Your Office Pool Just Hit $154B: How Prediction Markets Became the New Robinhood

Prediction markets have processed more than $154 billion in total volume, with daily trading on Polymarket alone often exceeding $300 million. At that scale, an uncomfortable question emerges: these platforms no longer look like niche betting venues. They increasingly resemble something closer to retail trading platforms—or perhaps the most degenerate Discord server ever to accidentally become a financial instrument.

This analysis uses on-chain data, primarily from Polymarket—the largest platform by users and transactions in a market dominated by a Polymarket–Kalshi duopoly—to test that shift directly. We're not here to moralize about whether guessing election results counts as "trading" or "gambling." We're here to look at the numbers and ask: who is actually using this thing, and do they look more like degenerates at a sportsbook or retail investors getting absolutely cooked in a different market?

Who Participates

The most revealing signal is not how much money flows through prediction markets. It is who is placing the trades. On Polymarket, the median bet size is $10, according to BeInCrypto's exclusive dashboard. The average sits at $89, but that figure is pulled upward by a thin tail of large participants. The underlying distribution paints a clearer picture: roughly 20% of all wallets trade in the $0 to $10 range, another 27% fall between $10 and $50, and about 11% sit in the $50 to $100 bracket. In total, over 57% of users trade for less than $100, and more than 80% trade for less than $500.

This is not a market shaped by whales. It is a market built on small, individual participants deploying modest amounts. The pattern mirrors what defined the rise of retail stock trading. Robinhood, for comparison, reported a median account size of $240, with the average around $5,000, according to CEO Vlad Tenev in 2021. The structural similarity is hard to miss: prediction markets are attracting the same class of small participants that reshaped equities over the past five years. Replace "I think NVDA goes to $800" with "I think Trump tweets something unhinged by Thursday" and you've basically got the same retail psychology, just with worse odds and more emotional damage.

How They Behave

Participation alone does not distinguish a financial platform from a betting one. Frequency of interaction does. A bettor places a wager and waits. A trader enters positions, adjusts exposure, exits, and re-enters. The transactions-per-active-user ratio captures this distinction directly. On Polymarket, this ratio currently stands at approximately 25 transactions per daily active user, meaning the average active participant executes 25 trades per day. Earlier this year, the figure peaked near 37. For context, through most of mid-2025, the ratio hovered between 3 and 5. The structural jump beginning in late 2025 represents a clear behavioral shift: users are no longer placing single predictions and walking away. They are actively managing positions across multiple markets. This is basically day trading, but instead of staring at candlesticks, you're staring at whether a Senate hearing gets adjourned early.

This pattern has a direct parallel in crypto markets. A Kaiko research report on Binance found that the exchange processed 61.9 million trades against $20 billion in spot volume on a single snapshot day in December 2025, implying small average trade sizes and frequent execution across its 300 million registered accounts. High-frequency, small-size trading is the behavioral signature of retail finance, whether the underlying asset is a stock, a token, or a prediction contract. The only difference is that crypto traders get to lose money on leverage, while prediction market degens lose money on being wrong about whether it rains in Austin next Tuesday.

How Capital Moves

If users behave like traders, the capital dynamics should confirm it. They do. Polymarket currently holds approximately $445 million in total value locked, while open interest stands at roughly $477 million. The near-parity between these two figures carries a specific implication: virtually all deposited capital is actively deployed in live positions rather than sitting idle. This is not passive liquidity. It is working capital. Your grandma's savings account doesn't look like this. This looks like a trading desk where everyone is simultaneously YOLOing and hedging in a way that would make a quantitative analyst weep.

The volume-to-open-interest ratio reinforces the point. With daily taker volume around $339 million and open interest at $477 million, the ratio is 0.71. Capital is not just deployed. It is rotating. Positions are being opened, closed, and re-entered at a pace that suggests continuous portfolio management rather than static, event-dependent exposure. A low vol-OI ratio would have suggested more betting-like activity. In a traditional betting market, capital tends to lock in and wait for resolution. Here, it circulates. That distinction is material: it signals a system in which participants treat capital as a tool for ongoing risk adjustment, not a one-time stake in a single outcome. This is literally what market makers do, except instead of capturing spread, these people are just trying to call whether Congress will pass a bill before their girlfriend's birthday.

At What Scale

The behavioral and capital patterns described above would be noteworthy even at modest volumes. But they are not operating at modest volumes. Polymarket's weekly notional volume has consistently exceeded $1 billion through Q1 2026, with recent weeks surpassing $2.5 billion. The 7-week rolling average has crossed $2 billion. Monthly volumes have climbed from around $1 billion in mid-2025 to over $8 billion by March 2026. To put this in perspective: that's more weekly volume than some small countries' entire GDP, and it's all being wagered on whether a celebrity will get cancelled or if Bitcoin hits a certain price by Friday.

The growth trajectory is not driven by any single event cycle. Volume is diversifying across categories: sports, crypto, and politics. Each contributed substantially in the most recent weekly data, with economy, weather, and culture adding further breadth. This diversification is what separates structural growth from event-driven spikes. A presidential election creates a temporary surge. Sustained, multi-category volume growth across sports, crypto, macro, and culture points to a user base that engages with prediction markets regularly, not just occasionally, as a typical retail habit. This isn't a once-every-four-years thing. This is