Copy trading is the obvious use case: find a wallet with a good track record, copy their bets, make money. It's also where most people lose money. This guide explains why, and what actually works.

Why naive copy trading fails

Three reasons the obvious version doesn't work:

When a whale opens a large position, their own order moves the price. By the time you see the trade, the price is worse than the price they got. On a shallow Polymarket market, a $100K whale buy can move the price 2–5 cents — that's 2–5 percentage points of probability. Your "copy" is no longer the same trade.

2. Survivorship bias

The whales on CrowdIntel's leaderboard are the ones who've won. You see their track record and think "look at this edge." But they're the survivors of a much larger pool of wallets that started the same way and blew up. Past performance of the survivors is not an estimate of ex ante edge for a random sharp — it's a ceiling, not a mean.

3. Information asymmetry doesn't persist

An insider wins because they know something others don't. The moment they bet, the information is partially leaked into the price. If the information was going to become public quickly, the edge decays fast. Copying 30 seconds late often means copying after the edge has burned off.

What actually works

Three adjustments turn copy trading from a loser into a reasonable strategy:

1. Copy early, or not at all

If you can be in within seconds of a whale's trade, slippage is low and information edge is maximum. If it takes you 10+ minutes, don't bother on short-horizon markets. Insider Radar surfaces flagged trades in real time; pair with browser notifications.

For long-horizon markets (weeks/months to resolution), the slippage tax is a smaller fraction of your expected hold, so lateness matters less. These are the better copy-trade targets for humans.

2. Copy direction, not size

A whale betting $500K at 35¢ doesn't tell you to bet $500K at 37¢. It tells you the whale thinks the true probability is higher than the market's 35¢. Translate that into your position size, based on your conviction and bankroll.

Useful heuristic: if the whale is sizing 1% of their (estimated) bankroll, you can size 1% of yours — but don't copy absolute dollars.

3. Copy clusters, not individuals

A single whale can be wrong. A cluster of wallets (or a basket of independent sharps) betting the same direction on the same market is a much stronger signal, because their independent information has to agree for them all to take the same side. Use CrowdIntel's investigations and insiders hub to find these.

Who to copy (and who not to)

Copy candidates

Shrinkage-adjusted win rate ≥ 65% over 100+ resolved bets, positive PnL. The floor for statistical credibility.
Category-concentrated edge. A wallet with 85% in one domain and 55% elsewhere gives you a clear lane.
Recent activity. If the wallet's last resolved bet was more than 90 days ago, the operator may have changed.

Avoid

High raw win rate, low sample size. A wallet with 9 wins in 10 bets has a shrinkage-adjusted rate closer to 60% — don't anchor on the 90%.
High win rate, negative PnL. Classic heavy-favorite bettor. Their wins pay small, their losses cost big. Don't copy.
Whales with obvious self-trade patterns. Suspiciously smooth PnL curves are sometimes artificial.
Solo insiders on one-off markets. High-variance copy with minimal recoverable edge.

Sizing rule of thumb

Assume half of a whale's historical edge will show up in your copy-trade PnL after slippage and delay. If their shrinkage-adjusted edge is +15 percentage points over the base rate, size as if yours is +7.5. This keeps you out of disaster when reality is worse than backtest.

A simple workflow

Pick one wallet from /whales that meets the "copy candidates" bar above. One at a time — don't try to copy a basket mentally.
Watchlist them. Enable the notification dot. If Telegram alerts have shipped, subscribe.
When they place a bet, open the market page immediately. Check if the price you can get is within 1–2 cents of the price they got. If yes, trade. If no, skip this one.
Record your entry, their entry, and your thesis. Revisit weekly.
After 20 copy-trades, review: what's your PnL vs theirs on the same trades? If you're capturing < 40% of their edge, either change wallet or quit copy-trading.

What not to do

How to find insiders — finding a good wallet to copy
How to track a whale — watchlists and alerts
/whales — ranked leaderboard
/investigations — published clusters (stronger signal than individuals)

Copy trading Polymarket (responsibly)

Why naive copy trading fails

1. Slippage