Polymarket Whales Data: What We Learned From 15,371 Wallets

We collected 15,371 Polymarket wallets, and 166 passed basic skill filters on 30 September 2026. Copying every position those wallets closed over 120 days gave a large average return, but only because a few 1¢ long shots won. The median copy made about $2 per $100, and three trade rules lifted it to $18. Which wallets were in the group decided whether the total ended up or down.

We do not publish wallet addresses or names. Live pool counts are on our public stats endpoint.

Key findings

1. About 1 in 93 collected wallets passes our default filters. Of 15,371 Polymarket wallets collected, 1,346 were scored in full and 166 passed (30 September 2026).
2. The 166 wallets closed 85,408 positions in the 120 days to 30 September 2026.
3. When every position was copied, the average copy looked large only because of a few winning long shots bought around 1¢. The median copy made about +$2 per $100.
4. Three trade rules (an entry price band, a minimum whale stake and a short market blocklist) kept 9,883 of the 85,408 positions, about one in nine.
5. With the rules, the median copy rose to about +$18 per $100, and 58% of copies ended in profit.
6. Across all 166 wallets, the dollar result with the rules over 120 days stayed close to zero (−1.9%), with gains from a handful of wallets offsetting losses elsewhere.
7. The wallets that looked best on the first 60 days were not reliably the best on the next 60.
8. Speed was not the bottleneck. Across 955 live fills, our copies paid a median of 0.3¢ per share above the whale's price.

How many Polymarket wallets pass a skill filter?

Collected means the wallet is in our pool. Scored in full means we have computed every metric below for it. To pass, a wallet needs all four:

• Return on investment of 15% or more, so it earns something on the money it puts in.
• 30 or more settled trades, so the record is longer than two or three calls.
• Max drawdown of 25% or less, so it has not given back large gains along the way.
• No more than half its profit from one trade, so a single long shot cannot carry the record.

We leave wallets used in our own preset strategies out of the pool.

Why does the Polymarket leaderboard reward luck?

Polymarket's leaderboard ranks wallets by profit or by volume, with Today, Weekly, Monthly and All views. Ranking by dollars favors size and cannot tell a repeatable edge from one big hit. Polymarket's Data API documentation also says the day, week and month boards track the change in marked equity, which moves with unrealized gains on open positions.

One hit can be enough because of how a cheap share pays out. Every Polymarket share pays $1 if its outcome wins and $0 if it loses. A $100 buy at 1¢ gets 10,000 shares. If that outcome wins, the position pays $10,000, a profit of $9,900, which covers 99 losing $100 bets.

This is the case the fourth filter removes: a wallet can top a leaderboard with one such win and still fail the "no more than half the profit from one trade" test.

What happens if you copy every trade a whale makes?

We replayed every position the 166 wallets closed in the 120 days to 30 September 2026, 85,408 in all, as if each one had been copied with $100. Each copy exited when the whale exited, so it earned the same percentage as the whale's own position.

The average result per $100 looked large because a few long shots bought around 1¢ won, the same arithmetic as above. The median copy, the one in the middle, made about +$2 per $100.

A copier cannot bank on an average built on rare long shots. Collecting it would mean copying every long shot the whales bought and paying for all the losers, with money still in the game when the rare winner arrives.

Which trades did the three rules remove?

We then applied three rules to the same 85,408 positions:

1. Entry price between 0.30 and 0.98.
2. The whale's own stake of $1,000 or more.
3. A blocklist of some sports markets (NBA, MLB, NHL, EPL, UCL and a few more).

Each position is counted under one reason, so the rows add up to the total (shares are rounded). The stake rule removed the most positions, with the price band close behind. By count, the market blocklist mattered least.

Did the trade rules make copying profitable?

They improved the typical copy. With the rules, the median copy made about +$18 per $100, up from about +$2, and 58% of copies ended in profit.

The group as a whole still did not end in profit. Across all 166 wallets, the dollar result over 120 days was close to zero (−1.9%), and gains from a handful of wallets offset losses elsewhere.

The gap between a positive median and a flat total comes from the payoff shape. A share bought at 80¢ gains 20¢ if it wins and can lose the full 80¢ if it does not. Per $100, that is +$25 against up to −$100. A copier can win most trades and still end flat if the losses are larger than the wins.

Which Polymarket wallets did best?

These four were the strongest of the 166 wallets over the 120 days, so they are not typical. Each was copied at $100 per trade with the three rules, exiting when the whale exited.

All four specialize in one or two sports or esports. Whale D was in profit on 91% of copies yet made the least per copy, because a high win rate built on small wins pays little. Whale C made the most money with a modest edge per copy because it traded far more often than the others.

The four wallets were picked after the fact, which flatters their results. Past results do not predict future ones. The four cards are also shown on our homepage under proof.

Does past performance pick next period's best wallets?

Not reliably. We split the 120 days in half, picked the best wallets on the first 60 days and checked them on the next 60. The first-half leaders did not reliably stay at the top.

For a copier this means:

• Score wallets again on a regular schedule.
• Prefer wallets with large samples over wallets with high but short records.
• Spread copies across several wallets that pass.
• Drop a wallet when it stops passing the filters.

Does copy speed matter as much as wallet choice?

No. Measured on 1 October 2026 over the previous 60 days, a whale's trade reached our signal in about 46 seconds (median), and a live order filled about 3.1 seconds after that (median, 77% under 5 seconds). Across 955 live fills we paid a median of 0.3¢ per share more than the whale did.

On a 60¢ share, 0.3¢ is 0.5% of the price. The trade rules took the median copy from +$2 to +$18, and which wallets were in the group decided whether the total ended up or down. Picking wallets mattered more than speed or volume.

How we measured: method and limits

Pool counts come from Whalidate's public stats endpoint on 30 September 2026. The positions are every position the 166 passing wallets closed in the 120 days to 30 September 2026, plus resolved positions not yet redeemed, all from Polymarket's public data. The copy model puts $100 into each position that passes the rules and exits when the whale exits.

We did not model slippage, a separate fee charge, stop-loss, take profit or days to resolution. The data has no entry time, so the days-to-resolution rule could not be replayed. The four strongest wallets were chosen by their result in the same window they are measured on. Wallet addresses and names are withheld.

What this means for Polymarket copy traders

Filter trades first, because the average from copying everything depends on a few long shots winning. Then spend most of your effort on which wallets you copy, and keep checking them. Our step-by-step guide to copying Polymarket whales turns these findings into settings, and what a Polymarket whale is covers the basics. To test the rules yourself without money at risk, open the app and start on paper, or read how we pick wallets.