2026-11-11 / essays / 7 min read

Prediction Market Volume Is the Wrong Headline Number

Reported volume is the wrong headline metric for prediction markets. In our November 11, 2026 pull, 68.9% of Kalshi's 24-hour volume came from contracts that open and close within a day, and Polymarket's intraday crypto series did 18.1% of its volume on $26,229 of open interest. We propose open interest plus contested depth.

Our position is that reported volume is the wrong headline number for prediction markets, because on November 11, 2026 most of it came from contracts that reset within a day, from programs that pay by the trade, and from bets on outcomes priced near zero, and that open interest plus depth near the mid on contested contracts measures what readers think volume measures: how much money stands behind a price.

Volume is the number both venues lead with. Pew Research, citing The Block, charted combined Kalshi and Polymarket volume rising from $2.0 billion in August 2025 to $53.0 billion in July 2026, and CNBC reported on September 30, 2026 that both companies cite volume as evidence of growth while raising money at valuations above $20 billion. We wanted to see how far that number sits from the money at risk.

What did we measure?

At 22:13 UTC on November 11, 2026 we pulled every open Kalshi event with its markets from Kalshi's public Trade API, plus every market that closed or settled in the prior 24 hours: 279,882 markets. Kalshi reports volume and open interest in contracts with a $1 face value, so the two are directly comparable. The public events listing leaves out multi-leg combos, so our Kalshi figures cover single contracts only.

At 22:38 UTC we pulled every open Polymarket event with any 24-hour volume from the gamma API (4,400 events), plus every event that closed in the prior 30 hours (6,048), and used each event's volume24hr, openInterest and liquidity. At 22:50 UTC we fetched the order book for all 14,573 open Polymarket contracts that had traded in the prior day. Categories are Kalshi's own and Polymarket's event tags. This is one Wednesday, and the 24-hour figure for a closed market is whatever the venue's API reports for it.

How far apart are volume and open interest?

On Kalshi, very far, and in opposite directions by category.

A contract in Kalshi's crypto category traded nearly 400 times as often, per unit of open interest, as one in elections. Polymarket shows the same split.

Ranked by volume, Kalshi looks like a crypto and sports exchange and Polymarket like a sportsbook with a politics sideline. Ranked by money committed, elections and politics are over a third of Kalshi and 60% of Polymarket. Both rankings are true, and only the first tends to reach headlines.

Where does the extra volume come from?

Four sources, each of which raises volume without raising the money behind any price.

Contracts that reset every few minutes. On Kalshi, 68.9% of 24-hour volume came from contracts that opened and closed within the same 24 hours. In crypto the figure was 99.2%, and the bitcoin 15-minute series alone traded 392.0 million contracts, 34.6% of all single-contract volume on the exchange. On Polymarket, crypto series running from 5 minutes to 4 hours traded $19.0 million, 18.1% of the venue's volume, while holding $26,229 of open interest at the moment we looked. DeFi Rate's tracker notes that Kalshi's 15-minute crypto markets settle 96 times a day, so the same money trades again and again. Each of those trades is real, but a dollar that turns over 96 times in a day is still one dollar at risk.

Incentives paid per trade. Kalshi ran a Volume Incentive Program that split a fixed pool by each trader's share of qualifying volume, until a September 28, 2026 filing moved its end date from October 1, 2027 to October 13, 2026, DeFi Rate reported. On August 12, CFTC staff issued Letter 26-23, which, per Willkie Farr & Gallagher, told exchanges to review "trading patterns that may exploit threshold- and rebate-based rewards (e.g., through wash trading, pre-arranged trading, or other trading misconduct)." Polymarket US still pays market makers by volume: its amended program, filed with the CFTC on September 30, 2026, awards rebates "based on each Program Participant's share of total maker notional volume within each opted-in Series," with a 5% share required to qualify. And Polymarket's chief marketing officer said in October 2025, per The Block, "There will be a token, there will be an airdrop," without published eligibility rules. CNBC quoted an analyst saying eligibility could depend on open interest or individual volume. Where a reward scales with volume, volume partly measures the reward.

