Forum claims, checked against the receipts

“Late money is the strongest signal”

Grouped by how long before the market's start or close the bet was placed. Bets placed in the final hour are left out of our main scoreboard because the outcome is often already visible, as in a game that has started.

“Fresh wallets on long shots know something”

Fresh means the wallet's first Polymarket trade was less than 7 days before the bet. Long shot means an entry under 30¢. Wallets whose age we haven't looked up yet are left out.

“Just copy the proven wallets”

Each bet re-priced as if you had bought the same side 1, 6 or 24 hours after the whale. A row only counts bets whose market was still trading then. Bets where the quoted price before the trade was more than 5¢ from the whale's fill are left out (thin books). Prices are hourly, so moves within minutes of the trade (and your own fill) are not captured; real copies pay a bit more. The first table is the honest one: wallets picked using only bets that resolved before Jun 10, 20:30 EDT, measured only on bets they placed after it.

Only 44 bets so far, and ±6.3 pts of noise: too few to say copying these wallets beats the price. The price itself barely moves after entry; the lag costs about 0.4¢ at 1 hour.

An independent study by 0xinsider on 1,585 large Polymarket buys (Sept 10 to Oct 9, 2026) found a similar day-late cost of about 1.1¢, and that 96% of large sports trades settled within a day. Their write-up ↗

“Copying works in politics, not sports”

How far the price moved against a late copier, by category, across every re-priced bet. Positive means you'd pay more than the whale did. Many sports bets drop out of the later rows because the game was already over.

The same test, minute by minute

Hourly prices can hide fast moves, so here are 1, 5 and 15 minutes after the whale, using minute prices. Minute prices are only kept for a few days upstream, so this table covers live flags from the last few days and grows every 15 minutes. "Price moved" is what a copier pays above the whale's own price, including the whale's own market impact.

Live $10k+ single-fill flags from the last few days, resolved or not. Markets whose quoted price just before the trade was more than 5¢ from the whale's fill are left out: there the quote isn't a price anyone could trade at. Quotes are midpoints, so a real copy also pays half the spread. Average price change on the whale's side; ± is one standard error.

“Bigger bets are better informed”

Grouped by the dollar size of the bet.

“Whales betting against the favorite are onto something”

Grouped by entry price. An underdog bet at 30¢ only needs to win 30% of the time to break even.

“Whales are sharper in some categories”

The same test, split by market category. When five groups are tested, one can clear 2 standard errors by chance, so treat a single "beats the price" here with care.