Skip to content

Research: Why Retail Loses

The core finding: 70% of prediction market trading addresses lose money. Fewer than 30% earn net positive returns across 124 million trades (Reichenbach & Walther). This is not bad luck — it’s structural.

The same study shows that sophisticated users do fine: traders transacting $500K+ median ROI is +2.6%. Traders transacting under $100 median ROI is −26.8%. The platform is a wealth transfer machine from unsophisticated to sophisticated participants.

Post-election 2024: Polymarket MAU dropped 57% from January to August 2025. Volume held at $7.9B. The casual users left; power users stayed. Monthly retention: 35%. That outperforms 85%+ of crypto protocols, but it means 65% of acquired users are gone within 30 days.

The question: What causes this? And what would fix it?


#BarrierSeverityWhat’s actually happening
1Loss aversion & churn★★★★★Users exit after first losing position. The pain of loss > pleasure of win at 2:1. No mechanism to recover emotionally or financially.
2Capital lockup★★★★Winning positions sit locked until resolution — sometimes weeks. Users can’t redeploy capital. Power users in particular feel this acutely.
3No skill recognition★★★★Good forecasters earn no credential. There’s no track record. Being right repeatedly has no compounding value — each market resets.
4Market integrity doubt★★★½Resolution failures (Zelensky suit, Oscar viewership, Venezuela) create distrust. Users who’ve been wrongly resolved exit and don’t return.
5Onboarding friction★★★Crypto wallet + fiat onramp + UI learning curve = 3 barriers before the first bet. Most casual users don’t clear all three.
6Liquidity gaps★★★63% of active Polymarket markets have zero 24-hour volume. 67.7% of all markets have lifecycle under 7 days. Thin markets = bad execution = losses.
7Information asymmetry★★½Power traders and AI agents have information advantages. Retail users are providing liquidity to better-informed counterparties.

Prediction market participation is heavily right-skewed:

PercentileTypical bet size (Polymarket)
Median$30–55
P75$200–400
P90$1,000–2,000
P99$20,000–50,000
Mean (2024 Trump contract)$210

28% of all bets are $10–50. These small-bet users are the first to churn — the $30 loss has high emotional salience relative to their budget, and the $30 win has almost no measurable financial impact.

The protocol’s 1% power users generate 70% of volume. Retention of this cohort is the survival metric.


Prospect theory predicts users weight losses 2× heavier than equivalent gains. For a prediction market:

  • A user who wins a $100 position resolving at $1.00 feels moderate satisfaction
  • A user who loses a $100 position to zero feels intense distress — approximately 2× the emotional magnitude

Standard prediction market design makes this worse:

  1. Binary resolution (win everything or lose everything)
  2. No mechanism to partially exit a losing position
  3. No insurance product to cap downside
  4. No recovery path — the loss is total and final

Churn after a losing experience is not irrational. It’s the rational response to a product that offers no loss mitigation.


Barrier fixedMechanismExpected retention improvement
Loss aversionREWIND insurance15–30% improvement in 30-day retention
Capital lockupFast-Forward early exit20–40% improvement in capital utilization / redeployment
No skill recognitionChrono Score30–50% improvement in long-term retention (6–12 months)
Market integrity doubtAI resolution + FREEZE dispute layer5–15% improvement in institutional participation
Onboarding frictionPRECOGNITION guided onboarding5–10% improvement in activation rate

Conservative estimate: addressing the top 3 barriers improves total addressable market by 15–30%. This is comparable to the effect Compound/Aave had on DeFi lending by adding yield to idle capital — it unlocked a category of users who wouldn’t participate otherwise.


The 70% loss rate is not just a user welfare problem. It’s a liquidity problem.

Prediction markets need a continuous supply of “incorrect” participants to function — someone has to be on the other side of the smart money. When retail users lose and churn, the market becomes dominated by sophisticated traders playing zero-sum games against each other. Volume concentrates in fewer markets, bid-ask spreads widen, and markets on long-tail topics disappear entirely.

Insurance doesn’t just protect retail users — it makes them willing to return to the market after losses. That’s the demand-side flywheel: more insured users → more diverse market participation → more liquid markets → better information → more user trust → more participation.

The protocol that solves retail retention doesn’t just earn insurance premiums. It earns the entire TAM expansion that follows.

Related: