Whale Lab

Only whales rated Worth following — not Promising, not Risky. A day is an India-time day (00:00–24:00 IST). Every buy of one outcome on one day is one entry, at its average price; smaller than $100 is left out as noise. A later sell closes the oldest entry first.

Won / lost counts only entries that are finished: sold, or the market settled. Open ones are shown at the last price and never counted. Parlays cannot be priced, so they are listed but not scored.

Price baseline is the chance the market itself gave: someone with no skill who pays 80¢ still wins about 80% of the time (a sell before the end counts as a coin flip). A whale only shows skill where the win rate beats it.

Per $100 copied uses the whale's own price and follows their exit, with taker fees charged. It is a ceiling: a real copy arrives later and pays more. Today and the two days before keep updating as markets settle; older days stay as last written.

The signal. When a whale rated Worth following puts $100 or more into one outcome in one India-time day, the fund buys $10 of the same outcome at the same price, in the same second. Every bet is $10, win or lose, whatever the fund is worth. If the fund already holds that outcome (another whale got there first), the signal is skipped.

Managing it. The fund follows the whale it copied: when that whale sells part of what it bought, the fund sells the same share at the same price. Otherwise it holds to the end and is paid $1 or $0 a share when the market settles, and that cash is ready for the next signal. Taker fees are charged on every buy and sell.

A, B and C. Fund A copies every signal. Fund B copies only markets scheduled to end within 2 days, so its money comes back and gets used again. Fund C copies every signal except crypto and esports markets (CS:GO, Valorant and other esports). C opened on 3 Oct 2026, a day after A and B, so read it from its own start date. A bet that finds no cash is counted as missed.

How real it is. Paper only: nothing is traded. Each update (about every 15 minutes) replays the whales' trades since the last one, in time order and to the second, with the whale list as it stood at the start of each stretch, so it trades exactly as a bot watching live would have. "Being copied right now" previews the trades since the last update. The same-second price is the best case; each bet is also scored as if copied a minute late, and that number is closer to what a person acting on an alert gets. The funds opened empty on 2 Oct 2026, with no backfill.

What it cannot see. Only the buys it watched: if a whale already held the outcome, a sale counts against the part the fund copied. A whale leaving by merging both sides instead of selling is not seen, and the copy holds to the end.

D, E and F: real prices. A, B and C buy and sell at the whale's own price, the best case. D, E and F use the same rules (D follows A's, E follows B's, F follows C's), but only on the signals the recorder on our server caught. That is most, not all, of A's, B's and C's signals, and each fund's page shows how many. Every buy and sell is priced from the live order book at the moment the recorder heard the whale's trade (about 0.1–0.2 s after it): $10 walked through the real sell orders, sales through the real buy orders. Each card also shows the same copies at the whale's own price, so the gap is the real cost of copying. No fill price is estimated: a copy the recorder could not price is left out. Fees are each market's own Polymarket fee (most sports have none; 3-7% x price x (1 - price) elsewhere), checked against the fees actually charged on the copied whale trades in Polygon's public record. Bets still open are valued at what they would sell for right now, to the buy orders really waiting, after the fee. They are rebuilt about every 15 minutes while their publisher runs (each page says when it last ran), and they start on 4 Oct 2026 at 22:35 India time, when the server's recorder began.

G: the forward test. G copies A's signals at real prices, like D, but skips any bet where the whale paid under 20¢. Our study of every copy from 21 Sep to 5 Oct kept only that one rule: no topic and no time of day paid once the real copy cost was counted. G starts on 5 Oct 2026 at 11:00 India time, and only bets from then count. Its page sets out the verdict, which was fixed before G opened, and a table that keeps testing every other idea the study looked at on new bets.

Ranked by realized profit within the selected topic, among screened whales. Each whale's most recent settled positions — sold, redeemed, or resolved at $0 and never redeemed — up to 200. A recent sample, not lifetime topic performance, and not picked by result. Losses on positions still open are not included. Topic labels are inferred from market text.

At least 20 distinct settled markets, positive sampled profit, results across at least 3 months, 60% of those months positive, and no single market providing over 35% of gross winning profit. A specialist also has at least 60% of sampled cost in this topic. These are screening rules, not proof of future skill.

Net share changes between successive observations. While Research or Holdings is open, live API checks rotate through whales and compare successful responses. Browser observations are session-only; saved history comes from the pipeline. Adds and reductions describe inventory, not intent; missing positions never prove an exit.

Followability checks whether $25 can still be bought within 2¢ of the whale’s fill, one minute after this browser sees a fresh live buy. Only followed whales are sampled, at most once per whale every 5 minutes. Keep this page open. The score requires 10 measured checks; failed checks are reported separately. Displayed depth is not a guaranteed fill, and fees are excluded. Observation spans are not exact holding times: gaps can hide close/reopen cycles.

Up to 40 open markets per window, refreshed straight from the market API while this view is open. Windows are cut in UTC: "end of today" is UTC midnight, which is 5:30 AM IST. A whale bet shown on a card is an observed fill, not a recommendation.

Connecting to the market API…

While your pointer is on the list, you've just touched it, or you're scrolled into it, new bets wait in a banner instead of pushing rows around — so a tap can't land on the wrong bet. Tap a name for their record, or ⚡ Copy this to track that you took the same bet. Nothing is ever traded for you.

Refreshed from the positions API while this view is open. All-whale mode rotates through the list a few whales at a time; pick one whale for focused updates. Holdings are capped samples, not every position, and markets that have already resolved are left out. If the API is unreachable, saved rows keep their original timestamps.

Connecting to the positions API…