The top trader on a copy trading leaderboard is usually the one who took the most risk, not the one with the most skill, because when luck drives short-run results the biggest bets produce the biggest returns. A copy trading leaderboard is a public ranking of traders by recent return that followers use to pick someone to copy automatically, and it drops the accounts those same bets destroyed. In an illustrative model of 1,000 traders with no edge, the riskiest third takes 99.7% of the top 10 seats and, under a rule that closes any account down 30%, loses 38% of its accounts within the year.
Ranking traders by raw return selects for volatility: among traders with no edge, the highest returns come from the largest risks, and the accounts those risks destroyed vanish from the board. A leaderboard can contain skilled traders but cannot tell them from lucky ones, so judge a trader to copy by drawdown, risk-adjusted return, sizing consistency and record length, not by rank.
Why do the riskiest traders rise to the top of a leaderboard?
The riskiest traders rise to the top because a ranking by return rewards the widest spread of outcomes, and without an edge the widest spread produces both the best and the worst results. Take 1,000 illustrative traders with zero skill (expected return 0%) in three equal groups by volatility, the standard deviation of monthly return: 2%, 6% and 12%. With normal, independent monthly returns, the 12-month standard deviation is the monthly figure times the square root of 12: 6.93%, 20.78% and 41.57%.
The expected 10th-best 12-month return in this population is about +78.2%, which is 11.3 standard deviations above zero for a 2% trader, 3.8 for a 6% trader and 1.9 for a 12% trader. About 1 in 33 high-volatility traders clears that bar by luck (a 2.99% chance), against about 1 in 12,000 mid-volatility traders and effectively none of the low group.
How much of the top 10 goes to the high-risk traders?
The 12% volatility group holds 99.72% of the expected top 10 seats in the model, although it is one third of the traders and none of them has an edge. The method finds the threshold return t at which the expected number of traders above t equals the leaderboard size k: Σ ng × (1 - Φ(t / σg)) = k, where ng = 333.33 traders per group, σg is the group's 12-month standard deviation and Φ is the standard normal cumulative distribution function. Each group's expected seats are its term in that sum.
| Leaderboard | 12-month return needed | Low vol (2%) seats | Mid vol (6%) seats | High vol (12%) seats | High-vol share |
|---|---|---|---|---|---|
| Top 10 | +78.23% | 0.0000 | 0.0279 | 9.9721 | 99.72% |
| Top 50 | +45.66% | 0.0000 | 4.6698 | 45.3302 | 90.66% |
| Top 100 | +30.70% | 0.0016 | 23.2874 | 76.7110 | 76.71% |
An exact order-statistic calculation (with groups of 334, 333 and 333) gives the high group 99.60% of the top 10, so the threshold shortcut holds. The shorter the board a follower reads, the purer the volatility filter. Real returns have fat tails and shifting volatility, so the table shows a mechanism, not a platform measurement.
What does research say about copying the top performers?
Laboratory evidence finds that seeing others' ranked results raises risk taking, and the option to copy raises it further. The abstract of Apesteguia, Oechssler and Weidenholzer (2020), an experimental asset-market study in Management Science, ends: "We conclude that copy trading leads to excessive risk taking."
The information screen in that experiment worked like a leaderboard; the authors' BSE Focus summary (2018) says "The information is ranked from highest to lowest selling prices." About a third of subjects chose to copy, and most of them copied subjects who had picked the riskiest assets. Whether copying pays at all is covered in does copy trading actually work.
eToro's own CopyTrader page warns that copying other traders "involves a high level of risks, even when following and/or copying or replicating the top-performing traders." Any list sorted by return carries the same bias, whatever the product format (compare copy trading, social trading, signals and PAMM).
What does a copy trading leaderboard hide?
A leaderboard hides the accounts that failed, and in a return ranking the failures come mostly from the same high-risk group that fills the top. Survivorship bias (judging a group only by the members still standing) can be sized with one illustrative rule: an account closes and leaves the board if its cumulative return is at or below -30% at any month-end.
