Systematic trading researchCopy-trading signals across 8 platforms: few clean survivors, and fees that weigh on small accounts
Huyen X. Vu · Quantex Research, AlgoTrade Pte. Ltd., Singapore · ORCID 0000-0002-2209-842X
Abstract
On 6 November 2026 we read eight copy-trading platforms with one filter: real money, at least three years of history, maximum drawdown of 40% or less, a trade in the last 30 days, and no more than 12 months of pre-listing history. Only 30 forex and gold records pass: 12 on SignalStart, 9 on MQL5 and 9 on Neotech (formerly FXCE). Two Collective2 records pass every other test but are labelled hypothetical. ZuluTrade's public leader lists, eToro and FundSeeder add none. Displayed age is often misleading: 58 of the 100 oldest SignalStart records and 19 of 28 long-running, low-drawdown MQL5 signals include more than a year of pre-listing history. Leaderboards also mix in non-retail money: 26 of 37 Neotech accounts with three or more years of history are firm-funded and 2 are demo. Clean survivors show modest drawdowns and mid-range returns. A SignalStart follower pays US$25 a month plus the signal price (median US$35 on our shortlist), so an account needs several thousand dollars before typical returns cover the costs.
Keywords: copy trading; signal providers; survivorship bias; back-filled history; trading fees; forex; gold
1. Introduction
Our first report (R1) reviewed 15 long-running signals. This report widens the view to eight platforms and asks two questions: how many records pass a strict filter, and what does copying them cost?
Fund studies show that databases which keep only live funds overstate returns (Liang, 2000; Brown, Goetzmann and Park, 2001; Baquero, ter Horst and Verbeek, 2005). Public signal lists have the same problem, and two more: history can be loaded from before a record was listed, and some platforms mix demo or firm-funded accounts with retail ones.
2. Data and method
- Platforms: MQL5 Signals, SignalStart, ZuluTrade, Collective2, Neotech (fxce.com now redirects there), eToro, FundSeeder (forex part) and Myfxbook AutoTrade. All pages were read on 6 November 2026 by normal browsing of public pages, at a user's pace.
- Lists read: MQL5 table view sorted by age (768 MT4 and MT5 rows); SignalStart sorted by age (oldest 100 of 802); ZuluTrade public leader lists (32 leaders); Collective2 systems sorted by age (about 1,000); Neotech traders sorted by profit (pages 1–10 of 18); eToro most-copied investors; FundSeeder's public trader page.
- Filter (fixed before reading): real account; history of three years or more; maximum drawdown of 40% or less (30% or less and five years or more for the MQL5 shortlist, to keep it short); a trade in the last 30 days; history from before listing of 12 months or less.
- Back-fill check: weeks or months of history shown, minus the time since the record was started on the platform. More than 12 months means the record was loaded with older history.
- Our arithmetic: compound annual growth from the platform's own growth figure and age. Drawdown is each platform's own measure, so values are not strictly comparable across platforms.
3. Results
| Platform (what we read) | Pass | Main reason others drop out |
|---|---|---|
| SignalStart Oldest 100 (all 4+ years) | 12 | 27 no longer ranked (11 at −90% or worse); 58 back-filled |
| MQL5 Signals 28 long-running, low-drawdown | 9 | 19 back-filled |
| Neotech (ex-FXCE) 37 accounts trading 3+ years | 9 | 26 firm-funded, 2 demo |
| Collective2 35 forex systems 3+ years old | 2 (hypothetical) | Most idle or test accounts; results labelled hypothetical |
| ZuluTrade 32 leaders on public lists | 0 | High drawdown, imported history or no recent trades |
| eToro Most-copied investors | 0 | Mostly stock investors, not forex |
| FundSeeder Public trader page (14 profiles) | 0 | None trades forex |
| Myfxbook AutoTrade | 0 | The page now redirects to the Myfxbook home page |
Listed age is often not live age
On SignalStart, 58 of the 100 oldest records had more than a year of history from before they were added. On MQL5, 19 of the 28 long-running, low-drawdown signals did; one showed 508 weeks of history but had been started on MQL5 a month before we read it. The same strategy can look clean on one platform and back-filled on another.
