A CNBC analysis published on July 2 found that most prediction markets on Polymarket attracted limited trading activity despite the platform's rapid overall growth in recent years.
Using data from Polymarket's Gamma API, CNBC found that about 70% of all closed markets between 2021 and the end of May 2026 recorded less than $10,000 in reported trading volume, while more than 45,000 markets, nearly 5% of all closed markets, had no reported volume. Fewer than 10% of completed markets generated between $100,000 and $1 million in reported volume.
The findings suggest that Polymarket's headline trading growth has been concentrated in a relatively small number of active markets, while most markets remained thinly traded.
CNBC also found that Kalshi, Polymarket's main competitor, recorded a large number of low-volume markets based on data analyzed through the on-chain platform Dune. Polymarket's Gamma API counts notional volume on both sides of each trade, while Kalshi's Dune data counts only one side.
Economists told CNBC that low-volume markets are generally more volatile, can expose traders to wider bid-ask spreads, and may increase trading costs for participants. The findings also provide context for how liquidity is distributed across prediction markets as the sector continues to grow.
Researchers Say Thin Prediction Markets Are More Volatile as Bots Dominate Low-Volume Trading
Low-volume markets can create difficult conditions for traders because small orders can move prices more sharply than they would in deeper markets. Constantin Bürgi, a professor of economics at University College Dublin, told CNBC that thin markets are “typically more volatile” because limited activity makes prices more sensitive to individual trades.
Eric Zitzewitz, a professor of economics at Dartmouth College, said newer traders can also face wider spreads between buy and sell prices in shallow markets. Wider spreads make positions more expensive to enter and exit, especially when traders need to move quickly before a market resolves.
Seasoned traders often prefer shorter and more active contracts. Logan Sudeith, a former financial risk analyst in Atlanta who began trading prediction markets full-time last fall, told CNBC he prefers higher-volume, short-term markets because they are “more capital efficient.” CNBC found that markets lasting up to one week produced the highest number of Polymarket contracts with at least $1 million in reported volume, including contracts tied to the Iran war, U.S. President Donald Trump and Elon Musk.
Automated trading also plays a major role in thinner markets. Joshua Della Vedova, a business professor at the University of San Diego, told CNBC that bots accounted for more than 80% of volume in Polymarket markets with less than $10,000 in volume. Della Vedova classified wallets as bots if they made more than 50 trades per day or more than 1,000 trades in total.
Using on-chain data from Polymarket, Della Vedova found that bots made about $1.2 million in shallow markets and about $35.1 million in markets with more than $10 million in trading volume from November 2022 to February 2026. He said bots made money across markets of all sizes, unlike retail traders, who have faced losses in both shallow and heavily traded markets.
Experts Disagree on Whether Low-Volume Prediction Markets Produce Reliable Signals
Researchers are split on whether thin prediction markets produce reliable signals. Evercore ISI strategists found that higher-volume markets produced more reliable probabilities after reviewing five years of completed markets on Polymarket and Kalshi. The strategists said only 8% of markets across both platforms reached $1 million in volume, leaving most quoted probabilities in thinner markets where calibration was weaker.
Other researchers said volume alone does not determine accuracy. Theis Ingerslev Jensen, a professor of finance at Yale University, told CNBC that market quality depends more on who is trading than how much trading takes place. Jensen and researchers at London Business School found that skilled or informed traders drove much of Polymarket’s accuracy.
“Thin markets are not automatically inaccurate, but they are less reliable,” Jensen told CNBC. He said the key issue is whether skilled traders still have enough incentive and ability to trade.
Harry Crane, a professor of statistics at Rutgers University, said low liquidity should be considered when reading prediction market prices, but it does not make a market useless. Crane said prediction markets can still provide signals as long as traders understand the limits of each contract.
Polymarket declined to comment on CNBC’s findings, while Kalshi did not respond to CNBC’s requests. Crane said larger markets are likely to keep growing as overall prediction market volume expands, while low-volume markets may remain shallow.
