October 2024. Polymarket odds spike on "Israel strikes Iran before November." Sharp movement. Large volume. Concentrated in narrow time window.
Hours later: strikes happen.
Someone knew. Someone bet. Someone profited.
This isn't prediction. It's information extraction.
What Prediction Markets Are
Prediction markets let people bet on real-world outcomes. Binary events. Will X happen by date Y? Yes or no.
Polymarket: crypto-based, offshore, technically banned for US users but accessible via VPN. $3.7 billion in volume during 2024 election cycle.
Kalshi: CFTC-regulated, USD-based, legal in United States after court battle. Focuses on economic and political events.
PredictIt: academic platform, small volume, research-focused. $850 limit per market.
Metaculus: forecasting platform, no money involved, just reputation.
The mechanism is simple. People bet on outcomes. More money on "yes" moves odds toward yes. More money on "no" moves odds toward no. Odds represent aggregated expectation of all participants weighted by how much they're willing to risk.
Traditional markets work similarly. Stock prices aggregate expectations about company performance. Prediction markets aggregate expectations about specific events.
The difference: stocks have intrinsic value (cash flows, assets). Prediction market contracts resolve to $1 or $0 based purely on whether event occurs.
Why They're More Accurate Than Polls
Prediction markets have track record of outperforming traditional polling.
2024 US presidential election:
- Traditional polls: Harris 48%, Trump 48%. Toss-up. Within margin of error.
- Polymarket: Trump 60-65% consistently for weeks before election.
- Outcome: Trump won decisively. Electoral college margin significant.
Polymarket was right. Polls were wrong.
Brexit (2016):
- Polls showed Remain ahead
- Prediction markets showed Leave gaining probability
- Final odds: 70-75% Remain (still wrong, but closer than polls)
- Outcome: Leave won 52-48%
Why are markets more accurate?
Skin in the game. Wrong prediction costs money. Polls cost nothing to get wrong. Respondents have no incentive for accuracy. Social desirability bias (lying to pollster). Sampling errors. Non-response bias.
Prediction markets force honesty. Your forecast is your bet. Wrong bet loses money. People care more about money than about answering pollster's questions.
Information aggregation. Markets collect information from everyone willing to bet. Including people with privileged access. Polls only collect from people willing to answer phone calls.
Continuous updating. Markets adjust in real-time as new information arrives. Polls are snapshots, often days old by publication.
Revealed preferences. Markets show what people actually believe will happen. Polls show what people claim they believe or want to happen. These are different.
The accuracy comes from incentive alignment. Get it right, make money. Get it wrong, lose money. Simple.
But this creates problem.
The Insider Trading Problem
Israel-Iran strikes (October 2024). Polymarket odds on "Israel strikes Iran before November" suddenly jump from 35% to 68%. Volume spikes. Concentrated betting in short window.
Hours later: Israel conducts strikes on Iranian military facilities.
The odds moved before public knew strikes were happening. Someone with operational knowledge likely bet. Made significant profit.
This is insider trading. Same concept as stock market insider trading. Information not available to public. Used for profit. Except prediction markets aren't securities. Legal framework is unclear.
Other examples:
Fed interest rate decisions. Unusual betting patterns hours before FOMC announcements. Odds shift before Jerome Powell speaks. Someone knows what the decision will be.
Corporate announcements. Prediction markets on CEO resignations move before official press releases. Markets on earnings surprises shift before quarterly reports.
Political resignations. Odds on UK Prime Minister resignation spiked hours before Boris Johnson announced (2022). Similar patterns with other political transitions.
COVID policy changes. Markets on lockdown extensions, vaccine mandates, travel bans moved before official announcements during pandemic.
The pattern is consistent. Those with access to information bet on it. Odds shift. Public sees shifted odds. By then, information is priced in. Opportunity is gone.
This isn't occasional. It's structural. Prediction markets extract privileged information and make it visible. But visibility comes after insiders profit.
What This Reveals About Information
Prediction markets are extraction tools. They convert privileged information into public signal.
