate has become increasingly tense because prediction markets currently sit under federal derivatives oversight rather than state gambling frameworks.
That distinction is now the heart of the legal fight. Prediction-market operators argue that event contracts are federally regulated derivatives, while several state-level critics view sports and election markets as gambling products that should be controlled locally. The outcome matters because it could decide whether platforms like Kalshi can operate under one national regulatory structure or face a fragmented state-by-state model similar to sports betting.
A recent breakdown of prediction-market pricing strategies focused heavily on risk management, probability pricing, and trader behaviour instead of gambling mechanics. That distinction sits at the centre of the regulatory argument. Reuters recently reported that prediction-market operators flagged more than 400 suspicious trades during 2026, more than double the previous year. Minnesota also attempted to block prediction markets entirely before the CFTC challenged the move in court. Professional sports leagues are getting involved as well; both the NHL and MLB reportedly signed integrity-monitoring agreements connected to prediction-market activity during 2026.
That sports angle is especially sensitive because league data, athlete availability, injury information, and insider knowledge can move markets quickly. Integrity agreements suggest that leagues and regulators are not treating prediction markets as a small side product anymore. They are preparing for a world where event contracts sit close to the same information flows that already shape sports betting, fantasy sports, broadcast data, and live analytics.
Prediction markets fit naturally into the same digital behaviour patterns already visible across esports, streaming, and financial-content communities. People now consume live information constantly through phones, livestreams, Discord groups, social feeds, and second-screen platforms running alongside major events. That constant flow of information creates ideal conditions for real-time event trading.
Esports audience statistics show that Twitch accounted for 71% of esports streaming hours during 2026, while mobile devices now drive roughly 45% of esports content consumption. TikTok esports viewing also rose 39% year-on-year. Prediction markets benefit from that same environment because users already expect live odds, instant information, rapid reactions, and continuous updates during major events.
AI systems simply accelerate that cycle by processing information faster and reacting to price movement almost immediately.
AI May Change Trading Faster Than Regulators Expect
The two biggest prediction markets have grown into something much larger than a niche internet curiosity. These markets reportedly surpassed $150 billion in lifetime trading volume during 2026, while institutional firms continue entering the sector alongside retail traders experimenting with AI-assisted forecasting tools. Researchers are already testing autonomous trading agents on live markets using real money, which means the next stage of development has already started.
The larger question now revolves around market structure rather than technology. Regulators still debate whether prediction markets belong inside gambling law, financial regulation, or something entirely separate.
Artificial intelligence complicates that discussion further because automated systems can process probability, pricing, and information at a pace human traders cannot realistically match.
That is why Mansour’s information-market argument now carries more weight than it did when prediction markets were smaller. If these platforms become widely used forecasting tools, AI will not just help users trade faster. It may influence how markets absorb news, how prices move, how regulators define manipulation, and how ordinary users interpret probability itself. The technology is already here. The unresolved question is whether the market rules can evolve quickly enough to keep up.
