Polymarket Liquidity Map: Where Real Depth Lives 2026

12-minute read · By Eli Marsh · Updated May 11, 2026

Polymarket listed 5,418 active markets on May 10, 2026. Only 184 of them — 3.4% — had at least $25,000 of resting depth on both sides at top-of-book during a typical trading hour. Another 612 markets crossed the bar intermittently. The remaining 4,622 markets, 85% of the listing surface, never showed sustained two-sided depth above $5,000 in our 30-day snapshot window. If you copy-trade, the practical map of Polymarket is much smaller than the marketing one. This post measures where the real liquidity lives, when it appears, and what the gap between “listed” depth and “capturable” depth actually costs you on each mirrored fill.

Why this matters before you copy a single trade

Copy-trading on Polymarket is execution-bound, not idea-bound. A leader can have a 9% expected edge on a thesis, but if the market they entered has $400 of resting depth opposite your $250 mirror, the difference between the screen quote and your actual fill eats most of the edge. Across our 30-day execution log, the single largest source of P&L leakage among copy traders was not bad leaders or bad timing — it was mirroring into thin books. So the first question worth answering is not “which leaders are good?” It is “where can my mirror actually fill?”

Methodology — how the depth map was built

For 30 days ending May 10, 2026, we polled the Polymarket order-book WebSocket on a 5-second cadence and captured top-of-book bid, ask, and depth at ±0.5¢, ±1¢, and ±2¢ from mid on every listed market. We computed per-market hourly medians, then categorised each market by Polymarket’s native taxonomy. We used hourly medians (not max) to avoid being fooled by transient depth spikes; the question is “can I rely on this depth being there when I need it,” not “does depth ever appear here.” The methodology mirrors the production pipeline that drives our leaderboard and is documented end-to-end on the methodology page.

A note on what is excluded: markets that resolved during the window without sustained activity (e.g. NBA player-prop markets that exist only on game day) are counted as “active” only during their live window. This avoids overstating the long-tail listing problem.

The headline distribution — how deep is deep

Of the 5,418 active markets across the 30-day window, the depth distribution looks like this:

The shape is what you would expect from a long-tail venue: a small head where almost all of the genuine activity lives, a thick middle of intermittent markets, and an enormous tail of effectively dormant listings. The honest number to internalise is not “5,400 markets” — it is “432 markets where a $250 mirror has a reasonable chance of filling without absorbing 30% of its edge.”

Where the depth lives — by category

The category breakdown is the most actionable view in this study. Aggregating by Polymarket’s native taxonomy:

A few things jump out. Politics carries more aggregate depth than sports despite listing 9× fewer markets. The reason is concentration: a handful of high-stakes binary races (party-control, regional outcomes) pull millions in resting orders, and the same wallet keeps re-loading the book as resolution approaches. Sports is the inverse — thousands of micro-markets, each with a few thousand dollars of depth, churning daily. Both are real businesses, but they require different copy-trade strategies. A leader specialising in politics gives you fewer, higher-quality fills; a sports leader gives you many small fills with thinner per-fill economics.

The second observation is the absence of $25k+ markets in policy politics, culture, climate, and the misc tail. These are categories where Polymarket has the listings but not the resting interest. If a leader in your basket plays these, your mirror will frequently see the screen quote and then watch the spread close before either leg fills.

Liquidity by time of day — the UTC cycle

Depth is not constant through the 24-hour cycle. Plotting top-of-book medians across all 432 actively-traded markets in UTC hourly bins:

Peak depth is the 13:00–17:00 UTC window — the EU/US overlap. Off-peak (03:00–06:00) depth is roughly 60% of peak. This matters for copy traders in a counter-intuitive way: many of the leaders worth following in our dataset trade during the thin-book windows because that is where their statistical edge is largest. If you mirror those leaders blindly, you will fill into the thinnest books in the cycle. The Poly Syncer execution layer applies a depth floor by default for this reason — it lets the leader take the off-peak trade and waits for adequate depth before mirroring, or sizes down proportionally. Mirroring like-for-like into a thin book is the single most common silent loss in our 30-day execution log.

The long-tail problem — 4,622 markets, 17% of aggregate depth

The 4,622 markets in the $1k-or-less tier collectively hold less than 17% of total Polymarket depth, but they account for 85% of the listed surface. From a copy-trader’s perspective these markets are essentially noise — if a leader you follow trades them, your mirror will fill at the wrong end of an inverted spread (we measured median execution shortfall of 4.2¢ per share in this tier vs. 0.42¢ in the deep tier). This is the same execution-shortfall finding from our arbitrage study, generalised: the further you move down the depth curve, the more of the screen edge you give back to spread.

The practical advice is simple but worth saying out loud: filter your copy-trade subscriptions by minimum depth before subscribing, not after. The leaderboard exposes a depth-floor filter that excludes leaders whose primary category is in the long tail; we recommend setting it to $5k for retail-scale copy-trading and to $25k if you mirror at $1,000+ per leg.

