Does copy trading work? We followed the best leaders for 14 quarters
Does copy trading work if you copy only the leaders with the best past year? We read the daily account records of 1,310 real-money leaders on ZuluTrade and followed the best fifth, picked afresh every quarter, from 2023 to June 2026. They lost less than the rest, but they still lost: 7.7% in the average quarter, before any fee.
The belief
“Copy the traders with the best record, and you will make money.”
In the average quarter, the best fifth of leaders by their past year lost 7.7%, all leaders still trading lost 11.1%, and the worst fifth lost 24.6%. A good record pointed to smaller losses, not to gains, and that was before any fee for following.
What copy trading promises
Copy trading links your account to another trader's, called a leader or signal provider, and every trade they make is placed in your account too. Platforms rank their leaders by past results, and the natural plan is to copy the best of them and drop the ones who fade. Does copy trading work that way? We tested the plan on ZuluTrade, a copy-trading platform that shows each leader's daily account record on public pages.
How we tested it
We started from ZuluTrade's list of leaders of 7 October 2026 and took every real-money leader who had been on the platform for at least 156 weeks, about three years: 1,314 leaders, of whom 1,310 had a daily record we could read. The list keeps leaders who stopped trading, so the failures are still in it. From each record we took what the account itself earned, day by day, with deposits and withdrawals taken out.
On the last day of every quarter from 31 December 2022 to 31 March 2026, 14 days in all, we ranked the leaders who were still trading and had at least a year of record behind them by their growth over the past year. The best fifth became the roster. Equal money went to each member and stayed for the next quarter; then the roster was formed again, the weakest out and the best in. We compared it with all leaders still trading, with the worst fifth, and with 200 rosters of the same size drawn at random.
Two choices make the test kind to the plan. Nothing was charged for following: no subscription, no share of profits, no slippage on copied trades. And the rules for a pass were written down before any leader's result was read. ZuluTrade's figures are not audited, so we cleaned them by fixed rules, set out at the end of the page.
What we found
The best fifth lost 7.7% in the average quarter. All leaders in trading lost 11.1%, and the worst fifth lost 24.6%. So the past year did tell something: the leaders with the best record went on to lose less than the rest, and the worst went on to lose the most. But losing less is not making money. The best fifth was below zero in both parts of the test: −5.7% a quarter for the rosters formed from December 2022 to September 2024, and −10.3% for those from December 2024 to March 2026.
The quarter after each roster was formed, average of 14 quarters, %
| Dec 2022 – Sep 2024 | Dec 2024 – Mar 2026 | All | |
|---|---|---|---|
| Rosters formed | 8 | 6 | 14 |
| Best fifth | −5.7% | −10.3% | −7.7% |
| All in trading | −10.0% | −12.7% | −11.1% |
| Worst fifth | −23.5% | −26.2% | −24.6% |
| Random rosters, middle | −10.0% | −12.5% | −11.1% |
Quarter by quarter
The best fifth ended a quarter ahead 5 times in 14, and did better than the whole group in 10 of them. Its two worst quarters came at the end: the roster formed on 30 September 2025 lost 39.6% in the last three months of 2025, and its middle member lost 53.8%; the next roster lost 32.2%. All leaders together lost money in every one of the 14 quarters.
| Quarter followed | Best fifth | All | Worst fifth |
|---|---|---|---|
| Jan–Mar 2023 | −8.4% | −6.3% | −4.8% |
| Apr–Jun 2023 | −8.7% | −13.0% | −18.8% |
| Jul–Sep 2023 | −4.9% | −11.6% | −37.2% |
| Oct–Dec 2023 | +0.3% | −8.4% | −25.1% |
| Jan–Mar 2024 | +9.1% | −3.7% | −19.3% |
| Apr–Jun 2024 | −11.2% | −11.0% | −14.9% |
| Jul–Sep 2024 | −12.9% | −13.4% | −43.5% |
| Oct–Dec 2024 | −8.7% | −12.2% | −24.4% |
| Jan–Mar 2025 | +2.1% | −11.2% | −35.2% |
| Apr–Jun 2025 | −6.7% | −12.1% | −25.1% |
| Jul–Sep 2025 | +14.1% | −4.3% | −17.4% |
| Oct–Dec 2025 | −39.6% | −15.2% | −17.2% |
| Jan–Mar 2026 | −32.2% | −22.3% | −30.3% |
| Apr–Jun 2026 | +0.2% | −11.2% | −31.9% |
Why the best still lost
Most of these records go badly. On a typical roster day, the middle leader of the whole pool had lost 18.1% over the past year. A gain of 28.2% was enough to make the best fifth, and the roster's middle member had gained 72.3%.
In the quarter after, 54% of the best fifth ended ahead on average, against 43% of all leaders; 16% lost half or more, against 19%; and 5% did not trade at all, against 11%. So winners were the majority on average, but the losers lost more than the winners gained, and the roster as a whole lost money.