Under that rule, 37.85% of the 12% group is wiped out within the year (126.2 of 333.3 accounts), against 11.08% of the 6% group (36.9 accounts) and 0.0009% of the 2% group. The high-volatility group supplies 77.35% of all 163.1 closed accounts. Its 207.2 survivors average +22.49% over 12 months, its 126.2 closed accounts average about -36.9% at closure, and the two together average exactly 0%, the group's true expectation. A follower who sees only survivors sees a high-risk group that looks profitable.
Removing the failures barely changes who tops the board: among survivors only, the top 10 cut is +78.16% and the high group still holds 9.97 of the 10 seats while making up 24.76% of the 836.9 visible accounts. Brown, Goetzmann, Ibbotson and Ross (1992) analyzed volatility and returns in mutual fund samples truncated by survivorship and showed that "this relationship gives rise to the appearance of predictability": past winners seem to keep winning once the losers are gone. The same arithmetic sets a single account's odds of breach, worked through in risk of ruin on a funded account.
Can a leaderboard tell a skilled trader from a lucky one?
A ranking by raw return cannot tell skill from luck, but a ranking by return per unit of risk can, when volatility is measured well. Give the model real differences: the low-volatility group earns +12% a year (+1% a month) with a 95.8% chance of a profitable year, the middle group earns 0%, and the high-volatility group loses 12% a year yet still ends 38.6% of years in profit.
Ranked by raw return with the same threshold method, the skilled group wins 0 of the 10 seats; the losing group takes 9.77 and the middle group 0.23, with a cut-off of +66.6%. Ranked by 12-month return divided by the group's true 12-month volatility (a Sharpe ratio with a zero risk-free rate, return per unit of risk, here 1.73 for the skilled group), the skilled group takes 9.93 seats, the middle group 0.05 and the losing group 0.02. Estimation noise does not reverse the result: in 20,000 simulated leaderboards ranked by a Sharpe ratio computed from only 12 monthly returns, the skilled group still took 9.72 seats. With zero skill everywhere, the same ranking splits the seats 3.33 each. Measuring results in R-multiples applies the same idea trade by trade.
Skill does exist, which makes a leaderboard noisy rather than empty. Barber, Lee, Liu and Odean (2014) sorted Taiwanese day traders from 1992 to 2006 by one year's return and found that the top 500 went on to earn 61.3 basis points (hundredths of a percent) a day before fees, 37.9 after, the following year, while the bottom-ranked earned -11.5 before fees and -28.9 after. They also concluded: "Less than 1% of the day trader population is able to predictably and reliably earn positive abnormal returns net of fees."
How do you choose a trader to copy without trusting the rank?
Choose on five measures that scale return by risk and take a long record to fake.
- Maximum drawdown (the largest peak-to-trough fall in equity) is the worst loss a copier would have lived through; compare it with what your account can absorb at your copy size.
- Risk-adjusted return, return divided by volatility, is the ranking that recovered the skilled group in the model.
- Time on the board matters because one year is a small sample and earlier closures are invisible.
- Position sizing consistency exposes a trader who doubles size after losses or puts half the account in one position, buying rank with risk.
- Number of trades separates edge from noise: hundreds of trades at a steady size say more than a few large bets.
Platform filters help at the extremes. As of September 2026, eToro's Pro Investor program (formerly Popular Investor) lists "Max daily risk score ≤ 7", "Max weekly drawdown ≥ -25%" and "Max single position size ≤ 50% of portfolio" among the standards members must maintain, though the page does not define how the risk score is calculated. Those caps remove the most extreme risk takers, yet a trader can still lose a quarter of the account in a week and stay inside them. Copying several top traders at once does not dilute the problem either, because correlated copied positions tend to fail together.
Is copying your own accounts safer than copying a leaderboard trader?