The process:
1. Someone gains information not available to public (operational knowledge, policy decision, insider access)
2. That person bets significant amount on outcome they know is likely
3. Their bet moves odds
4. Public sees odds movement
5. Public updates beliefs based on odds
6. By time public sees signal, profit opportunity is gone
Markets make hidden knowledge visible. But not accessible. You learn something will happen. You learn after insiders already positioned themselves.
This creates two tiers:
- Tier 1: Those with privileged access. Profit from certainty.
- Tier 2: Those who watch markets. Learn from odds shifts. React too late.
Information asymmetry gets monetized. The asymmetry doesn't disappear. Just becomes visible after it's already been exploited.
Traditional insider trading (securities) is illegal because it undermines market integrity. Those with inside information have unfair advantage. Markets only work if participants have roughly equal access to information.
Prediction markets operate in legal gray area. Not securities. Not regulated same way. Insider trading happens openly. Nobody stops it. Can't stop it. No requirement to disclose who's betting. No trading restrictions for insiders.
The markets work precisely because insiders trade. Their knowledge gets incorporated into prices. Makes markets accurate. But accuracy benefits watchers, not participants. By time you see odds shift, the information is already priced in.
The Regulatory Battle
Prediction markets exist in regulatory gray zone. Different jurisdictions, different rules.
Kalshi vs CFTC:
Commodity Futures Trading Commission initially banned election betting. Reasoning: elections are not economic events, betting on them is gaming, raises public interest concerns.
Kalshi sued. Argument: elections are economic events. Policy decisions affect markets. People should be able to hedge political risk. Same way they hedge currency risk or interest rate risk.
Federal court sided with Kalshi (September 2023, affirmed 2024). Elections are events people can have economic interest in. Prohibiting betting violated Administrative Procedure Act.
Now: legal to bet on US elections through regulated platforms like Kalshi in United States.
Polymarket's situation:
Offshore. Crypto-based. Technically not available to US users. CFTC settlement (2022): $1.4 million fine, agreed to block US users, implement geofencing.
But: accessible via VPN. Everyone knows this. $3.7 billion volume in 2024 election cycle suggests significant US participation. Enforcement is difficult. Crypto makes it harder to identify users.
The regulatory split: Kalshi operates legally, openly. Polymarket operates in gray zone, offshore, with plausible deniability about US users.
What's allowed:
Kalshi permits betting on:
- Elections (now)
- Federal Reserve decisions
- Economic indicators (jobs reports, GDP)
- Congressional votes
- Supreme Court rulings
What's banned:
Markets cannot exist on:
- Assassinations
- Terrorist attacks
- Deaths of specific individuals
- Anything deemed "against public interest"
The line between allowed and banned is arbitrary. Changes based on political pressure and public reaction.
The 2024 Election Case Study
2024 US presidential election demonstrated prediction market accuracy and raised questions about their influence.
Traditional polling landscape:
Most major polls showed race within margin of error. Harris 48-49%, Trump 48-49%. Swing states too close to call. Expert consensus: genuine toss-up. Either candidate could win.
Nate Silver's model: 50-50. The Economist model: slight Harris edge. Cook Political Report: toss-up. Every major forecaster said too close to call.
Polymarket odds:
Trump maintained 60-65% probability for weeks before election. Brief dip to 55% after debate performance. Recovered to 62% by election day.
Market priced in Trump victory consistently. While polls said toss-up, markets said Trump favored.
Outcome:
Trump won decisively. Electoral college margin significant. Swing states went Republican. Senate flipped. House maintained Republican control.
Polymarket was correct. Polls were wrong. Markets captured something polls missed.
Why the divergence?
Polls measure stated preferences. People tell pollsters what they're willing to say. Social desirability bias. Some Trump voters don't admit it to pollsters.
Markets measure revealed preferences. People bet money on what they actually believe will happen. Can't lie with money. Bet reflects true expectation.
Polls sample population. Markets sample people willing to bet. Different populations. Bettors include people with better information, stronger convictions, willingness to put money on beliefs.
