2026-11-11 / market structure / 6 min read

Polymarket Liquidity by Category: Where Order Books Are Thin

As of November 11, 2026, Polymarket's 300 deepest open events carried $406.7 million of reported liquidity across 20,004 contracts, by our count. 71% sat on contracts priced under 5 cents, and 87% of contracts had no trades in 24 hours. Our view is that most of that depth is paid-for inventory, and most prices on the venue are unpriced.

Our thesis is that Polymarket's headline liquidity mostly measures capital the venue pays to sit on near-certain longshots, while the contracts where a price would tell you something are thin or empty, so a Polymarket price should be read as a probability only when a tight, trading book stands behind it.

What did we measure?

At 05:07 UTC on November 11, 2026 we pulled the 300 open events with the most liquidity from Polymarket's public gamma API, sorted by liquidity, in three pages of 100. Inside them we kept every open contract with positive reported liquidity: 20,004 contracts. For each we used four fields. liquidityNum is what Polymarket's market documentation calls "reported market liquidity," without a formula. bestBid and bestAsk give the spread, which we computed only where both sides existed. volume24hr gives daily volume, and volume divided by liquidity gives turnover: how many times the resting book traded over in a day. We grouped events by their first matching tag among sports, politics, crypto, economy, geopolitics, pop-culture and tech, and called the rest "other."

Totals: $406.7 million of reported liquidity and $33.5 million of 24-hour volume, an aggregate turnover of 0.08 times a day. Our live liquidity page runs the same pull on every build with a few more categories split out, so its numbers will move away from the ones below.

Where is the liquidity, by category?

The median turnover of zero in politics, sports and pop culture is literal. Across all categories, 17,340 of the 20,004 contracts (86.7%) had no trades in the prior 24 hours. The median contract carried $439 of liquidity, and 10,510 carried less than $1,000.

Where does the depth concentrate?

The 10 deepest contracts held 6.2% of all liquidity, which looks diffuse until you read them. Every one is a 2028 US presidential longshot: Hillary Clinton, Oprah Winfrey, Kim Kardashian, Phil Murphy, Barack Obama, Beto O'Rourke, Liz Cheney and Graham Platner for the Democratic nomination, and Greg Abbott and Tim Walz for the presidency. By event, the 10 deepest held 54.9% of the total, and the four deepest (the three 2028 US presidential events and the next French presidential election) held 42.5% of liquidity while drawing 11.2% of the day's volume.

The Kim Kardashian nomination contract had $2.71 million of reported liquidity, a YES price of 0.05 cents, and $2.90 of volume in 24 hours. In the Democratic nomination event, the favorite in our snapshot, Jon Ossoff at 26.6 cents bid and 26.7 cents ask, had $1.05 million of liquidity. Clinton, at 0.2 to 0.3 cents, had $2.79 million. The 27 contracts in that event priced under 1 cent held $29.3 million of its $65.6 million.

Across the whole pull, contracts with a midpoint under 5 cents held 71.3% of all liquidity. Contracts priced between 20 and 80 cents, where a probability is most contested, held 16.2%.

Why is the depth on longshots?

Our read is that Polymarket pays for it, in three ways its documentation describes.

First, holding rewards: Polymarket "pays a 4.00% annualized Holding Reward based on your total position value in eligible markets." The API flags eligibility per market. In our pull, 275 contracts had it enabled, 1.4% of the total, and they held 40.1% of all liquidity and 11.4% of daily volume. All but two expire in 2027 or 2028: the presidential fields and the full-year crypto price ladders.

Second, liquidity rewards score resting orders by distance from the midpoint and, when the midpoint is below 10 cents or above 90 cents, count only two-sided quotes. A longshot priced at a fraction of a cent is close to the cheapest place to keep a two-sided quote alive, because the price barely moves.

Third, fees fall to almost nothing at the extremes. Polymarket's maker-rebate page gives the fee formula as shares times rate times p times (1 minus p). At a 4% politics rate, a taker pays 1 cent per share at 50 cents and 0.004 cents at 0.1 cents. Makers in politics also receive 25% of taker fees back as rebates.

None of this is improper. It does mean the $406.7 million figure mostly describes capital earning a yield on outcomes almost nobody thinks will happen.

What do thin books do to price quality?

Sports shows the other end. Sports had 17,463 of the 20,004 contracts, 93% of them with no trades in 24 hours, and a median spread of 90 cents. 9,189 sports contracts were quoted at least 50 cents wide, with a combined $1.1 million of liquidity and a median of $2 each. A League of Legends prop on "Both Teams Slay Baron Nashor?" sat at 3 cents bid and 97 cents ask with $2.32 of liquidity, and the API reported its price as 50%. No trade produced that number.

The 20-to-80-cent band makes the point sharper: 14,029 contracts, median spread 97 cents. Remove sports and 224 contracts remain, with a median spread of 2 cents and $13.7 million of liquidity. Outside the headline games, much of what Polymarket lists in the contested range has no price at all.

Where flow is real, it is concentrated. "Fed Decision in December?" traded $7.26 million in 24 hours on $2.28 million of liquidity, turning over 3.2 times and producing 21.7% of all volume in the 300 events. Thunder vs. Trail Blazers had 400 contracts, of which 250 had no trades; the moneyline alone, at 57 cents bid and 58 cents ask with $325,000 of liquidity, traded $2.0 million. Our wallet-concentration essay found a similar shape in who trades; this pull finds it in where the books are.

Who do the venues serve?

On this data, two groups. Takers on a small set of flagship questions (a Fed meeting, a primetime game, a bitcoin price ladder) get tight books with real turnover. Capital that wants to earn rewards gets a large, quiet home in long-dated longshots. The long tail of props and minor markets serves a third purpose that the data cannot measure directly: a catalog. Our view is that the catalog is where the "markets know" framing breaks down, because most of its prices come from a quote that nobody has hit.

What is the strongest case against our view?

The strongest objection is timing and purpose. Sports props are listed days ahead and may fill near game time; a 05:07 UTC snapshot catches them asleep. And deep two-sided books on longshots may improve accuracy, because a wall of offers at 0.1 cents stops a contract like Kardashian from drifting to 3 cents on a few small buys. On that reading, paid depth is doing its job.

We accept both points in part. Polymarket's reward rules count a pre-game sports order as soon as it rests on the book, so the venue already pays for early quotes, and thousands of props still sat at 3 and 97. A price like that is a placeholder for as long as it lasts. And keeping a 0.05-cent contract at 0.05 cents is a narrow service. The calibration that matters to readers is in the 20-to-80-cent range, and that is where we found the least depth per contract outside a handful of events.

What would change our mind

• A time series of sports props showing spreads under 5 cents in the hour before start for most contracts, which would make our snapshot unrepresentative.
• Evidence that longshot depth stays after holding rewards end on an event, which would weaken the claim that incentives explain it.
• Calibration data showing contracts with spreads above 10 cents resolve in line with their displayed midpoints.
• A comparable pull from Kalshi showing the same shape, which would make this a feature of event contracts generally and less a Polymarket design choice.