Concept · liquidity-and-trading
Retail Flow
Quick definition. Trading activity from non-professional participants · generally less informed, more predictable, and the desired counterparty for market makers. Retail flow is the economic engine that makes the rest of prediction-market microstructure work; without it, MMs can't profitably absorb adverse selection from informed traders.
Key insights
- Kunal Doshi on Polymarket fast crypto: 19 algo addresses extract consistent profits; 69% of retail traders lose money. Fast markets account for 16% of volume but ~40% of fees.
- Isaac Rose-Berman ("Kalshi's Favorite Lie"): the exchange model still depends on retail losing. Every Kalshi trade has a maker (offers price) and a taker (accepts price); Kalshi takes a fee from both sides. Because fee = 0.07 × C × P × (1-P) peaks at 50/50, and retail almost always crosses the spread as the taker, a fair coin-flip bet becomes a 3.4% expected loss after fees to the retail taker. Maker dominance is "structural, not stylistic" · only the fastest systems and best models can profitably make.
- Rose-Berman: Kalshi's pitch ("we don't care who wins") is true but misleading · Kalshi profits from trading volume, and trading volume requires retail to keep losing slow and coming back. The marketing claim that PMs are fundamentally different from sportsbooks is "a much bigger fiction." Sportsbook = casino; Kalshi = casino-funder model.
- Momin: Polymarket data shows 70% of 1.7M addresses lost money; the top 0.04% captured >70% of $3.7B realized profits. Platforms predictably funnel retail into informed counterparties, including platform-operated MM desks at Kalshi and Crypto.com.
- Jordan Bender (Wall Street equity research): median PM user ROI is -8%, worse than sports bettors (-5%); only >$500K volume traders hit positive returns (+2.6%). PMs attract sharper competition than regulated sportsbooks, creating worse outcomes for casual retail.
- functionSPACE: noisy traders fund the probability space · they are not exit liquidity but the necessary fuel for MM economics.
- Becker (72.1M Kalshi trades, $18.26B volume): systematic taker→maker wealth transfer of 1.12% excess returns on each side; takers disproportionately buy YES longshots, accepting -64 pp worse returns. The transfer only emerged after October 2024 when algorithmic MMs arrived.
- Becker on category-level retail vulnerability: Finance gap 0.17pp (efficient · "filters out emotional bettors"); Sports 2.23pp; Crypto 2.69pp; Weather 2.57pp; Entertainment 4.79pp; Media 7.28pp; World Events 7.32pp. The retail loss rate scales directly with the category's emotional engagement.
- Becker on the YES/NO asymmetry: at 1-cent prices, YES contracts have a -41% expected return while NO contracts have +23% · a 64-percentage-point gap at "identical" prices. Takers account for 41-47% of YES volume at 1-10¢ but only 23% of NO volume at 99¢. Retail is paying for the framing.
- Della Vedova (222M Polymarket trades): forecasting accuracy does not predict profitability. Even retail who pick the right side lose because they arrive late at unfavorable prices. Automated traders profit by paying 2.52¢ less per contract despite near-random directional skill.
- functionSPACE on yes-bias: traders buy whichever token is cheaper, not whichever is labeled YES · the apparent yes bias is a compound effect of longshot preference channeled through Polymarket's "Will X happen?" framing.
- sealaunch: 2% of users generate ~90% of platform volume. Crypto markets dominated by algo execution; politics markets by event-driven casual participants. Different user-growth vs. volume-growth product strategies.
- 4casters/sportsbook square-vs-sharp: Kalshi (3.5% fee) and Polymarket (1.5%+) monetize price-insensitive retail takers. Raising fees from 0.5% → 0.75% on 4casters had no volume impact · sports bettors are price-insensitive.
- Hariharan Dopamine 2025 letter: retail financial speculation has permanently shifted from investment to entertainment; 0DTE options = 59% of options volume in 2025. Retail participation went from a trivial share to 100M+ monthly active traders over the past decade.
- Hariharan on the cultural shift: "the smartphone has put a casino in every pocket" · the path is "to a billion active traders." Investing has converged with entertainment; the SEC has been replaced by a technology-forward permissive regime; BREAKING news now comes with live odds.