Wash trading. A Columbia University study led by Allen Sirolly, covered by Decrypt on November 7, 2025, estimated that about 25% of Polymarket's volume over three years was likely wash trading, about $4.5 billion, flagged in 14% of 1.26 million wallets. It peaked near 60% of weekly volume in December 2024, reached 90% in sports markets in the week of October 21, 2024, and 95% in election markets in the week of March 24, 2025. The paper, titled "Network-Based Detection of Wash-Trading" per Cointelegraph, had not been peer reviewed. CNBC reported that the measure had fallen to a negligible level by April 2026, which Polymarket credits to surveillance and new fees. Separately, Paradigm's Storm Slivkoff showed on December 8, 2025 that many dashboards summed both the maker and taker records of each Polymarket fill, so "a simple sale of YES tokens for $4.13 gets recorded as $8.26 worth of volume." Paradigm discloses that it invests in Kalshi.

Volume on outcomes almost nobody expects. In our pull, 41.7% of 24-hour volume on open Polymarket contracts traded at prices of 5 cents or less, or 95 cents or more. CNBC found $158 million traded on Egypt winning the World Cup, a contract never priced above 0.5%, against $152 million on Spain. Polymarket's Kyle Gesuelli attributed such patterns to "sharps" correcting mispricings.

What should the headline be instead?

We propose two numbers, with volume demoted to a ratio beside them.

Open interest. It counts money committed to an outcome once, however often it changes hands. Polymarket had $292.1 million on November 11; Kalshi had 1.33 billion contracts. DeFi Rate's 30-day figures to November 10 show the gap at venue level: Kalshi traded $107.1 billion against $1.4 billion of open interest, about 76 times, and Polymarket US traded $15.9 billion against $66 million, about 241 times.

Contested depth. By this we mean dollars resting within 2 cents of the midpoint on contracts priced between 10 and 90 cents, where a probability is in dispute. Across the 14,573 traded Polymarket contracts, $116.0 million sat within 2 cents of the mid, but $80.9 million of it (69.7%) was on contracts priced under 10 or over 90 cents. Contested depth was $35.1 million. Politics had $63.5 million near the mid and $2.8 million of it contested; sports held 85.1% of all contested depth. Kalshi's listing shows only the best bid and ask: on contested contracts quoted 2 cents wide or tighter, $86.6 million sat at the touch, $78.4 million of it in sports and $208,461 in crypto, the category that did 44.3% of the day's volume. Our liquidity page and our essay on where order books are thin track the same shape from the depth side.

Volume then appears as turnover: volume divided by open interest, by category. A turnover of 0.04 and one of 74.6 describe two different products, and readers deserve to know which one a headline is counting.

What is the strongest case against our view?

The strongest objection is that volume is what pays the bills, and that our replacements can be gamed too. Fees are charged per trade, so volume tracks a venue's revenue. And Polymarket pays a "4.00% annualized" holding reward on positions in eligible markets, a direct subsidy to open interest, while both venues pay for resting orders, a subsidy to depth. Our companion essay on how market makers set prices finds depth bunched inside Polymarket's reward bands. On this view we would swap one inflated number for two.

We accept the first point: volume is the right number for a venue's income statement, and investors should see it. Headlines use it for something else, as a proxy for how much the market knows and how much is at stake, and for that it is the weakest of the three. On gaming, the costs differ. Inflating open interest or depth means keeping capital at risk for as long as the claim stands. Inflating volume costs only the fee, and the Columbia authors noted Polymarket charged no transaction fees when they wrote. The contested filter also strips out the cheapest subsidized depth, the longshot walls. Venues that want their depth believed can publish reward spend beside it.

What would change our mind

• A study of resolved markets showing 24-hour volume predicts calibration better than open interest or contested depth.
• Venues publishing one-sided, wash-filtered volume with an outside audit, and that figure staying close to open-interest measures.
• Contracts lasting under a day falling below a fifth of volume on both venues while headline volume holds up.
• Evidence that holding and liquidity rewards inflate open interest and depth as much as volume rewards inflated volume.