Methodology

How we measure whales

What we record

Two recorders on our own server watch Polymarket and Limitless trades around the clock. On Polymarket we keep every trade of $1,000+; on Limitless every trade of $50+. A whale, on this site, is any wallet placing trades that size. Records run Polymarket since Aug 20, 2026, Limitless since Sep 5, 2026.

The live feed on the home page shows the bigger ones: $5.0K+ on Polymarket and $50+ on Limitless. Fills by the same wallet on the same side of the same outcome, each within 2 minutes of the last, are combined into one bet.

Settled bets and profit

A bet counts once its market has resolved: closed, with a winning outcome priced at 99¢ or more. Profit is what the position paid at resolution, plus anything sold before, minus what it cost. We add it up in two-week periods, which is what the bars on the Victors and Fallen pages show. Return is profit divided by the money put in.

So far 92.7% of 115,424 Polymarket positions and 99.7% of 16,092 Limitless positions have resolved. The rest wait, and join the record when their markets close.

Win rate and its range

Win rate is the share of a wallet's settled bets that paid. A small sample can't pin it down, so every win rate comes with a 95% range (a Wilson interval): the top Victor on Polymarket won 55% of 60 bets, and its true rate could be anywhere from 43% to 67%. A high win rate isn't skill either: buying favourites wins often and can still lose money.

The Victors board only ranks wallets with at least 20 settled bets. The Fallen are wallets that were up at least $10K (Polymarket) or $1.0K (Limitless) at their peak, then gave back at least half.

Skill or luck?

A bet bought at 30¢ should win about 30% of the time. For each wallet we take the bets it held to the end and compare the wins it got with the wins its prices implied. Prices are first calibrated to how often bets at that price actually won across all whales, which strips out the market-wide favourite–longshot bias, so what's left is the wallet's own edge.

The gap is measured in standard deviations (a z-score):

• Beat the odds* above +1.96: it won more often than its odds implied.
• Worse than the odds below −1.96: it won less often.
• Can't tell from luck in between.
• Too few bets fewer than 10 bets held to the end.

By chance alone about 1 wallet in 40 (2.5%) clears +1.96, so the badge is a flag, not proof. So far 3.2% of 1,343 Polymarket whales tested and 4.0% of 149 Limitless whales tested beat the odds, close to what luck alone produces.

The copy-trading test

We pick the top 20% of whales by return on earlier weeks, then copy their next trades, entries and exits, off the order book as it stood moments after each trade, and score the copier against the whale on the same trades. The whales are picked before the period they're copied in, as a real follower would have to. Full results: Does copying whales work?

Do winners keep winning?

For each pair of back-to-back two-week periods we check three things: whether a whale's rank in one predicts its rank in the next, whether whales in profit stay in profit more often than whales in general, and whether the top 20% stay in the top 20%. Each is compared with a shuffled baseline, and checked again with the markets a wallet traded in both periods removed, so one shared result can't fake a streak. The results are on the copy-trading page and in the FAQ.

Wash trades

Some trades are priced far from fair value: someone paying close to even odds for an outcome the price feed says is already decided. That's a sign of a wash trade or a value transfer between wallets, not a real bet. On Limitless crypto markets we check every trade against Binance prices, and exclude the 91 flagged trades ($187K) from every Limitless number. On Polymarket, trades far from the order book are reported (1,670 so far) but not excluded, because the book is captured a few minutes after the trade. Flagged trades are marked on wallet pages.

When it updates

• The live feed: every 5 minutes.
• Records, boards, the Hopium Index and the copy test: every Monday at 03:00 UTC. Each page names the weekly run its numbers came from in “About this data” at the bottom.

Limits

• Polymarket trades under $1,000+ aren't recorded, so a wallet's record covers only its big trades.
• Limitless prices are recorded since Sep 5, so Limitless records are shorter.
• The latest two-week period has the most markets still waiting to resolve; its numbers move the most between updates.
• Wallets are addresses. One person can run several, and we can't see that.

Every number on the site comes from the raw trades, through the weekly run or the 5-minute feed: nothing is rounded up, estimated or typed in by hand. If a dataset is missing, the page says so instead of showing an old number. The data is free to reuse: see About.