- Ranger Global biases: longshot (5¢ contracts win only 4.18% of the time), maker-taker asymmetry (makers outperform at 80 of 99 price levels), YES/NO asymmetry (YES buyers -1.02% vs. NO buyers +0.83%).
- Whitaker & Mazlish on retail's structural fit: gamblers want immediate resolution (42% of 2020 election volume traded in the final week); long-term investors prefer compounding; MMs need consistent retail flow that only materializes for elections/sports. Outside those niches, retail demand collapses and the markets stay illiquid.
In their words
70% of 1.7 million addresses lost money on Polymarket; the top 0.04% captured over 70% of the $3.7B in realized profits.· Momin
Forecasting accuracy does not predict profitability.· Della Vedova
Whales are not the sharpest participants.· Deleep et al.
Retail financial speculation has permanently shifted from investment to entertainment.· Hariharan
Kalshi uses those truths to sell a much bigger fiction: that because it profits from trading volume rather than directly from user losses, its business does not depend on users losing money.· Rose-Berman
Takers exhibit negative excess returns at 80 of 99 price levels. Makers exhibit positive excess returns at the same 80 levels.· Becker
The smartphone has put a casino in every pocket, and people can't help but spin the wheel.· Hariharan
Where it matters
Retail flow is the precondition for everything else · without it, there's no spread to harvest and no liquidity to lean against. But every microstructure study agrees the modal retail trader loses, often more than they would at a regulated sportsbook, because PMs attract sharper sophisticated competition. This makes consumer-protection regulation, fee design, and category mix (politics vs. sports vs. entertainment) decisions about how much retail is allowed to lose, and to whom.
Connections
- Market making / adverse selection · retail flow is the LP's profit; informed flow is the LP's loss.
- Toxic flow · the opposite of retail flow.
- Execution quality · what separates retail from pros.
- Longshot bias / yes bias · empirical retail-flow behavioral footprints.
- Bid-ask spread · the tax retail pays for immediacy.
- Platform competition · design choices about how to attract or shield retail.
- Regulatory classification · determines what consumer protections apply.
Platforms linked to this concept
- Kalshi · implements · Mentioned in Retail Flow content as an implementing platform
- Polymarket · implements · Mentioned in Retail Flow content as an implementing platform
Related concepts
- Market Making
- Adverse Selection
- Toxic Flow
- Execution Quality
- Longshot Bias
- Yes Bias
- Bid-Ask Spread
- Platform competition
- Regulatory classification
Sources
- A Game of Volatility · Kunal Doshi · May 12, 2026
- Kalshi's Favorite Lie · Isaac Rose-Berman · Apr 29, 2026
- The Prediction Market Epidemic: Who's Actually Winning · Momin · Apr 21, 2026
- What Happens When Institutional Liquidity Enters Prediction Markets · Daedalus Research · Apr 20, 2026
- Faster, Shorter, More Automated: Anatomy of Polymarket's Fastest Markets · Dune · Apr 14, 2026
- The Two Kinds of Prediction Markets · 4casters · Apr 9, 2026
- The Yes Bias Might Not Exist · functionSPACE · Mar 27, 2026
- Is Polymarket a Retail Product or a Pro Trading Venue? · sealaunch intelligence · Mar 27, 2026
- Prediction Markets vs. Sports Betting: Market Dynamics, ROI by Cohorts, and Competitive Implications · Jordan Bender · Mar 23, 2026
- How Wise Is the Crowd? Bias and Edge in Prediction Markets · Deleep, Lee, Bai, Suresh, Dhawan · Feb 28, 2026
- Prediction Markets Are Not Good Markets (Yet) · Nic Carter · Feb 21, 2026
- Who Profits from Prediction Markets? Execution, Not Information · Joshua Della Vedova · Feb 7, 2026
- Prediction Market Biases Revealed in 72 Million Trades · Ranger Global · Jan 29, 2026
- The Microstructure of Wealth Transfer in Prediction Markets · Jonathan Becker · Jan 18, 2026
- Dopamine Markets: 2025 Annual Letter · Shreyas Hariharan · Jan 8, 2026