The rare huge quarters
A quarter of several hundred percent is not rare on these records, and one of them can carry an average. So in the main figures each leader's quarter is cut at +100%. The cut touched 72 of the 2,985 leader-quarters in the pools, but only 3 of the 601 in the rosters. Left uncut, the picture turns: all leaders in trading averaged −6.4% a quarter and the best fifth −7.5%, and the best fifth did better than only 32% of the random rosters. The few windfalls came from leaders a past-year ranking did not pick.
Following the roster for three and a half years
A follower who formed the roster again every quarter, from January 2023 to June 2026, would have lost 72.5% of the money, before any fee. Following all leaders in trading the same way lost 81.2%. With the huge quarters left uncut, the best fifth lost 71.9% and all leaders 62.7%. The roster's worst single quarter was −39.6%.
Following each group for 14 quarters, January 2023 to June 2026, %
Does copy trading work, then?
Not as a way to make money from these records. Ranking leaders by their past year did sort them: the agreement between a leader's rank and its next quarter was positive on 12 of the 14 roster days, +0.166 on average on a scale of −1 to +1, and the best fifth lost less than the worst. But the best fifth still lost money in both parts of the test, and of the nine conditions set before the run, four were met. We counted no fees; a follower who pays a subscription or a share of profits would have done worse.
The test covers one platform and three and a half years, and the records are not audited. It does not show that no leader is worth following. It shows that a strong past year was not enough to find one.
Questions people ask
Does copy trading actually work?
Not on ZuluTrade's public records from 2023 to June 2026. Copying the leaders with the best past year lost 7.7% in the average quarter and 72.5% over 14 quarters, before any fee. They did lose less than the average leader, who lost 11.1% a quarter.
Is copy trading profitable?
For the leaders as a group, no. The real-money leaders still trading lost money in every one of the 14 quarters we followed, 11.1% in the average quarter. In an average quarter 43% of them ended ahead.
How much can you make from copy trading?
On these records a follower of the best fifth lost money on average. The roster's best quarter was +14.1%, from July to September 2025; its worst was −39.6%, from October to December 2025.
Do copy traders with the best record keep winning?
Only in part. The best fifth by the past year did better than all leaders in 10 of the 14 quarters, but ended ahead in only 5. A strong past year pointed to a smaller loss, not to a gain.
Is copy trading worth it?
On this evidence, no. Even the best-ranked leaders lost money before fees, and any subscription or share of profits comes on top. The test covers one platform and three and a half years, so it cannot rule out that some leaders are worth following; it shows that a strong past year did not find them.
The bottom line
Copying the leaders with the best past year did better than copying the average leader, but it still lost money: 7.7% in the average quarter on ZuluTrade's records from 2023 to June 2026, before any fee. A good record pointed to smaller losses, not to profits. MYTH.
How we tested
- Belief
- Following the copy-trading leaders with the best past results makes money.
- Data
- ZuluTrade's public leader pages, read in October 2026: every real-money leader on the list of 7 October 2026 with at least 156 weeks on the platform, 1,314 leaders, 1,310 with a readable daily record. The list keeps leaders who stopped.
- Period
- Rosters formed on the last day of each quarter, 31 December 2022 to 31 March 2026 (14), each held for the next quarter, to June 2026. Two parts: 8 rosters to September 2024, 6 from December 2024.
- Rule
- On each roster day, the leaders with a year of record behind them whose account moved on five or more of the last 90 days were ranked by growth over the past 365 days. The best fifth was held with equal money for the next quarter.
- Cleaning
- Growth is the account's own equity day by day, with recorded deposits and withdrawals taken out. A rise of more than 100% in one day was taken as an unrecorded deposit and counted as no change (2,352 days on 886 leaders). In the main figures each leader's quarter is cut at +100%; the uncut figures are given beside them. Falls stay in, so a withdrawal the record does not show counts as a loss.
- Costs
- None: no subscription, share of profits or slippage on copied trades. Real followers pay some of these.
- Compared with
- All leaders in trading on the same day; the worst fifth; 200 random rosters of the same size.
- Pass rules
- Nine conditions written before any leader's result was read, with one change to the cleaning made before the run: among them, the roster above zero over the whole period and in both parts, above all leaders, and above 95% of the random rosters. Four were met.
Leader records: ZuluTrade's public leader pages (zulutrade.com), read in October 2026. ZuluTrade's figures are not audited. The cleaning, the rosters and the statistics are our own; no leader is named and no single record is shown. ZuluTrade has not reviewed this test.
This is research on past prices. It is not advice to buy or sell anything, and past results do not promise future ones.
Keep reading
More tests
- Test 005MythDoubling the trade after every loss is meant to win everything back. On EUR/USD since 1999, about half of the martingale accounts were wiped out within a year.
- Test 004MythA famous 2003 study found stock markets did better on sunny mornings. In ten markets from 2004 to 2025 the sunny days did not win, and trading the weather lost money even before costs.
- Test 003MythThe fear and greed index says extreme fear is a time to buy Bitcoin. From 2018 to 2026 the day and the week after extreme fear were no better than after any other reading.