Field notes
Why Polymarket copy-trading tools can't touch a Polymarket US account
Updated 5 October 2026 · based on 2,400+ real copies and 10,023 observed trader bets
Every copy-trading service for Polymarket asks you to connect a wallet. A Polymarket US account doesn't have one. That single fact explains why none of them work on the US exchange, and it changes what a copy tool has to do to function there at all.
A wallet and an exchange account are different objects
International Polymarket settles on-chain. Your positions belong to a crypto wallet, and a copy tool signs transactions on that wallet's behalf once you connect it. Polymarket US is a different animal: a CFTC-regulated designated contract market where your account is identity-verified, your balance is cash held at the exchange, and access is granted by a developer API key rather than a wallet signature.
| Polymarket (international) | Polymarket US | |
|---|---|---|
| Account is | A crypto wallet | A verified exchange account |
| Third-party access via | Wallet connection / session keys | Key ID + Ed25519 secret from polymarket.us/developer |
| Each request | Signed transaction | Signed over timestamp + METHOD + path, sent in X-PM-* headers |
| Key can withdraw funds | Depends on the grant | No — trading permission only |
So a tool built for the wallet model has nothing to connect to on Polymarket US. It isn't a missing feature; the authentication model it depends on doesn't exist there.
How much of a trader's activity can actually be copied
This is the question nobody answers before you pay. Over one week we watched 10,023 bets from the traders we follow and recorded why each one did or didn't become a copy on Polymarket US.
| Reason a bet wasn't copied | Count | What it means |
|---|---|---|
| Game not listed on US | 2,561 | Whole leagues absent — Valorant, many soccer competitions |
| Bet type not offered | 2,362 | Player props, map handicaps, child moneylines |
| Underdog, below our price floor | 2,004 | Our rule, not a platform limit — see below |
| Trader had bet both sides | 587 | They're trading the swing, not picking a winner |
| Already at our per-game cap | 556 | Sizing limit, deliberate |
| Parlays | 0 | In 10,023 bets, these traders placed none |
Roughly one bet in six survives to become a copy. Any service implying you'll mirror a trader's whole book on Polymarket US is selling something it can't deliver.
What execution actually costs
Measured across our own copies, not estimated:
- Price difference versus the trader: a median of half a cent. Copying is not where the money leaks.
- Delay: median 21 seconds from their bet to ours, with traders checked every 30 seconds.
- Exchange fee: 2.6% of everything staked. Polymarket US charges takers
Θ × contracts × p × (1−p), so the cost per dollar falls as the price rises — a 75¢ favourite costs about half what a 50¢ coin-flip does.
That 2.6% is the number any copy strategy has to beat before it earns a cent. It is larger than the edge most traders show on the bets you can actually copy.
What a month of real money taught us
We ran this with our own money before selling anything. It lost $355 of $1,020 across roughly 270 settled games. The losses were not mysterious:
| Price we paid | Copies | Won | Result per $ |
|---|---|---|---|
| Under 50¢ (underdogs) | 215 | 27% | −42% |
| 50¢ and above (favourites) | 165 | 68% | +6% |
Underdog copies lost 42 cents per dollar staked, consistently, across two separate groups of traders. The reason is ordinary: a 38¢ underdog needs to win 38% of the time to break even, and these won 27%. Copying a profitable trader's underdog bets does not make them profitable for you, because a large part of their return on those positions comes from selling before the final whistle — which a copy account rarely replicates.
On correctly matched favourites, the picture inverted: 305 copies won 62% against an implied 61%, returning +$31.51 on $1,180 staked. Same traders, same software, different subset.
A record is evidence about the past. None of these figures predict what any trader does next, and a month is a small sample. We publish them because the alternative — a marketing page with no numbers on it — tells you even less.
What this means if you're evaluating a copy tool
- Ask what it connects to. If the answer is "your wallet", it cannot trade your US account.
- Ask what share of bets it can mirror, and what it does with the rest. "We copy everything" is false on this exchange.
- Ask to see the skips. A tool that won't show you the bets it declined is hiding its own failure rate.
- Ask how it sizes. One game taking a whole budget is how copy accounts die; our cap is 25% of a bot's budget per game, after learning the hard way.
Taildraft is in private testing
We're not taking payments yet. Leave your email and we'll tell you when accounts open.
Join the waitlist Read how market matching works