Live

· 5 min read · polymarket · whales

Half of Polymarket's $72 billion runs through 2,900 wallets

We rebuilt Polymarket's entire trading history from the Polygon blockchain — 1.5 billion fills — and the concentration is staggering: the top 0.1% of wallets moved 55% of all volume. To make sure the number was real, we checked our total against Polymarket's reported figures. It matched to within 2%.

By CrowdIntel

Polymarket sells itself on the wisdom of crowds: millions of people, each with a little money and a little information, collectively pricing the future better than the experts. That's the pitch.

We rebuilt the entire thing from the blockchain to see if it's true. Our indexer pulls every Polymarket contract on Polygon into a database — about 1.5 billion fills across 2.9 million wallets, from September 2020 through mid-2026. Then we asked one question: who actually moves the money?

The answer is not a crowd.

A field of 1,000 dots with one lit green — the top 0.1% of Polymarket wallets moved 55% of all trading volume
Each dot is 1 of every 1,000 Polymarket wallets. The single lit one — the top 0.1% — moved 55% of all $72B ever traded. The other 999 split the rest.

On this page

The number

Of the 2,899,788 wallets that have ever traded, the top 0.1% — about 2,900 wallets — moved 55% of all volume. Here's how steep the curve is:

Top 0.1%55%(of all volume (~$39B))Top 10,00069%(0.34% of wallets)Volume Gini0.96(near-total concentration)

Put another way: 1,000 wallets account for 41% of every dollar traded, and the top 10,000 account for more than two-thirds. The bottom 99.9% — roughly 2.9 million accounts — split what's left. The median wallet holds just 9 positions over its entire lifetime.

This is the shape of a venue dominated by a small core of high-frequency, mostly-automated accounts — not a town square of millions of independent forecasters.

We checked it against Polymarket itself

A number this dramatic invites one fair question: is your reconstruction even right?

So we validated it. We summed our independently-indexed volume month by month and compared it to Polymarket's monthly volume as reported by industry trackers:

March 2026 — ours$10.38B(from on-chain data)March 2026 — reported$10.57B(industry trackers)Agreement98%(independent reconstruction)

Our lifetime total comes to about $73.6 billion, scaling exactly the way Polymarket's growth did: ~$8.8B in 2024, ~$21B in 2025, and $43.5B in the first half of 2026 alone. Because our independent count lines up with the publicly reported figures, the concentration above isn't our estimate of the market — it's a slice of the real market.

Why "volume," not "profit"

We're deliberately measuring volume — dollars traded — and not profit. Volume is the single most directly verifiable figure on a blockchain: every buy and sell is a settled, immutable event. Profit requires reconstructing who claimed which payout, and on-chain that attribution is genuinely hard (winnings are often claimed through batch-routing contracts that obscure the end recipient).

So when we say 0.1% of wallets moved half the volume, that's arithmetic on settled trades — not an inference about anyone's intent or P&L. It's the claim that survives any fact-check.

More unequal than any country

The cleanest way to summarize concentration is the Gini coefficient — 0 is perfect equality, 1 is one entity holding everything. National income Gini tops out around 0.63 (South Africa, the most unequal country on Earth, per World Bank data). By the same statistical measure, Polymarket's trading volume scores 0.96 — a different quantity than income, but a striking yardstick for just how concentrated the activity is.

That's not hyperbole; it's the math of a market where a few thousand machines do most of the trading and millions of people place a handful of bets and leave.

See it / try it

All of this is live and free during beta (no card) on CrowdIntel:

  • Look up any wallet — real volume, P&L, win rate, and funding trail: whale leaderboard.
  • Watch the smart money move before the crowd: insider radar.
  • Ask in plain English — hit ⌘K and ask Pulse "who moved the most volume this week?" over the same ledger behind this post.

Get the same data this analysis is built on — free during beta. Start exploring →

Need the raw ledger — every wallet, every fill — piped into your own models or a REST API? That's Terminal: full programmatic access to the 1.5B-fill dataset, for funds and quant desks. Live now, with founding pricing:

Live now · MCP + API + raw ledger

Wire the ledger into your model.

Terminal exposes the CrowdIntel MCP server + a REST API over all 72M trades — the exact tools Claude used in this post, in Claude Desktop, Cursor, or Zed. Founding pricing while it’s new.

Not ready to wire it up? Get founding updates:
● 12 read-only tools● Cancel anytime● Founding pricing

FAQ

How concentrated is trading on Polymarket? Extremely. Of ~2.9 million wallets, the top 0.1% (about 2,900) moved 55% of all volume; the top 10,000 (0.34%) account for 69%. The volume Gini coefficient is 0.96.

How much has been traded on Polymarket? About $72–73.6 billion in secondary-market volume across its history, reconstructed from on-chain Polygon data — a figure that reconciles to within 2% of Polymarket's own reported monthly volume.

How many wallets have traded on Polymarket? Roughly 2.9 million distinct proxy wallets. The median wallet has just 9 positions over its lifetime.

Is the volume real or is it bots? A large share of the top volume appears to be automated market-making rather than directional bettors — the extreme concentration itself (0.1% of wallets moving half the volume) is a signature of machine-driven trading.

How was this measured? From a full on-chain reconstruction of every Polymarket contract on Polygon — ~1.5 billion fills — aggregated per wallet, then validated against Polymarket's own reported volume.

More from the blog

Live
Beta
···v0.1.0
CrowdIntel needs a newer browser

Your device is on iOS 15 or older. CrowdIntel uses features (color-mix, WebGL2, modern auth) that require iOS 16+ / Safari 16+.

iPhone 7 maxes out at iOS 15, so the site can't render here. Open CrowdIntel on a desktop browser or a newer phone.

crowdintel.xyz

◆ legacy browser fallback