Systematic trading researchHow long do funds and trading signals survive? What the survivors share
Huyen X. Vu · Quantex Research, AlgoTrade Pte. Ltd., Singapore · ORCID 0000-0002-2209-842X
Abstract
We bring together published studies of hedge fund attrition, reports on large failures, and public records from 15 long-running trading signals on MQL5, Darwinex, and Collective2. Funds close at about 8% to 9% a year, and about a third of Nordic funds were still running after eleven years. Poor returns, small size, high leverage, and weak operations raise the risk of closing. The records that last eight years or more share low margin use, risk spread across markets or small fixed lots, and stops; the ones that collapsed used high margin and large averaging adds. Signal age on MQL5 can include history published after the fact, so the start date needs checking.
Keywords: fund survival; hedge fund attrition; copy trading; signal providers; survivorship bias; leverage; family office
1. Introduction
Most funds and public trading records do not last. This report asks what the few that last have in common, using published fund studies alongside the public pages and trade histories of long-running signals.
Research question
How often do funds and long-running public trading records stop, what predicts it, and what do the records that last eight years or more have in common?
2. Data and method
- Funds: published studies of hedge fund attrition (TASS and HFR databases, a Nordic fund census, an industry report) and public reports on large failures. Each number is taken from the cited source.
- Family offices: the UBS Global Family Office Report 2025 and public summaries of the Singapore (13O / 13U) and DIFC (Dubai) family office rules.
- Public trading records: 11 of the longest-running signals on MQL5 (pages read 1 and 3 November 2026; full trade history exported on 5 November 2026), three Darwinex DARWINs and one Collective2 strategy. We read growth, drawdown, margin use, symbols, deposits and withdrawals from the public pages and counted lot sizes per year from the exported trade history.
- Percentages of growth are the platform's own (time-weighted). Compound annual growth (CAGR) is our arithmetic from growth and weeks of history.
3. Funds: how often they die and why
| Source | Data | Finding |
|---|---|---|
| Getmansky, Lo and Mei (2004) | TASS, 1994–2003 | 8.8% of funds die per year (11.4% in 2001). Managed futures 14.4%, global macro 12.6%, long/short equity 7.6%. Median age at liquidation 45 months; half of liquidated funds never reached their fourth year. Median size at liquidation US$6.3 million. |
| Liang (2000) | TASS 1994–1998, HFR 1994–1997 | Attrition 8.3% per year (TASS) and 2.72% (HFR). |
| HedgeNordic (Guzun, May 2026) | Nordic funds closed 2015–2026 | Closed funds lived 7.3 years on average (median 5.3). Of 169 funds active at the start of 2015, 55 still run (about one third). |
| HFR (2Q 2025) | Global industry | 2024: 479 launches and 406 liquidations. |
At 8.8% a year, about 40% of funds would still be alive after ten years; at the managed-futures rate, about 21%. The Nordic census (about one third alive after eleven years) is close to this.
What predicts death
- Poor returns, small size, young age and high leverage raise the chance of closing; more of the manager's own money in the fund lowers it (Liang 2000).
- Higher volatility goes with higher attrition in every style (Getmansky, Lo and Mei 2004).
- In a study of 100 fund failures over 20 years, half failed from operational problems alone, and misrepresentation, fraud or unauthorised trading played a part in 85% of those (Capco, 2003).
Two large failures and one change of form
- Archegos (2021): a family office that built exposure of about six times its capital through total return swaps, with five stocks making up 80% of its long book. It defaulted on margin calls in March 2021; its banks lost more than US$10 billion (ESMA, 2022).
- Amaranth (2006): energy trades were about half of its capital. It lost about half of its assets in a few weeks of September 2006, about US$6.6 billion (Till, 2008).
- Soros Fund Management (2011): returned outside money and became a family office for its own family's capital.