Resilient depth vs displayed depth

The most important distinction in this study is between displayed depth (what the order book shows right now) and resilient depth (what the book actually holds when you start to fill). We measured this by sweeping a simulated $500 marketable order through the book and recording how much of the listed depth survived the impact:

Two takeaways. First, the $50k+ tier is the only one where slippage is materially negligible — below 0.5% on a $500 order. Second, the $1k–$5k tier is functionally broken for copy-trading at meaningful size: 43% of a $500 order shows up as slippage, which means more than 40% of the leader’s edge has to be given back just to enter. No leader is good enough to overcome that on a sustained basis.

Persistence — does today’s liquid market stay liquid tomorrow?

We tagged every market in the $25k+ tier on day 1 and checked whether it was still in the same tier on day 30. The persistence rate was 68% — meaning roughly one in three “deep” markets had migrated to a thinner tier by month-end. The migration is largely driven by the underlying event calendar: an NBA conference-winner market thins after the conference resolves, an earnings market thins after the earnings release, a Fed-rate market thins after the FOMC meeting. The implication for copy traders is that your “capturable surface” is a moving window of 100–200 markets at any given time, not a fixed list.

This matters because most copy-trade subscriptions are configured once and then run for weeks. If your filter is “only mirror trades in deep markets” but you never re-evaluate which markets are currently deep, your fill rate degrades silently. Our leaderboard runs the depth-tier classification on a daily cron and updates each leader’s “capturable rate” (the fraction of their last 30 trades that landed in markets you can actually mirror) in real time; the same logic is exposed via the API for self-hosted operators (bot architecture post covers the implementation).

What this means for a copy-trade subscriber

1. Anchor on the head, not the tail. The 184 markets in the $25k+ tier hold most of the genuine edge. Configure your subscription to mirror only these, even if it means following fewer leaders.
2. Match the leader’s typical category to your size. A politics leader is right for $500+ mirror sizes. A sports leader is right for $50–$250 mirror sizes. Mismatching size and category is the most common configuration error we see in the dashboard logs.
3. Set a depth floor in your settings. Default is $5k; consider $10k if you mirror at $500+/leg, $25k at $1,000+/leg. The leaderboard’s capturable-rate column reflects your current floor automatically.
4. Treat the off-peak UTC window with respect. 03:00–06:00 UTC has the thinnest books; if you follow off-peak specialists, the executor’s adaptive sizing is doing real work and is worth keeping on.
5. Re-evaluate every 30 days. A third of deep markets migrate out each month. Set a calendar reminder, or use the leaderboard’s “your basket health” weekly digest (auto-mailed if you opt in on Pro).

The marketing surface of Polymarket is “5,400 active markets.” The trading surface, where copy-trading actually pays, is closer to 200. Knowing which 200 is the difference between a viable strategy and a slow bleed to slippage.

How this connects to the rest of the work

Three earlier posts dovetail with this one. The arbitrage study explains why the screen edge does not equal the realised edge once you account for execution. The resolution-time study explains why annualised return matters more than raw P&L. And the wallet-scoring methodology explains how we fold all three signals — depth-tier exposure, time-to-resolution, and edge-adjusted hit rate — into a single composite rank. Each by itself is incomplete; together they describe an honest map of what is and is not capturable on Polymarket in 2026.

Frequently asked questions

How much depth does a Polymarket market need before I can copy-trade it safely?

For $250 per-leg mirror sizes, $5,000 of sustained top-of-book depth on both sides is the practical floor; below that, slippage absorbs more than 15% of the leader’s edge on average. For $1,000+ per-leg sizes the floor rises to $25,000. Our measured slippage was 0.4% in the $50k+ tier and 43.2% in the $1k–$5k tier — a 100× difference.

Which Polymarket category has the most liquidity right now?

Politics — horse race — carries the most aggregate depth at roughly $12.8 million across 248 active markets in May 2026, despite listing 9× fewer markets than sports. Crypto price markets are second at $4.9 million. Sports has more total markets but much thinner per-market depth, with a median of $3.8k on NBA and $2.9k on soccer.

When are Polymarket order books deepest during the day?

The 13:00–17:00 UTC window — the EU/US overlap — is peak liquidity, with order-book depth roughly 1.3× the 24-hour average. The 03:00–06:00 UTC window is the thinnest, at about 0.78× average. Off-peak windows are where statistical edge is largest for skilled traders but where copy-trade execution risk is highest.

Why are so many Polymarket markets effectively dormant?

Polymarket lists thousands of markets to compete for keyword and event coverage, but professional liquidity providers concentrate on the markets with sustained two-sided interest. As a result, 63% of listed markets carry less than $1,000 of top-of-book depth and are functionally untradable at meaningful retail size. This is normal for a long-tail prediction-market venue.

Does Poly Syncer filter for liquidity automatically?

Yes. The executor applies a default depth floor of $5,000 of sustained two-sided top-of-book before firing a mirror order. The floor is configurable per-user in the dashboard settings, and the leaderboard’s “capturable rate” column shows, for each leader, the fraction of their last 30 trades that would have landed inside your current floor.