But questions remain:
Was Polymarket accurate or influential? Did odds shape narrative? Media coverage referenced Polymarket odds constantly. "Markets say Trump winning." Does this create momentum? Self-fulfilling prophecy?
If markets show candidate at 65%, does media coverage shift? Do donors react? Do voters perceive momentum? Does perception become reality?
Impossible to separate prediction from influence.
The Manipulation Question
Théo. French trader. Bet over $30 million on Trump winning 2024 election across multiple Polymarket accounts.
Accounts: Fredi9999, Theo4, PrincessCaro, Michie. Four accounts, one person, $30+ million in Trump bets.
Claimed reasoning: "Superior polling analysis." Said mainstream polls underestimated Trump support. Used "neighbor polling" (asking people who their neighbors will vote for, reduces social desirability bias).
Outcome: Trump won. Théo made $79 million profit.
Questions:
Was this genuine forecast? Did Théo have analysis that predicted Trump victory? Possible. Some pollsters said Trump support was underestimated.
Was this privileged information? Did Théo have access to internal campaign data, donor information, ground game intelligence? Possible. No way to verify.
Was this manipulation? Did Théo bet large amounts to shift odds, influence media narrative, create perception of Trump momentum? Possible. Large bets moved odds significantly.
The problem: can't distinguish between these.
Same action (large bet) looks identical whether motivation is:
- Genuine forecast (I believe Trump will win based on analysis)
- Insider knowledge (I know Trump will win based on privileged information)
- Manipulation (I want Trump to win and I'm using market to shift narrative)
Large bettor moves odds. Moved odds influence media. Media influences perception. Perception influences behavior. Market stops being purely predictive. Becomes interventionist.
The manipulation mechanism:
1. Bet $30 million on Trump
2. Odds shift from 50% to 65%
3. Media reports "Markets say Trump favored"
4. Creates perception of momentum
5. Donors become more confident, give more
6. Volunteers become more motivated, work harder
7. Undecided voters see momentum, some shift toward Trump
8. Trump actually wins (partially because market signaled he would)
The bet becomes self-fulfilling. Not pure prediction. Active intervention disguised as forecast.
No way to prevent this. Can't ban large bets (reduces market liquidity, makes odds less accurate). Can't require disclosure of reasoning (unprovable, unenforceable). Can't distinguish good-faith forecast from manipulation attempt.
Geopolitical Intelligence Value
Prediction markets reveal information valuable to intelligence agencies and governments.
What markets show:
When military strikes will occur (Israel-Iran odds spike before strikes)
Whether regime change is imminent (odds on government collapse shift before coups)
If war is escalating (defense contractor bets signal insider knowledge)
When central banks will pivot (Fed rate decision odds move before announcements)
Which political figures will resign (odds shift hours before official announcements)
Intelligence agencies watch these markets:
Free signal aggregation. No need to pay sources. No need to infiltrate organizations. Just watch where money flows.
Markets capture insider trading by connected individuals. Someone in Israeli defense establishment bets on strikes before they happen. Someone in Fed system bets on rate decision before announcement. Someone in White House bets on policy change before release.
Their bets reveal what insiders expect. Cheaper than running spy networks. More reliable than single sources. Markets aggregate multiple insider signals.
But also creates counterintelligence problem:
Foreign adversaries can manipulate. Bet large amounts to create false signal. Make intelligence agencies think event is likely when it's not. Information warfare through prediction markets.
Example: Russia wants US to think they're withdrawing from Ukraine. Bets large sums on "Russia withdraws by December." Odds shift. US intelligence sees signal. Interprets as insider knowledge. Adjusts strategy based on false signal. Russia gains advantage.
Or reverse: Russia wants US to think they're escalating. Bets on "Russia uses tactical nuclear weapon." Odds shift. Media panics. Public pressure on US government to respond. Russia achieves intimidation without actual escalation.
Prediction markets become battlefield. Information warfare conducted through betting. Signal and noise become impossible to separate.
The Future Trajectory
Prediction markets are expanding. More platforms. More markets. More liquidity. More mainstream adoption.
