Hot right now
A market is here because it moved fast, moved a lot today, is trading several times what it traded on its earlier days this week, is new, or ranked traders bought it. That makes it worth a look; it does not make it an edge. Only markets with a real two-sided price and $10,000 traded today.
Edge board
Every idea being tracked: what is live now, and how much evidence it has. Strongest evidence first
EvidenceIdeaLive nowRecord so far
Every signal that is on
A signal stays here only while its condition holds. Open one to see why it is here, what takes it off, how its kind has done so far, and a prompt to run in Nick.
Who is buying
Buys of $250 or more over the last 7 days by the top 50 wallets on Polymarket's monthly profit leaderboards, on the events followed in full. Sells are left out. A high rank says a wallet made money this month, not that copying it does: the signal built on this is scored on paper first.
Tests
Points per $1 share, on paper
Crypto
Hyperliquid
| Market | Price | 24 h | Funding | 2 h |
|---|
Signal scorecard
Does each kind of signal turn out right? Read on 26 November
| Signal | Logged | Against the closing price | Held to the result | Verdict |
| Scored | Avg pts | ± | Beat it | Settled | Avg pts | ± |
The closing price is Polymarket's price when the game starts. A signal that keeps beating it is seeing something before the market does. The verdict comes after 100 scored signals.
Signals with no closing price
Day moves, funding and pairs: the price is read again 1, 4 and 24 hours later
| Signal | The question | Logged | After 1 hour | After 4 hours | After 24 hours | Drift |
| Scored | Avg | Right | Scored | Avg | Right | Scored | Avg | Right |
A positive average means yes to the question. Day moves are in points; funding and pairs in percent. Measured mid to mid, before any spread or fee. One entry per market per day, so entries never overlap. The last column uses the 24-hour figure and needs 100 scored. It reports a drift before costs, not a profit.
| Time (UTC) | Signal | Market | Signal value | Price then | 1 h | 4 h | 24 h |
|---|
Ledger
Every signal logged, newest first
| Time (UTC) | Signal | Game | Outcome | Signal value | Price then | Closing price | Vs close | Result |
|---|
Fires
The price-move rule fired
| Time (UTC) | Market | Outcome | From | To | Change | Nick workflow |
|---|
Moves
2 points or more in the last 15 minutes
| Market | Outcome | From | To | Change |
|---|
Scoreboard
Paper only. Results in points per $1 share
| Test | What it does | Trades | Won | Avg pts | ± | Running total | Verdict | Read on |
|---|
A verdict needs 100 trades and an average more than twice its ± figure. Before that, a good or bad number is luck. The three "leads" rows are measured before fees, so they can show that a price drifts, not that a trade pays: they are marked before costs and never get a positive verdict.
Bitcoin 5-minute markets
Latest windows. Buy with 3 minutes left, hold to the result
| Window (UTC) | Winner | Up price | Down price | Bitcoin move | Model: chance of Up | With implied vol | Favourite rule | Result | Binance rule | Result |
|---|
Buying No in week-old markets
| Bought (UTC) | Market | Topic | Yes traded at | Paid for No | Bid-ask gap | No price now | Settled | Result |
|---|
From our study of 61,000 settled markets: in week-old markets that were really trading, with Yes between 20c and 60c, Yes settled about 3 points less often than the price paid. This tests it forward. A market is looked at once, when it turns 7 days old. It needs 20 trades and $1,000 in the last 3 days, Yes traded at 20c to 60c and still there now, and a No bid and ask at most 4 points apart. Then $100 of No is bought at the ask on offer, the market's own fee is paid, and it is held to settlement. About two markets a day pass every condition and money is tied up for about a month, so the verdict at 100 settled trades is about three months away.
Who moves first
| Moved first | Times | Other followed | Typical delay | Move held | Buying the slower one: avg pts | ± | Drift |
|---|
A venue moves first when its price jumps 1.5 points in 30 seconds and the other has not moved half as far. The other is given 2 minutes to follow. The two sides of a game move together, so a jump is counted once per game. Buying the slower one is the gain from buying it at the moment of the jump, after its entry spread and before fees.
| Pinnacle moved 1 point or more | Times | Pinnacle first | Polymarket first | Cannot tell | No follow | Polymarket's next 3 min: avg pts | ± |
|---|
Pinnacle is read once a minute, so its move is only known to lie between two readings. A Polymarket move in that same minute counts as cannot tell, never as a lead. Champions League odds are read every 10 minutes and are left out. The last column is what Polymarket did after each Pinnacle move it had not matched.
| Time (UTC) | Game | Outcome | When | Moved first | Jump | Other followed after | Buying the slower one |
|---|
Kalshi against Polymarket
| Started (UTC) | Game | Outcome | When | Cheaper on | Largest gap | Lasted | Size at the price |
|---|
Price after a score
| Time (UTC) | Game | Score | Before | After 5 s | After 30 s | After 60 s | Follow | Fade |
|---|
Buying the winner before settlement
| Bought (UTC) | Game | Outcome | Price | Outcome won | Result | Hours to settle |
|---|
Hyperliquid markets
Coins, and perps on stocks and gold
| Market | Price | 24 h | Bid-ask spread | Funding per hour | Funding per year | Open interest | 24 h volume | Last 2 h | |
|---|
Funding is what holders of long positions pay holders of short positions each hour (negative means shorts pay longs). The yearly figure is that hourly rate held for a year.
Polymarket Perps
| Market | Price | Against Hyperliquid | Funding per hour | Funding per year | Open interest | Long / short accounts | |
|---|
The same assets on Polymarket's own perps exchange. Its Anthropic market takes its reference price from the Entropy perp on Hyperliquid, so the two should sit close together; it is quoted as fully diluted company value until Monday 16 November 2026, then as a price per share.
Pair ratios
A readout, not a trading rule
| Pair | Ratio now | Change, 1 h | Distance from average | Average over |
|---|
Distance is how far the ratio sits from its own average, in standard deviations. Beyond 2 is unusual.