How to Judge a Copy-trading, Signal or EA Track Record Before You Follow

Before you pay for signals, follow a copy-trading provider or buy an EA, what you are really buying is someone else’s past results. The trouble is that a track record is very easy to dress up without a single false number. The seller just chooses which account, which period and which statistic to show. This article collects checks that work for anyone selling results: signal sellers, copy-trading providers and EA sellers.

Screenshots versus verified accounts

A screenshot of profits proves nothing. The numbers can be edited, it can be a demo account, or it can be the one account that survived out of ten. A track record you can start to trust looks like this:

• The account is connected directly to a tracking site. The data comes from the broker’s server, not typed in by the owner, usually through a read-only login (Investor password)
• Every trade is visible, with open and close times, prices, lots and Stop loss, not just a nice profit curve
• It says whether the account is real or demo. A demo account has no slippage problems, so its results always look better than reality
• Deposits and withdrawals are visible, because they can distort the % gain a lot. See the section below

Even a verified account only proves that the results happened. It doesn’t prove they can happen again. The rest of this article is about checking that second part.

Account age and number of trades

Few trades means the numbers can still swing a lot. Take an 84% win rate from 66 trades. How far off can that be?

1. Standard error = √(0.84 × 0.16 ÷ 66) = 4.5%
2. Rough 95% confidence range = 84% ± 1.96 × 4.5% = 75.2% to 92.8%

With only 66 trades, the true win rate could be anywhere in a range almost 18 points wide. With 500 trades the range narrows to 80.8% to 87.2%, and that is before the market itself changes character.

A rough rule: at least 12 months, several hundred trades, and more than one kind of market. A system that shows only 2–3 months of a one-way market hasn’t really been tested. Read more in When the market regime changes.

Drawdown: look at equity, not balance

Drawdown is how far the account fell from its previous peak. The figure to look at is equity drawdown (counting trades still open), not balance drawdown (closed trades only). A system that refuses to close losing trades always shows a misleadingly low balance drawdown. The details are in Drawdown: equity vs balance.

If a results page shows only profit and balance drawdown, without equity drawdown, treat the real risk as unknown, and plan for the future being worse than anything you have seen.

Signs of martingale and grid

Martingale doubles the lot after a loss. Grid adds trades at intervals as price moves against you. Both make the profit curve look beautiful, until the day price moves far enough the wrong way. The full mechanics are in Grid and martingale.

We simulated a simple martingale system on an illustrative 300-bar price series. Balance $1,000. It opens a Buy of 1 unit (a 1-point price move = $1 profit or loss). Each time price falls 2 points below the latest trade, it opens another Buy at double the lot. The whole batch closes once it is in profit overall. No Stop loss.

This run survived. Price bounced in time, the 7-trade batch closed with a large profit, and the balance jumped up. But the low point was only about 4 points from wiping out the account. When we ran the same system on 40 random price series, equity reached zero in 3 of them before the 300 bars were over, and every one of the 40 showed a balance drawdown of 0% along the way. The balance figure tells you nothing here.

Signs you can see on a results page:

• A balance curve that is almost a straight line, while equity drawdown is several times the balance drawdown
• Lots growing after losing trades, or several trades in the same direction stacked up and closed together
• No Stop loss, or one so far away it is never hit
• A win rate of 90% or more with an average loss many times the average win. Read A high win rate doesn’t mean profit

Survivorship bias: the ones on display are the lucky ones

Survivorship bias means seeing only the survivors and concluding their method works, without seeing the many who disappeared. In the signal business, losing accounts are closed quietly and lucky accounts go on the front page.

We simulated 1,000 traders with no skill at all. Each takes one trade a day, 252 trades over 12 months. Every trade is a coin flip: a win makes 1% of starting balance, a loss costs 1% (RR 1:1, 50/50 odds, no spread). Everyone’s expected result is 0%.

What the simulation shows:

• The top 10 made +38% to +62%. The best won 157 of 252 trades (62.3%) by flipping a coin
• Check with the probability formula: reaching +38% takes at least 145 wins. The chance of that is 0.98%, about 9.8 traders per 1,000, matching the 10 we got
• 16 traders had at least 10 profitable months out of 12. The formula expects 19.3 (79 ÷ 4,096 × 1,000). They look “consistently profitable”, but it is luck
• When the top 10 traded another 12 months, their average was +3%, ranging from −38% to +28%. Back to coin flipping

If enough people sell results, some will always have excellent ones, even if nobody has any skill. A good record on its own is not yet evidence of skill.

This bias isn’t only on sales pages. Our own memory prefers the winners too. Read Trading biases.

