Ranked signals
all bets, highest conviction first · last 48hTracked specialists
top wallets we follow
Signal history
every signal logged · 0 total
Research
how the signals are scored and how they would have done · three lenses: Backtest, Edge, ConvictionWhat this tab is Everything here is backward-looking and informational — it explains the methodology and shows how resolved bets played out. Nothing here places a trade, and none of it is a prediction. Three lenses below: Backtest (what a flat stake would have returned), Edge (entry-price vs proven win rate), and Conviction (what the 0–100 score means).
1 · Backtest
$100 per bet on resolved signals only
Definition. Take every signal whose market has already resolved, pretend you
staked a flat $100 on each, and tally the result. A win pays
$100 ÷ entry price; a void (disputed / 50-50) refunds the stake.
Read this first These returns are idealized and optimistic. They assume you filled at the whale’s entry price (you’d realistically enter later, at a worse price), with no fees or slippage. Only markets that have already resolved are included — most live signals are excluded. Small, recent, top-whale sample: past performance, not a prediction.
Consensus depth × conviction (same flat $100/bet · win rate & ROI over resolved bets · alert volume over all tracked bets)
2 · Edge / ROI sweep
entry-price band × edge threshold · resolved bets only · proven whales (≥20)Definition. “Edge” = a proven whale’s current market-level win rate minus the price they paid to enter. A positive edge means they tend to win more often than the entry price implied. This sweep buckets resolved bets by entry-price band and edge threshold to find which filter rules historically cleared a 10% modeled-fill ROI.
Read this first Headline ROI is modeled fill (you enter 2¢ worse than the whale) — still optimistic vs reality, and idealized is shown only for contrast. Only resolved bets from proven whales count, so the sample is small. A cell is a candidate rule only at n≥20 and ≥10% modeled ROI; the Wilson lower bound guards lucky small samples. Nothing here places a trade.
3 · What “conviction” means
a 0–100 score — how much a proven whale is backing this betDefinition. Conviction is not a win probability and not advice. It answers one question: “how seriously should I take this signal?” The biggest input is whether the whale has actually won before — a huge bet from someone with no track record no longer scores high.
Honest limits The win rate is market-level settlement (which outcome the market resolved to) — publicly knowable and honest — not a per-wallet profit figure, which the free Polymarket data can’t report truthfully. A whale tagged unproven simply hasn’t had 20 of their tracked bets resolve yet. Informational only — nothing here places a trade.
Crypto whales
top crypto-leaderboard wallets · their open (unresolved) positionsHeads up These are the highest-PnL wallets from Polymarket’s crypto leaderboard. “Crypto” is just the board we sourced them from — their open bets can be on any topic, not only crypto. Realized PnL is the wallet’s lifetime figure from the leaderboard; it is not a claim about these specific open bets. Nothing here places a trade.
Predictions
market-implied views from Polymarket · aggregate vs proven-whaleRead this first These are market-implied readings, not forecasts. The aggregate view reads the current Polymarket prices as probabilities; the whale view re-weights using only proven whales’ (≥20 resolved) entry prices, and falls back to all tracked whales when the proven sample is thin. Prices move constantly and samples are small — informational only, nothing here places a trade.