How to Evaluate a Trader Before Copying

Evaluate the record, not the rank. A usable track record needs enough independent trades, a clear split between closed and open profit, returns measured against capital actually at risk, a drawdown you could survive, and a strategy you can explain in one paragraph. Leaderboard position is a reason to start reading, not a conclusion.

Many copy-trading losses begin with a number that looked impressive in a short window. The job is to find out whether that number came from skill, risk, luck, or a definition you do not share.

Start with sample size, not rank

A win rate is an estimate, and estimates from small samples are noisy. If a trader has 60 resolved trades and a 55 percent win rate, the standard error is the square root of 0.55 times 0.45 divided by 60, which is about 6.4 percentage points. A rough 95 percent confidence interval runs from 42 percent to 68 percent. That range includes a coin flip.

At 200 resolved trades, the standard error falls to about 3.5 points, and the interval narrows to roughly 48 to 62 percent. Better, still not proof. A useful minimum is 100 resolved trades over six to twelve months, across more than one market regime.

Count independent decisions, not legs. Twenty positions that all depend on the same election, the same match, or the same weather system are closer to one bet than twenty. A trader who buys five correlated markets in one event has not produced five independent observations.

Separate realized from unrealized profit

A wallet can show a large gain on a position it still holds. That gain is not cash until the position is sold or the market resolves. Ask how much of the profit is closed, how much exposure remains open, and how old the open positions are. Polymarket's Data API exposes positions and trade history at wallet level, so this is checkable rather than a matter of trust.

The distinction changes the risk picture. A trader up $80,000 with no open positions has banked the result. A trader up $80,000 with $200,000 of open exposure in a longshot market has a mark-to-market gain that can disappear. The first record is a set of completed outcomes; the second is a snapshot.

Also separate profit from volume. Polymarket's leaderboard can rank traders by profit and loss or by volume, and volume counts both sides traded. High volume can mean many small edges, or it can mean churn that generates fees. Volume is not a proxy for skill.

Read the return correctly

ROI versus absolute profit

Absolute profit favors large bankrolls. A trader who makes $1 million on $50 million of average capital has earned 2 percent. A trader who makes $20,000 on $100,000 has earned 20 percent. Neither number is impressive on its own.

Return on capital has the opposite distortion. A tiny account can show a huge percentage by taking enormous risk, and a martingale can look brilliant until the one loss that ends it. The useful measure is return on average capital deployed, alongside the path it took to get there. Ask for profit per trade as well, because a 300 percent return produced by one position is a different object from the same return produced by 200 positions.

Consider two records. Trader A made $120,000 on $4 million of average capital, a 3 percent return, with an 8 percent maximum drawdown. Trader B made $18,000 on $60,000, a 30 percent return, with a 55 percent maximum drawdown. B has ten times the raw return and a risk profile that would wipe out many followers. The comparison is not which number is larger. It is which return you could hold through the drawdown.

The win-rate trap

Win rate without payoff is close to meaningless. Suppose a trader buys Yes at $0.80 and wins 70 percent of the time. A $1 stake buys 1.25 shares. A win pays $0.25; a loss costs $1.00. The expected value per dollar is 0.70 times $0.25 minus 0.30 times $1.00, which is minus $0.125, or a loss of 12.5 percent before fees. A 70 percent win rate can lose money.

Now suppose a trader buys at $0.20 and wins only 30 percent of the time. A $1 stake buys 5 shares. A win pays $4.00; a loss costs $1.00. The expected value is 0.30 times $4.00 minus 0.70 times $1.00, which is plus $0.50, or 50 percent before fees. A 30 percent win rate can be profitable.

Neither number tells you whether the market price was fair. The first trader may have been buying overpriced favorites; the second may have been buying underpriced longshots. What the numbers do show is that you need the average win, the average loss and the win rate together. A high win rate often describes a strategy that collects small gains and carries a large tail loss. That can work for years and then give back everything in one resolution.

The practical formula is expected value per trade: win rate times average win, minus loss rate times average loss, minus fees and slippage. If that number is not positive after costs, the win rate is decoration.

Drawdown, concentration and recency

Maximum drawdown is the largest peak-to-trough decline in account value. If a $100,000 account rises to $160,000 and falls to $96,000, the drawdown from the peak is 40 percent. Getting back to $160,000 from $96,000 requires a 66.7 percent gain. That asymmetry is why drawdown matters more than average return.

Look at the duration and the cause. A 40 percent drawdown spread over a year of many small losses is a different problem from a 40 percent drawdown caused by one market. The second tells you the sizing was concentrated. The first tells you the edge may be decaying.

Concentration shows up in three places: the largest position as a share of the portfolio, the top three positions as a share of total profit, and the share of volume in one event. A trader whose entire profit came from one lucky market is not a diversified strategy. A trader whose positions are large relative to the order book may not be able to exit at the prices shown.

Recency is a common trap. Leaderboards offer day, week, month and longer windows, and the top of a 30-day board is selected for recent performance. Compare the last 30 days with the last 12 months and with the full history. If the short window is spectacular and the long window is flat, you are looking at a streak, not a record. IOSCO also warns that leaders can change strategy or products without notifying followers, so check whether the recent trades resemble the old ones.

From checklist to decision

A practical checklist

1. At least 100 resolved trades over six to twelve months, across more than one market regime.
2. Closed profit separated from open profit, with open exposure and position age disclosed.
3. Return on average capital, not absolute profit, plus profit per trade.
4. Win rate, average win, average loss and expected value after fees.
5. Maximum drawdown, recovery time and worst month.
6. Largest position, top-three profit concentration and share of volume in one event.
7. 30-day, 12-month and full-history consistency.
8. A one-paragraph explanation of why the trader buys and when they sell.
9. A written answer to what would make you stop copying.
10. A 30-day paper copy that compares your fills with the leader's.

What the record cannot tell you

It cannot tell you intent. A wallet may be hedging another position, making a market, testing a model, or managing outside money. It cannot tell you about incentives or conflicts, especially when the trader is pseudonymous. It cannot tell you how the next market will resolve, and FINRA Rule 2210 explicitly prohibits communications that imply past performance will recur.

It also cannot tell you what your fills will be. The record belongs to the leader. Your copy is a new order in a shared book.

A good record is necessary, not sufficient. Copy only when the record is long enough to read, the risk is visible enough to size, and the strategy is legible enough that you would know when it stopped working.