Copying your own accounts removes the leaderboard's selection problem, because the master account's sizing, drawdown and every trade are known rather than inferred from a rank, but the result is only as safe as your own strategy. A copier multiplies whatever the master does: an untested strategy copied to five accounts produces five correlated results, not five chances.
A trade copier is not the answer if you have no tested edge yet, or if the goal is to borrow someone else's skill. Many prop firms ban copying other people's signals into funded accounts (see why prop firms ban copying signal providers), and the rules on copying others differ by country (see is copy trading legal). Thor, this blog's own product, copies a trader's master account to that trader's other accounts server-side; it does not rank, recommend or supply traders to copy.
Go deeper
- Does Copy Trading Actually Work? An Honest Answer
- Risk of Ruin: The Odds You Blow a Funded Account
- Copy Trading Correlation Risk: Why Ten Funded Accounts Copying One Master Isn't Ten Times the Diversification
- Copying Signal Providers on Prop Firms: Why Paid Signals Get You Banned
Frequently asked questions
Should I copy the number one trader on a copy trading leaderboard?
Not on rank alone, because first place is the purest volatility signal on the board. By the same threshold method, the illustrative high-volatility group takes 99.9994% of the expected first place, with a winning 12-month return near +114%.
Why do top copy traders blow up after people start copying them?
Because the ranking picked them for volatility, and the volatility continues after you join. In the illustrative model past returns do not change future odds, so a trader with 12% monthly volatility faces the same 38% chance of being down 30% or more at some month-end over the next 12 months, whether they ranked first or last.
How long a track record do I need to judge a trader?
Several years, not one. Under normal returns and a modest Sharpe ratio, the standard error of an annualized Sharpe ratio is roughly 1 divided by the square root of the years of data: about 1.0 after one year, 0.71 after two and 0.5 after four. A one-year Sharpe of 1.5 therefore fits, within two standard errors, a true value anywhere from about -0.5 to 3.5.
How many trades should a trader have before the results mean anything?
Enough that a lucky streak cannot explain them. Among 1,000 traders flipping fair coins, 1,000 x 0.5 to the fifth power = 31.25 would be expected to go five for five, so a short perfect record appears by chance in any large crowd.
Is a high return always a sign of excessive risk?
No, but a return shown without a risk measure beside it cannot be interpreted. A +30% year at 6.93% annual volatility sits 4.3 standard deviations from zero and is hard to get by luck, while the same +30% at 41.57% volatility sits 0.7 standard deviations away, which luck delivers about one year in four.
Does a platform risk score fix the leaderboard problem?
No, a cap only trims the extremes. Inside the allowed band, a return ranking still favors whoever runs closest to the cap and still cannot separate skill from luck.
Can traders game a copy trading leaderboard?
Yes. A trader can run several high-risk accounts and promote whichever one wins, turning luck into an apparent record. Ranking contests also push traders who are behind to add risk: Brown, Harlow and Starks (1996) found that among 334 growth-oriented mutual funds from 1976 to 1991, mid-year losers raised volatility later in the year more than mid-year winners did.
Does a high copier count mean a trader is good?
No, a copier count measures visibility, not skill. Followers arrive after a trader ranks high, so the count amplifies the return ranking instead of checking it.
Do prop firm leaderboards have the same bias?
Yes, any ranking by raw profit over a short window does. A board of funded traders ranked by profit or payouts shows the winners of a volatility contest and leaves out the accounts that breached their drawdown on the way.
Sources
- Apesteguia, Oechssler and Weidenholzer (2020), Copy Trading, Management Science 66(12)
- Apesteguia, Oechssler and Weidenholzer (2018), Copy Trading and its Implications for Risk Taking, BSE Focus
- eToro (2026), CopyTrader
- Brown, Goetzmann, Ibbotson and Ross (1992), Survivorship Bias in Performance Studies, Review of Financial Studies 5(4)
- Barber, Lee, Liu and Odean (2014), The cross-section of speculator skill: Evidence from day trading, Journal of Financial Markets 18
- eToro (2026), Pro Investor Program