The 100 biggest Polymarket wallets we grade trade half the volume. None of them is a copy candidate.
The 100 largest wallets CopyGrade grades traded $778M in the last 60 days, 49.6% of the volume of all 13,976 graded wallets. 87 of them carry a farming-risk flag, 53 of them severe. Their median realistic edge is −18.05%. Not one has a positive copy verdict. They took 71% of the graded set's 60-day profit, and 97% of that went to the flagged wallets among them.
The pattern is a gradient, not a cliff. Sort the graded set by 60-day volume and the flag rate climbs at every tier: 28% below $10k, 64% from $10k to $100k, 77% from $100k to $1M, 83% above $1M. The 172 wallets that clear the full copyable bar sit mostly at the bottom: 134 of them trade under $10k in 60 days. Their median 60-day volume is $1,376, and between them they trade 0.24% of the volume.
All figures are a point-in-time pull from CopyGrade's live scored set via the read-only database, dated 2026-10-23. "60-day volume" is traded volume (size × price) of a wallet's fills in the trailing 60 days of its tracked tape. "Real edge" is our post-fee, post-latency estimate of what a copier would have kept, per the published methodology. The set is the wallets CopyGrade covers, not every wallet on Polymarket. Its discovery passes start from leaderboards and the largest markets, so it likely over-represents the large, active wallets this post is about.
How concentrated is Polymarket volume?
| Graded wallets by 60-day volume, 2026-10-23 | Wallets | Share of 60-day volume | Entry threshold (60-day volume) |
|---|---|---|---|
| The largest wallet | 1 | 3.6% | — |
| Top 100 | 100 | 49.6% | $3.32M |
| Top 1% | 140 | 57.1% | — |
| Top 1,000 | 1,000 | 96.5% | $134k |
| Top 10% | 1,398 | 98.6% | — |
| All graded wallets | 13,976 | 100% ($1.57B) | — |
One hundred wallets carry half the graded set's trading, and a thousand carry all but 3.5% of it. The other 12,976 graded wallets, nine-tenths of the set, trade the remaining sliver.
What do the 100 biggest look like?
| Top 100 by 60-day volume, 2026-10-23 | |
|---|---|
| Farming-risk flag (any level) | 87 |
| Severe flag | 53 |
| Pass the farming check | 13 |
| Pass the farming check and positive real edge | 1 |
| Negative real edge | 70 |
| Median real edge | −18.05% |
| Median CopyGrade Score | 30 |
| Verdicts: negative / caution / positive | 64 / 36 / 0 |
| Clear the full copyable bar | 0 |
| Profitable over the last 60 days | 43 |
| Combined 60-day realized profit | +$31.5M |
| …of which, flagged wallets | +$30.6M |
| Median trades in the scored 90-day window | 1,607 |
| Median distinct markets traded | 624 |
The 100 largest wallets made money over 60 days, $31.5M between them, but 43 of the 100 were profitable, and the flagged 87 took almost all of it. The profit is real on their books. The question a copier asks is different: what would a wallet that mirrored them have kept? At a median realistic edge of −18%, the model's answer is that a copy loses, because the record is not one a follower can reproduce at the leader's prices.
The one wallet in the hundred that passes the farming check with a positive realistic edge still fails the bar elsewhere. That is why the "copyable" row reads zero.
What the detectors raised
| Signal kind, top 100 by 60-day volume | Wallets with at least one signal |
|---|---|
| Manufactured-volume churn | 55 |
| Pump-and-dump | 55 |
| Iceberg accumulation | 42 |
| Self-trade wash | 26 |
| Copy-bait | 19 |
| YES/NO stealth merge | 19 |
| Decoy wallet cluster | 4 |
These are the names of our detectors. Each count is wallets on which that detector fired at least once. A wallet can carry several kinds. The anatomy post explains what each one reads in the public trade record. The two most common, volume churn and pump-and-dump, both describe a record whose volume and price action a follower would be buying into, not alongside.
