Prediction markets are sold as truth machines. Critics reply they are casinos with charts, vulnerable to manipulation, inequality, and moral hazard. Both sides overstate.
This closing chapter of Foundations of Prediction Markets names the real limitations—technical, ethical, and social—so you use market odds as tools, not idols.
“It’s just gambling with extra steps”
Retail does lose money on sports and event contracts; fees extract edge; addiction externalities mirror wagering apps with push alerts. That criticism is fair for many users who chase losses and trade emotionally.
The nuance is that institutional use (hedging, risk desks, research) pursues information, not entertainment. Regulated venues argue market integrity rules and surveillance exceed unlicensed bookmakers.
Ask whether you are forecasting or gambling on variance. If you cannot state resolution rules and base rates, it is likely the latter.
Manipulation and insider trading
Thin markets let whales, campaigns, or insiders move headline odds for signaling or profit before news is public. Suspicious flow has been reported to prosecutors; oracle governance fights erupt on crypto venues; wash trading fakes volume on obscure listings.
Defenses are partial: position limits, surveillance, KYC tiers, reporting to authorities, and market maker programs on headline contracts.
Require volume, spread, and cross-market consistency before citing a price. Ninety percent YES on $1,200 is not “the market believes”—it is one trade away from comedy.
The crowd isn’t diverse—it’s weird
Crypto-global traders pricing US school-board races are not the “wisdom of crowds” in the textbook sense—they are a niche fandom with different demographics and incentives than voters. Participant selection bias breaks aggregation theorems from the wisdom-of-crowds chapter.
Mitigations include regulated US pools, stake caps that widen participation, and multiple venues for comparison. Label whose beliefs you are seeing: Kalshi traders, Polymarket wallets, PredictIt academics—not “America.”
Prices oversell precision
Media prints 62.4% from a three-cent-wide market on a fuzzy event—false precision breeds false confidence. Implied probabilities ignore spread, fees, and model error; tail events in the 1–5% bucket are especially unreliable.
Round for public communication; internally stress-test wide moves when liquidity is mediocre.
Ethical and social harm
Some listings commodify tragedy—assassinations, pandemics, wars, personal suffering—and can incentivize awful outcomes. Industry responses include listing policies, delisting, and geographic bans, still inconsistent.
Separate “can we price this?” from “should we?” Platforms make that call; traders vote with participation.
Political legitimacy and democracy
Election markets can undermine faith in democracy if odds move on leaks, or if officials trade on inside information. Perception risk is real even when markets are directionally accurate; Congress explores trading bans for military and federal officials.
Treat election prices as inputs to analysis, not election administration tools.
Resolution and oracle failure
You are not always trading the event—you are trading rules lawyers and token votes. Ambiguous wording, disputed certifications, and contentious oracle majorities overturning obvious outcomes have all happened.
Read resolution before entry; size down when settlement is novel or governance is thin.
Inequality of information and capital
Well-capitalized firms with faster data feed on retail flow—markets aggregate money-weighted beliefs, not democratic equality. Superforecasters without capital cannot move prices.
If prices move before public headlines, you are late unless you have a niche edge.
Poor track record on long horizons
Markets shine on short, high-attention events; multi-year structural questions stay illiquid and stale. A 2026 GDP regime shift does not trade like tonight’s playoff—do not expect poll-grade discipline everywhere. Match instrument to horizon; use experts and models for decadal questions.
Regulatory arbitrage distorts global odds
The “same” election trades at different prices because different people can access different venues under different rules—not because information differs. Arbitrage is incomplete; compare venues only after harmonizing resolution and fees.
A balanced view
| Claim | Kernel of truth | Overreach |
|---|---|---|
| Gambling | Retail harm exists | Ignores hedging/research use |
| Manipulation | Real on thin books | Headline markets have surveillance |
| Bad crowd | Selection bias | Can improve with access/rules |
| False precision | Spread ignored | Still useful ordinal signal |
A disciplined checklist
Before you stake capital or cite a contract in an article, ask seven questions. Is there enough liquidity that a $10k order would not materially move price? Is resolution clear to two honest lawyers? Who are the participants legally allowed in this pool? Is there independence, or is everyone on the same tweet thread? Do polls, fundamentals, or sister markets corroborate? Is your purpose forecast, hedge, or entertainment? Should this question exist at all?
Pass most of them and the price deserves a place in your model. Fail several and treat it as noise.
What good critics still concede
Even skeptics often admit that continuous, incentivized prices beat silence during fast-moving crises—if you read them with humility. The fight is over claims: “markets predict winners” oversells; “markets are useless” undersells when liquidity and rules are sound.
Your job after Module 01 is to cite odds with context: venue, volume, resolution, and corroboration. That discipline is what separates informed users from meme screenshots.
Closing Module 01
You now have vocabulary, price–probability logic, crowd conditions, history, comparison to polls and experts, applications, law, and skepticism.
Module 02 — Market Mechanisms & Trading Infrastructure moves from what markets are to how they are built and traded: order books, AMMs, liquidity, and infrastructure. Bring this chapter’s doubt with you—the best builders know exactly where their machines fail.