Current trajectory:
Kalshi expanding market offerings. Now includes elections, Fed decisions, economic data, congressional votes, Supreme Court cases. Planning to add more event types.
Polymarket growing despite regulatory uncertainty. $3.7 billion volume in 2024 election. More than doubled from previous cycle.
Traditional finance entering space. Hedge funds using prediction market data. Asset managers monitoring odds. Integration with traditional forecasting.
Media citing markets more frequently. Odds become news. "Markets say X is likely" becomes headline. Feedback loop strengthens.
Two possible futures:
Future 1: Expansion and Integration
Prediction markets become primary forecasting tool. Replace polls for elections. Replace expert consensus for major events. Governments use for policy planning. Corporations use for strategic decisions.
Deeper liquidity makes manipulation harder. More participants, larger volume, harder for single actor to move odds significantly.
Better regulation creates legitimacy. Clear rules about what's allowed. Insider trading still happens but within framework. Markets accepted as information aggregation tools despite imperfections.
Integration with traditional finance. Prediction market derivatives. Options on prediction market outcomes. Full financialization of event forecasting.
Future 2: Restriction and Underground
Governments ban or heavily regulate prediction markets. Too much insider trading. Too much manipulation. National security concerns. Public policy concerns.
Markets deemed threats to democratic process. Betting on elections seen as corrupting. Forecasting geopolitical events seen as intelligence risk.
Heavy restrictions or complete bans. Markets go underground. Crypto-based, anonymous, untraceable. Offshore platforms. Decentralized protocols.
Accuracy remains but accessibility decreases. Only sophisticated users participate. Public loses access to information signal. Markets continue in shadows.
Which future occurs depends on government response. If markets stay useful and relatively clean, Future 1. If abuse becomes too obvious, Future 2.
What drives the decision:
Visible insider trading scandals. If someone obviously trades on classified information and profits millions, public backlash forces regulation.
Manipulation affecting elections. If evidence emerges that large bettors shifted election outcomes through market manipulation, governments ban.
Geopolitical incidents. If prediction markets give adversaries intelligence advantage or enable information warfare, national security apparatus demands shutdown.
Revenue and taxation. If governments can tax betting profits, incentive to allow markets. If revenue is significant, incentive to regulate rather than ban.
The trajectory isn't predetermined. Depends on how problems manifest and whether benefits outweigh costs in eyes of regulators.
Who Benefits
Not the general public.
Public sees odds shift. Learns something is likely to happen. But learns too late to act. Information already priced in. Opportunity gone.
Retail bettors lose money on average. Like casino, house has edge. Sophisticated bettors and insiders extract value from retail participants.
Insiders benefit most.
Those with privileged access profit directly. Convert knowledge advantage into money. Legal gray area protects them.
Israeli official knows strikes are coming. Bets large amount. Makes profit. No prosecution. No oversight. No consequences.
Fed employee knows rate decision. Bets accordingly. Profits from certainty. Untraceable if done carefully. Low risk, high reward.
Corporate executive knows resignation announcement. Bets on timing. Guaranteed profit. Who's checking prediction market trades?
Platforms benefit.
Polymarket charges approximately 2% fee on winning bets. More volume equals more revenue. $3.7 billion volume equals $74 million in fees.
Platforms incentivized to maximize volume. Volume increases with insider trading (creates opportunities for profit-seeking). Platforms don't prevent insider trading. They profit from it.
Kalshi charges fees on contracts. More trading, more revenue. Same incentive structure.
Intelligence agencies benefit.
Free aggregated signal about insider expectations. Better than paying sources. More reliable than single informants.
CIA, Mossad, MI6 likely monitor prediction markets. Track which events show unusual betting patterns. Investigate who's betting. Follow the money to information sources.
Markets do intelligence work for free. Reveal what connected people expect to happen.
Financial institutions benefit.
Hedge funds incorporate prediction market data into models. Better forecasts mean better positioning.
Asset managers use markets for risk assessment. Political risk, policy risk, geopolitical risk all reflected in prediction market odds.