The top of a leaderboard is often not retail money
Neotech's list sorted by profit is led by demo accounts. Of the 37 accounts trading three years or more in the first ten pages, 26 run on firm-provided capital under the firm's own drawdown rules, 2 are demo and 9 are real retail cent accounts. Collective2 marks its results as hypothetical. A record's account type has to be checked before its numbers mean anything.
What the clean survivors look like
| Platform | n | Max drawdown | Annual growth, median |
|---|---|---|---|
| Neotech cent accounts | 9 | 1.8%–16.0% (balance) | Growth 11%–208% over 3–5 years |
| MQL5 | 9 | 8%–30%, median 21% | 14% a year (range 4%–53%) |
| SignalStart | 12 | 2%–39%, median 29% | 31% a year (range 21%–46%) |
Most clean MQL5 survivors have no subscribers (8 of 9) and small equity, often a few thousand dollars or less. Several strategy families appear on two or three platforms, sometimes with very different results per account, so a name match is not proof of the same account.
4. The cost of copying
SignalStart states a service fee of US$25 a month for connecting a follower account, a 30% fee taken from the provider's subscription income, and a starting balance of US$1,000 (support page, read 6 Nov 2026). Signal prices on our 12 SignalStart survivors range from US$30 to US$500 a month, median US$35.
| Monthly cost | Yearly cost | At 14% a year | At 31% a year |
|---|---|---|---|
| US$60 (US$25 + median US$35 signal) US$720 | US$5,143 | US$2,323 | |
| US$100 (US$25 + US$75 signal) US$1,200 | US$8,571 | US$3,871 |
Below these sizes the follower pays more in fees than the signal earns, before spread, slippage and copy delay. Past growth is not a forecast, so the real break-even is likely higher. For the provider, a US$35 signal leaves about US$24.50 a month per subscriber after the 30% fee.
5. What it means
- For followers: check a record's listed age against its start date on the platform, confirm it is real retail money, prefer drawdowns under 20%, and keep enough capital that fixed fees are a small share of expected growth.
- For system builders: records that pass this filter are real from day one, with low drawdown and modest growth; listing on more than one platform lets others check them.
6. Limitations
- Public lists show mostly survivors; records that were removed are not seen. The 27 unranked SignalStart records are the only direct view of failures here.
- One reading date. Drawdown definitions differ by platform; ZuluTrade often shows 100% drawdown even with positive returns, and we did not resolve its definition.
- We read the first pages of each list only (oldest or most profitable), not the full population.
- Fees are from one platform (SignalStart); other platforms price differently.
7. Next step
We will re-read the same 32 records (30 real, 2 hypothetical) once a month for six months and report how many stop, so that failures are seen as they happen rather than inferred from who is left.
Sources
- Baquero, G., ter Horst, J., Verbeek, M. (2005). Survival, look-ahead bias, and persistence in hedge fund performance. Journal of Financial and Quantitative Analysis 40(3), 493–517.
- Brown, S. J., Goetzmann, W. N., Park, J. (2001). Careers and survival: competition and risk in the hedge fund and CTA industry. Journal of Finance 56(5), 1869–1886.
- Liang, B. (2000). Hedge funds: the living and the dead. Journal of Financial and Quantitative Analysis 35(3), 309–326.
- SignalStart support page, signalstart.com/support (read 6 Nov 2026).
- Public signal lists of MQL5, SignalStart, ZuluTrade, Collective2, Neotech, eToro and FundSeeder (read 6 Nov 2026).
- Quantex Research R1: How long do funds and trading signals survive?
Declarations
- Author: Huyen X. Vu, Quantex Research, AlgoTrade Pte. Ltd., Singapore (ORCID 0000-0002-2209-842X).
- Type: Quantex internal research report, published on this site only; not peer reviewed.
- Use of AI tools: Data collection and drafting were assisted by AI tools (Claude, Anthropic). The author checked the results and is responsible for the content.
- Conflict of interest: The author is the founder of AlgoTrade Pte. Ltd., which develops and runs systematic trading models. No platform named here paid for or reviewed this report.
- Funding: No external funding.
- Comments: welcome at [email protected].
- How to cite: Vu, H. X. (2026). Copy-trading signals across 8 platforms: few clean survivors, and fees that weigh on small accounts. Quantex Research, AlgoTrade Pte. Ltd. https://quantex.com.sg/research/copy-trading-survivors/