Family offices: size and rules
Single family offices in the UBS survey (317 offices, 2025) manage US$1.1 billion on average for families worth US$2.7 billion. Public summaries of the rules give these entry points: Singapore 13O at least S$20 million under management with two investment professionals, 13U at least S$50 million; DIFC (Dubai) at least US$50 million of family net assets. In both places, managing money for outsiders falls under licensed activity. These come from secondary summaries; check the regulators' own texts before use.
4. Long-running public trading records
| Signal (ID) | Weeks | Growth | Relative drawdown | Peak margin use | What the record shows |
|---|---|---|---|---|---|
| SFE Portfolio1 (138420) | 574 | 1,952% | 24.3% (equity) | 44.1% | 45 symbols incl. crypto, FX and gold |
| Commodity Accumulation Plan (2270674) | 505 | 390% | 19.6% (equity) | 39.3% | Six commodities, long only, 0.01 lot every year |
| ICTrading Signal 2 (926651) | 477 | 494% | 18.9% (equity) | 5.8% | Ten FX crosses; more withdrawn than deposited |
| SCR EURAUD (545538) | 296 | 288% | 11.5% (equity) | 15.6% | Stop loss on every position (provider) |
| Waka Waka EA (995287) | 438 | −95% | 91.1% (equity) | 184.6% | Grid on three pairs (user review); record collapsed |
| JINNIUFX7 (2330430) | 526 | 685% | 6.6% (equity) | 8.4% | GBPUSD only |
| MySingalStart 3 (1760039) | 488 | 540% | 21.7% (equity) | 3.8% | Seven FX crosses |
| CP1 (733875) | 414 | 1,086% | 11.1% balance / 37.2% equity | 8.7% | Losing positions held open (equity drawdown far above balance drawdown) |
| TP TRADER (1932739) | 359 | 2,852% | 2.0% (equity) | 6% | Returns rose sharply from 2024 after gold was added |
| Vektor (2300556) | 537 | 1,237% | 36.4% (equity) | 34.9% | No trades in 2020–2024 |
| Torus (879621) | 408 | 284% | 14.3% (equity), 83.7% max | 98.3% | One year (2022, +516%) carries the record |
How lot sizes changed over the years
| Signal (ID) | First year → latest year (median / largest lot) | Pattern |
|---|---|---|
| Commodity Accumulation Plan (2270674) | 2017: 0.01 / 0.01 → 2026: 0.01 / 0.01 | Fixed lot |
| ICTrading Signal 2 (926651) | 2017: 0.01 / 0.01 → 2026: 0.01 / 0.02 | Almost fixed |
| MySingalStart 3 (1760039) | 2017: 0.01 / 0.01 → 2026: 0.01 / 0.06 | Fixed base, adds up to 10× |
| JINNIUFX7 (2330430) | 2016: 0.01 / 0.16 → 2026: 0.02 / 1.28 | Fixed base, adds up to 64× |
| Torus (879621) | 2019: 0.21 / 7 → 2026: 0.26 / 1.2 | Roughly fixed median |
| Waka Waka EA (995287) | 2018: 0.02 / 0.24 → 2026: 0.05 / 2.62 | Adds up to about 30× the median |
| Vektor (2300556) | 2016: 0.02 / 0.72 → 2026: 0.01 / 0.36 | Adds up to 144× in 2017 |
| SFE Portfolio1 (138420) | 2015: 0.01 / 0.01 → 2026: 0.12 / 100 | Median ×10 from 2022; a few very large deals |
| CP1 (733875) | 2018: 2.17 / 2.72 → 2022: 49.7 / 194 → 2026: 0.04 / 21 | Lot follows the balance; cut after large withdrawals |
| TP TRADER (1932739) | 2019: 0.01 / 0.1 → 2026: 1.15 / 1.15 | Step-ups in 2022 and 2024 |
The trade history of SCR EURAUD (545538) could not be exported on 5 November 2026.