Proprietary trading firms arbitrage between prediction markets and traditional markets. Profit from discrepancies.
Institutions get information before retail. Faster data feeds. Better analysis. More capital to deploy. Retail gets information late. Price already moved.
Who loses:
Retail bettors without insider access. They bet on public information. Compete against insiders betting on privileged information. Lose money systematically.
General public trying to understand world. Markets show odds but not reasoning. See "70% chance of strike" but not why. Information without context. Signal without meaning.
People who need actionable intelligence. By time odds shift, action opportunity is gone. Too late to prepare, hedge, respond.
The Efficiency Paradox
Prediction markets are accurate because insiders trade on privileged information.
Remove insider trading, markets become less accurate. Lose their forecasting value. Just aggregation of public information, no better than polls or expert surveys.
Allow insider trading, markets extract hidden knowledge. Create two-tier system. Those who know profit. Those who watch learn nothing actionable.
The paradox:
Can't have both democratic access to information and market efficiency.
Efficiency requires information asymmetry. Someone must know more than others. That knowledge must be tradable. Trading on it makes markets accurate.
But trading on asymmetric information creates inequality. Those with access profit. Those without lose. The efficient market is systematically unfair.
Attempts to solve this fail:
Ban insider trading: enforcement impossible, markets lose accuracy, activity moves offshore.
Require disclosure: participants hide identity, use intermediaries, route through crypto.
Limit position sizes: reduces market depth, makes odds less reliable, creates multiple small accounts.
Every solution undermines the mechanism that makes markets work. Markets are accurate precisely because they're unequal.
This isn't fixable. It's fundamental. Information has value. Value gets captured. Capture creates inequality. Inequality enables efficiency.
Traditional securities markets handle this through regulation. Insider trading illegal, enforcement uneven but exists, reduces worst abuses.
Prediction markets have no such framework. Insider trading is the feature, not the bug. Remove it and markets stop working.
What Prediction Markets Actually Do
They don't democratize information. They monetize asymmetry.
Someone knows strike is coming. Bets on it. Odds shift. Public sees shift. Public learns strike is likely. But can't act on information. Already priced in.
The market extracted hidden knowledge and made it visible. But visibility comes after opportunity. Learning something will happen is different from learning before you can act.
The two-tier system:
Tier 1: Know the information directly. Trade on certainty. Profit guaranteed (barring execution risk).
Tier 2: See market odds shift. Learn something is likely. Too late to profit. Maybe too late to prepare.
This is efficient for aggregating information. Not equitable for distributing it.
Markets know before you do. By time you know they know, the information is worthless to you.
The value proposition:
For intelligence agencies: free signal about insider expectations.
For institutions: data to incorporate into forecasts and risk models.
For platforms: transaction fees on volume.
For insiders: convert privileged access into profit.
For public: entertainment and delayed information.
The public's value proposition is weakest. Pay to watch odds shift. Learn what insiders already knew. Bet against people with better information.
Prediction markets aren't scam. They work as advertised. Aggregate information efficiently. Forecast accurately. Reveal insider expectations.
Just don't expect to benefit from them unless you're insider, institution, or platform. Everyone else is paying for the privilege of seeing information after it matters.
The Question That Remains
Not whether prediction markets are accurate. They are. Track record proves it.
Not whether they should be legal. Courts decided they are (in US, for now).
Not whether they're useful. They provide value to intelligence agencies, institutions, researchers.
The question: who benefits from that accuracy?
Insiders profit from privileged knowledge. Platforms profit from transaction volume. Institutions profit from incorporated data. Intelligence agencies profit from free signals.
The public gets accurate odds on events. But accuracy without ability to act on it is entertainment, not advantage.
Markets reveal truth. Just reveal it to those who already knew, by extracting profit from those who didn't.
Not conspiracy. Just incentives. Markets work because people trade on information gaps. Information gaps create inequality. Inequality makes markets efficient.
The accuracy and the unfairness are inseparable. Can't have one without the other.
Prediction markets extract truth. Also extract value from those who possess it first.
By time odds shift, you're not early. You're the exit liquidity.