CrowdIntel· 3 min read · polymarket · profitability
The median Polymarket winner has made $14. The top 50 took $291 million.
We randomly sampled 800 wallets and asked Polymarket's own profit API for each one's all-time PnL. 71% are net losers, the median account is down $2.62, and the median winning account ever made $14.38. Meanwhile the top 50 accounts on Polymarket's own leaderboard took a combined $291M.
By CrowdIntel
There's a number Polymarket's marketing will never show you, and it isn't the loser rate. It's what winning looks like.
We took a uniform random sample of 800 wallets out of the 2.9 million that have ever traded on Polymarket — drawn from our full on-chain reconstruction, so no cherry-picking — and asked Polymarket's own profit API for each one's all-time PnL. Not our ledger, not an estimate. Their number, for a statistically representative crowd.
The crowd, by Polymarket's own math
| Finding | Value |
|---|---|
| Net losers | 71.2% (95% CI 67.9–74.3%) |
| Median account, all-time | −$2.62 |
| Median winner, all-time | +$14.38 |
| Accounts that ever made >$1,000 | ~3% |
| Accounts that ever made >$100 | ~7.5% |
Read that middle row again. Even if you win on Polymarket — and 7 in 10 accounts don't — the median winning career nets you fourteen dollars.
This lines up with our earlier realized-cashflow analysis, which found only ~2% of traders ever cleared $1,000. Two independent methods, one from on-chain cashflows and one from Polymarket's own API, land on the same picture.
Scale doesn't save you
The obvious objection: "sure, tourists lose — serious traders win." Their API says otherwise:
| Lifetime volume | Net losers | Median PnL |
|---|---|---|
| Under $100 | 75.9% | −$1.60 |
| $100–10K | 70.2% | −$2.60 |
| $10K–1M | 65.0% | −$10.15 |
Trading a hundred times more money buys you a six-point better loss rate — and a median result that's four times worse. Bigger accounts don't lose less. They lose more, slower.
Meanwhile, at the top
Polymarket also publishes an all-time profit leaderboard. Same API family, same source of truth:
- Top 10 winners: $118.5M combined — Theo4 ($22.1M), Fredi9999 ($16.6M), swisstony ($13.4M), kch123 ($11.4M), RN1 ($10.5M)…
- Top 50 winners: $291.3M combined
- Entry ticket to the top 50: $2.67M — roughly 185,000× the median winner's lifetime profit
And the money itself flows through a funnel just as narrow: the top 0.1% of accounts move 55% of all volume, and in 2026 most of the volume is machines anyway.
What this means
Polymarket's pitch is that the crowd is smart, and being right pays. Their own numbers describe something else: a market where the crowd is the product — 71% of it net-negative, its median winner up $14 — and a microscopic elite of a few dozen accounts collects hundreds of millions.
That's not an accusation. Casinos are legal and lotteries are popular. But nobody calls the lottery the wisdom of crowds.
Methodology
Sample: uniform random draw of 800 proxy wallets (ORDER BY cityHash64(proxy_wallet) — deterministic and reproducible) from all 2.9M wallets in our full Polygon reconstruction of Polymarket (1.52B fills, volume reconciled to Polymarket's reported figures within 2%). Each wallet's all-time PnL fetched from lb-api.polymarket.com/profit; 775 of 800 resolved, 11 had no Polymarket record, 14 unresolved — too few to move the headline outside its confidence interval. Leaderboard figures from the same API, window=all, retrieved 2026-07-01. Raw survey results are archived and every number in this post is machine-checked against live sources before publication.
CrowdIntel runs the only independent full-history reconstruction of Polymarket's on-chain activity — the same data behind our public API. Check our math.