See what the sharpest
Polymarket wallets are
loading up on.
Six independent signal engines watch every active Polymarket market every two minutes. Each alert includes the exact Kelly bet size, streak wallet co-occurrence, and the edge vs the crowd price. News catalysts fire before the whales reprice. You decide and execute, or just watch.
Edge and hit rate from a 2-year point-in-time backtest. Streak wallet win rate from a separate verified deep backtest. Not financial advice.
Three things happening right now while you look elsewhere.
The sharpest wallets are loading up right now.
Top Polymarket wallets don't post their trades. By the time a move surfaces in discussion threads, the edge is already priced in. The window to act is measured in minutes, not hours.
Knowing the direction is only half the edge.
Without the Kelly formula, you're sizing by gut. Too small and you leave real money on the table. Too large and one wrong market undoes the whole run. The math exists -- most tools ignore it.
An entry without an exit is still a guess.
Most signal tools fire once and disappear. You got in on the alert. Now the market has moved. Is it a winner to hold or a loser to cut? Without an exit signal, you're back to guessing.
One clean text. The whole story.
It watches the whole board
Every active Polymarket market, polled every two minutes. Six independent signal engines score activity in real time. Nothing about your wallet is ever touched.
It scores, fuses, and sizes
Each signal gets a 0 to 100 score with the crowd's price next to the model's estimate. When sources agree, they fuse into a COMBO alert with a boosted score. Every directional alert includes half-Kelly sizing: exact dollar amounts for three bankroll sizes so you know what to bet, not just that something is happening.
It texts you entry and exit
A clean Telegram message lands on your phone with the market, the side, the score, the edge, and the Kelly size. When a streak wallet agrees, the alert says so. When the market later drifts ten-plus points, an exit alert follows. You decide and execute, or just watch.
The smart-money signal beat the market price.
We reconstructed the actual accumulation signal across 2 years of history. 289 proven wallets, 541,289 fills examined, settled against Polymarket's gold-standard outcome flag. This is the honest, point-in-time number, with each wallet required to prove itself first and zero look-ahead.
Net ROI +8.9% after costs on the qualified set. The optimistic, survivorship-inflated number is higher. We publish the conservative one on purpose. Past results do not guarantee future outcomes.
Five wallets. 84.9% combined win rate. 96.4% when they agree.
A separate verified deep backtest identified five specific Polymarket wallets with statistically anomalous win rates, settled against Polymarket's gold-standard resolved outcome flag. When two or more of them hold the same side as a signal, the win rate in the backtest was 96.4%. That co-occurrence is flagged in every alert.
Deep point-in-time backtest, 2026-06. Resolved against Polymarket's gold-standard outcome flag. Past results do not guarantee future outcomes.
Proven forward, in public.
The backtest is history. This is the live bot's own calls, resolved against Polymarket settlement as they close. It starts small on purpose. Nothing here is back-dated.
Links open the real Polymarket market. Arbitrage signals are outcome-independent and do not count toward the hit rate. Not advice.
Not one signal. Six, all watching at once.
Each engine is independent. When two or more agree on the same market and side, they fuse into a high-conviction COMBO alert with a boosted score.
Top wallets, freshly loading the same side
Tracks the leaderboard's sharpest wallets and fires the moment three or more of them buy the same market and side inside a short window. Snapshot diffing isolates fresh accumulation, not stale positions. Polled every two minutes. This is the headline signal.
84.9% win rate. 96.4% when two or more agree.
Five wallets with statistically anomalous win rates, verified in a deep point-in-time backtest. When any of them holds the same side as a signal, the alert is flagged and the score is boosted. When two or more agree, the win rate in the backtest was 96.4%. The most powerful combo signal.
Fresh headlines matched to live markets
Polls Reuters, BBC, and the New York Times RSS feeds every cycle. When a headline published in the last five minutes shares meaningful keywords with an active Polymarket market, a [CATALYST] alert fires. The window is designed to beat the whales to the reprice, not follow them.
Outcome-independent setups under $1
Binary markets where buying both sides costs less than a dollar, and multi-outcome events where every Yes ask sums to under one. A stale-book guard rejects fake arbs before they ever reach your phone.
Lopsided capital from the biggest holders
Surfaces markets where the largest holders' money is stacked heavily on one side, so you see conviction-weighted positioning instead of raw share counts.
Follow-up when price drifts from your entry
Every signal the bot fires becomes an open entry to track. When the market price moves ten or more percentage points from the entry price, a follow-up alert arrives. Entry and exit, not just entry.
Plug in or own it. You choose.
The signal feed delivers alerts directly to your Telegram with nothing to install. The source gives you the full engine to run yourself.
Signal Feed
Hosted. Alerts straight to you.
- All six signal engines: smart money, streak wallets, news catalyst, arbitrage, whale concentration, and exit alerts
- Half-Kelly sizing in every directional alert -- exact dollar amounts for three bankroll sizes
- Streak wallet co-occurrence flag with win-rate badge when two or more agree
- Alerts land directly in your Telegram. Zero setup, no server, no code.
- Cancel any time
Cancel any time. No lock-in.
Full Source
Self-hosted. Own the engine.
- Full Python source: API client, all six signal engines, fair-value model, and half-Kelly bet sizing
- Streak wallet backtest: five verified wallets with anomalous win rates. 84.9% combined win rate, 96.4% when two or more agree.
- One-prompt Claude Code setup guide -- live in about five minutes
- config.yaml to tune every threshold without touching code
- Two ready-made bot avatar images
- Single-seat commercial license and full disclaimer
First first 50 buyers. Goes to $35 after.
7-day watch-it-work guarantee
Run it for a week. Watch the alerts land. If the radar is not what you wanted, email for a full refund, no questions asked. Honest products with refunds is how this should work.