Prediction markets look like a casino. The typical user behaves like a fan with a $6 bet.

Polymarket and Kalshi together clear nearly $24 billion in monthly trading volume, and a thin tail of heavy accounts drives most of it. The typical account on Polymarket looks nothing like that headline. A new Pew Research Center analysis of about 12,000 accounts that traded on ten high-volume events in early 2026 puts three numbers in front of readers: 46 trades, an average bet of $6.50, and a net loss of about $2 over six weeks.

The distribution is bimodal, and the second mode does the heavy lifting. Eleven percent of accounts placed 1,000 or more trades in the same window. Nine percent lost more than $1,000. Seven percent made more than $1,000. The tail cuts both ways, and the tail, not the median, is what explains the $24-billion headline. Sports, politics, and crypto account for nearly all of Polymarket's volume.

Topic concentration sharpens the picture. The median user put roughly three-quarters of their trades on a single topic, and a quarter of all users traded on only one. The split varies sharply by lane: sports traders averaged 69 trades over the period, crypto traders 59, and politics traders 13. Sports traders are the largest of the three groups, and the median sports user places more than five times as many trades as the median politics user.

The scope is narrow and worth naming. Pew's data covers one platform, ten high-volume events, six weeks of activity from May 7 to June 19, 2026, and on-chain public activity that may include bots, market makers, or wash trading rather than ordinary retail users. The dataset does not break out age, country, or profession. Long-tail markets and quieter weeks sit outside the read.

The benchmark is concrete enough for a non-beat reader to place themselves: 46 trades, $6.50 a bet, a few dollars net, and one topic the trader actually cares about.