Cherry-picked periods

Another way to mislead without lying is choosing the window: showing “the last 6 months” because the 6 months before were a heavy loss, or opening a fresh account after blowing one up and starting the count again. To protect yourself:

• Ask for results from the first day of the account, and ask whether the same system runs on other accounts
• Look at every month’s result, not only the total
• Be wary of a young account from someone who says they have “traded for 10 years”

The same thing happens with backtests sold alongside EAs. The period used for tuning usually looks best. See how to separate test periods in Walk-forward testing and Overfitting.

Deposits and withdrawals distort % gain

The % gain you see may not match the money actually made. Example:

1. Deposit $1,000 and make $500 profit: +50%
2. Deposit another $10,000. The account is now $11,500
3. Lose $2,000: −2,000 ÷ 11,500 = −17.4%
4. Chained % gain (deposits excluded) = 1.5 × (1 − 0.174) − 1 = +23.9%
5. Actual money: $11,000 deposited, $9,500 left, a loss of $1,500

The chart says +24% while 1,500 dollars disappeared. The gain happened while the account was small and the loss while it was large. The other way round, naively comparing the final balance with the first deposit, $9,500 ÷ $1,000, gives +850%, which is even more wrong. So look at the % gain, the deposits and withdrawals, and the profit in money, together.

Copiers don’t get the source’s results

Even when every number is genuine, a copier ends up somewhere else, because of:

• Slippage. The copier’s order always goes out after the source’s, at a worse price, and possibly at a different broker with a different spread. Systems that make only a few dollars per trade are hit hardest
• Lot rounding. A $20,000 source account opens 0.30 lot. A $1,000 copier should get 0.30 × 1,000 ÷ 20,000 = 0.015 lot, but the minimum step is 0.01, so it gets 0.01 (33% less risk) or 0.02 (33% more). If the source opens 0.05 lot, the copier should get 0.0025 but is rounded up to 0.01: 4 times the source’s risk
• Missed trades. If you start copying while the source has open trades, or the connection drops for a while, the results can split apart completely

In our backtest, we took a frequently-winning scalping system whose source showed 66 trades with an 84% win rate, and a drawdown of about 61% on the source account itself. We wrote the rules as we could read them from the source’s trades (a tight $5 Stop loss, a tight trailing stop, frequent re-entries) and tested them on gold M1 for 2026 with real tick data, a $3,000 balance and 0.10 lot. The result: a 35.7% win rate over 1,954 trades and a loss of $2,928, or −98% of the balance. Most of the gap came from execution. A spread of about $0.24–0.30 is 5–6% of the Stop loss on every trade, and the tight trailing stop was hit before price moved on. Part of it may be that our reading of the rules was not exact, but the point stands: a system like this depends more on execution quality than on the strategy. Backtest results do not predict future results, and they are not returns you should expect.

In our own planning, we assume live results may come in at around 40–60% of the backtest. That figure is an estimate from expected costs, not a measured result. The same principle applies to anyone else’s record: expect less than you see. The full reasoning is in Why backtest and live results differ.

Fees

Measure fees against your account size, not the source’s.

• Monthly subscription. $30 a month on a $500 account is 30 × 12 ÷ 500 = 72% a year. The system has to make more than 72% just for you to break even
• Performance fee. Charged only on new profit above the previous peak (a high-water mark), it is reasonable. Charged every month regardless of last month’s loss, you pay again on profit that merely recovers an earlier loss
• Added spread or commission. Some schemes require an account opened through a link with a wider spread, which hurts frequent-trading systems most

The most important question isn’t “how much did they make” but “after everything is deducted, what is left for an account my size, and could I sit through their worst equity drawdown?”

Checklist before you follow

1. The account is connected directly to a tracking site, it is a real account, and every trade and every deposit and withdrawal is visible
2. At least 12 months, several hundred trades, more than one kind of market
3. Equity drawdown is not far from balance drawdown, and there is no sign of lot doubling
4. Monthly results from day one, not a chosen window
5. Fees measured against your account still leave it worthwhile after you discount the results
6. Size your lots so that the source’s worst equity drawdown (plus a margin) is money you can afford to lose. The lot size calculator helps

If the seller guarantees profit, pressures you to decide, or asks you to transfer money so they can trade for you, read Forex scam red flags first.

Common mistakes

• Trusting profit screenshots or videos. Evidence you can’t check isn’t evidence
• Looking only at % gain and win rate without equity drawdown. A martingale system always wins often and looks smooth, right up to the day it blows up
• Picking the top of a leaderboard. The top of a large group usually includes the lucky, as the 1,000-trader simulation shows
• Copying the source’s lot size with a much smaller account, or not checking how many times the minimum lot rounding multiplies your risk
• Adding money after a winning run. The best stretch is often the luckiest one, and it doesn’t mean the next stretch will look the same. Read Losing streaks are normal