The gradient: flags rise with size
| Graded wallets by 60-day volume, 2026-10-23 | Wallets | Flagged | Severe | Median real edge | Median score | Positive verdicts | Copyable |
|---|---|---|---|---|---|---|---|
| $1M and above | 357 | 82.9% | 56.0% | −15.90% | 30 | 0 | 0 |
| $100k – $1M | 766 | 76.9% | 57.6% | −13.25% | 30 | 7 | 7 |
| $10k – $100k | 942 | 64.5% | 49.0% | −5.20% | 30 | 31 | 31 |
| Under $10k | 11,911 | 28.0% | 18.5% | −1.00% | 48 | 134 | 134 |
| All graded | 13,976 | 34.6% | 23.7% | −1.40% | 48 | 172 | 172 |
Part of the gradient is mechanical, and we say so: a wallet with more fills gives the detectors more to read, and a thin record can pass the check simply because there is little in it. Below $10k the median wallet has six trades in its scored window, too few for the copyable bar in most cases. Still, the direction is the finding. In the graded set, size buys detector exposure faster than it buys realistic edge. The median realistic edge is −1.0% at the bottom tier and −15.9% at the top.
Where the copyable wallets are
| The 172 wallets clearing the full copyable bar, 2026-10-23 | |
|---|---|
| Median 60-day volume | $1,376 |
| 90th-percentile 60-day volume | $46,274 |
| Above $100k of 60-day volume | 7 |
| Above $10k of 60-day volume | 38 |
| Combined 60-day volume | $3.8M |
| Share of the graded set's 60-day volume | 0.24% |
The wallets CopyGrade's model would put on a shortlist trade a quarter of one percent of the volume. Even the most active of them sits far below the $3.3M of 60-day volume it takes to enter the top 100. Being small is not a virtue in itself, and many of these records are short. It does mean that a "follow the whales" feed and a vetted shortlist barely overlap.
How the biggest wallets trade
The 86 top-100 wallets that traded between October 9 and October 23 placed 112,190 fills in those two weeks, about 1,300 each. The fills were not large. The median was $165, 42% were under $100, and only 3.1% were $10,000 or more. They traded $212M of notional in the fortnight.
So a "whale" in this data is mostly a high-frequency wallet, not a wallet placing a few big bets. To mirror one, a copy-bot would place on the order of a hundred fills a day. It pays the taker fee on every one: 0.05 × p × (1 − p) per share on sports, 2.5% of the notional per leg at 50¢. It arrives behind every price the leader moved.
Their markets are overwhelmingly sports. About 96% of the top 100's category-attributed volume is in sports markets. Seventy-four of them qualify for the soccer board, 61 for tennis, and the median wallet sits on six of our 26 boards. The single wallet that tops all three soccer boards by profit is one of them.
Are they on Polymarket's own leaderboards?
Mostly not. Of the top 100 by our 60-day volume, 23 appear in the top 50 of Polymarket's 30-day volume leaderboard in our 2026-10-23 capture. 17 appear in the top 50 of its 30-day profit leaderboard. A public leaderboard shows a few dozen names per window. It is a different list from "who trades the most", and both are different from "who is worth copying". Our leaderboard post and the leaderboard-versus-score guide cover that gap.
What this means for copying a whale
Size measures activity, not skill. In the graded set, the biggest wallets are the most likely to carry a farming-risk flag, the least likely to have a realistic edge a follower could keep, and the hardest to mirror at their own prices. None of the top 100 clears the copyable bar. Our guide to copying a Polymarket whale sets out the checks that separate the rare large wallet worth reading from the rest. The vetting checklist applies to every wallet, whatever its size.
A farming-risk flag is an algorithmic risk assessment, not an accusation of conduct, and a positive verdict is not a recommendation. Our own validation of whether the score predicts copier returns is still null. CopyGrade is analysis-only. It never executes trades, holds funds, or custodies keys. Not financial advice.