Edge Score

Most whale trackers sell “copy the whales”. Copying has a cost they leave out: by the time you see an alert and act, the price has often moved. Edge Score answers the question a follower actually has: had you copied this trader’s entries a minute after they made them, would you have made money, and did it hold up beyond the month that put them on the leaderboard?

You see it on a wallet’s card in /list and next to traders in /leaders.

The labels

Label	What it means
🟢 Copyable	Copying each entry 60 seconds later returned more than +2% on average on entries older than 30 days, and stayed positive over the last 30 days
🟡 Mixed	Positive overall, but not by enough before the last 30 days, or not in the last 30 days
🔴 Not copyable	Copying 60 seconds later lost money on average
⚪ Not enough data	Fewer than 20 entries we could price, or less than 14 days of history

The card also shows the number: copy after 60 s: +6% (trader +12%). The trader’s own figure uses their real fill prices, so the gap between the two shows how much of the edge the delay costs you.

How it is calculated

• Entries. An entry starts when the trader buys an outcome they did not hold, and ends when they sell out of it or merge it back into cash (holding both sides of a market and merging them is how many active traders exit). Both market and limit-order fills count. If they still hold it, it is valued at what it pays: 1 or 0 once the market resolved, the current price while it is open. A market that closed but whose result is still disputed is left out. Adding to or trimming a position along the way is ignored: the copier holds one fixed stake.
• The delay. The copy buys at the market price 60 seconds after the trader’s first buy, and sells 60 seconds after the trader sells out. A 5-minute delay is computed too: that is roughly how late alerts arrive on the Free plan.
• Costs. One cent per share is added to the copy’s buy price (never past $1) and taken off its sell price, to stand in for the spread and fees.
• Honest gaps. Prices come from Polymarket’s public price history, about one point a minute. If there is no price within 10 minutes of the moment we need, that entry is left out rather than guessed.
• Why “older than 30 days”. Most traders we score were picked by their profit over the last 30 days, so those 30 days always look good. The entries before them are the honest test.
• The window. Up to 60 days, read as six 10-day slices. For most traders we see every trade in each slice. Very active traders make hundreds of trades a day, so for them we take a sample in every slice: a run of their trades from a different moment each night, and for each market in that run their whole history in it, so we know exactly when the position opened and closed. Their card says “sampled entries”. At most 50 entries older than 30 days and 50 recent ones are scored, spread evenly in time. Scores are recalculated every night.

What it does not model

• Order book depth and slippage beyond the one-cent cost, so large copies would do worse.
• Taker fees above one cent, which some categories charge.
• Adds and trims inside a position.
• Positions the trader opened before the window starts.
• Every trade of a very active trader: their score rests on a sample of entries spread over the window, not on all of them.
• Markets where the trader created or converted shares (a split or a conversion), because the price they paid for each side is not known. Those markets are left out.

Read it the right way

Edge Score describes what copying would have returned in the past. Past results, not a forecast, and not financial advice. A trader can be copyable for 60 days and lose the next week. Use it to tell traders whose edge survives a delay from traders whose edge is gone by the time you hear about it.