Other platforms
| Record | Investable since | Return a year (gross) | Investors' money |
|---|---|---|---|
| Darwinex THA | Jan 2017 | 20.35% | US$9.8 million, 445 investors |
| Darwinex SYO | Feb 2018 | 15.43% | US$12.8 million, 552 investors |
| Darwinex TRO | Aug 2017 | 7.01% | US$0.7 million, 23 investors |
| Collective2 Sage Volatility IRA | May 2016 | 22.2% (11.6% excluding fees; labelled hypothetical); max drawdown 35% | – |
All three Darwinex records trade under the platform's monthly risk target (VaR 95% of 6.5%). Investors' money follows the return: the two records above 15% a year hold far more than the one at 7%.
5. What the survivors share
| Trait | Survivors | Closed or collapsed |
|---|---|---|
| Leverage and margin use | Low (ICTrading 5.8%, JINNIUFX7 8.4%) | High (Waka Waka 184.6%; leverage raises fund death risk) |
| Volatility | Lower | Higher |
| Concentration | Spread across markets or a fixed small risk per trade | One market or a few names (Archegos, Amaranth) |
| Position sizing | Fixed base lot for years, or lot that follows the balance | Large averaging adds on a small base |
| Stops | A stop on each trade (SCR EURAUD) | Grid or averaging with no hard stop (Waka Waka) |
| Operations | Clean, verified records | Half of fund failures are operational |
The same few things end both funds and public records: too much leverage, one large bet, high volatility and weak operations. The long-running records in this sample are modest: about 18–32% a year with 11–25% relative drawdown for the older MQL5 signals read on 1 November.
6. Lessons for reading a track record
- Check the real start date. All six MQL5 signals read on 3 November show trade history from before their registration as signals; three registered in 2025 with history back to 2016–2019. Age on a signal page can include history published after the fact.
- Look at equity, not only balance. CP1 shows 11% drawdown on balance but 37% on equity, because losing positions stay open.
- Read money in and money out. Time-weighted growth on an account with large deposits and withdrawals says little about money made.
- Watch margin use and lot growth. Peak margin use above 50% and adds of 30× or more on a small base go with the records that collapsed.
- Check for gap years and regime changes. A long label can hide years without trades (Vektor) or a recent jump in risk (TP TRADER).
7. Limitations
- Only records still listed were read, so the signal sample has survivor bias.
- The fund studies are old (TASS 1994–2004, Capco 2003) or regional (Nordic 2015–2026).
- The family office rules come from secondary summaries.
- Traits are associations, not proven causes.
Sources
- Getmansky, Lo and Mei (2004), "Sifting Through the Wreckage": PDF
- Liang (2000), "Hedge Funds: The Living and the Dead": PDF
- Guzun (2026), "The Illusion of Longevity", HedgeNordic: link
- HFR Market Microstructure Industry Report 2Q 2025: PDF
- Capco study (Hedgeweek, 2003): link
- ESMA (2022), "Leverage and derivatives – the case of Archegos": PDF
- Till (2008), "Amaranth Lessons Thus Far": PDF
- WealthBriefing (2011), Soros converts hedge fund to family office: link
- UBS Global Family Office Report 2025: link
- Singapore 13O and DIFC summaries (wiki.private.law): 13O · DIFC
- Darwinex: THA · SYO · TRO; Collective2: Sage Volatility IRA
- MQL5 signals: 138420 · 2270674 · 926651 · 545538 · 995287 · 2330430 · 1760039 · 733875 · 1932739 · 2300556 · 879621
Declarations
- Author: Huyen X. Vu, Quantex Research, AlgoTrade Pte. Ltd., Singapore (ORCID 0000-0002-2209-842X).
- Data availability: All figures come from the public sources listed above, read on the dates given.
- Use of AI tools: Data collection, analysis 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.
- Funding: No external funding.
- Peer review: Not yet peer reviewed. Comments are welcome at [email protected].
- How to cite: Vu, H. X. (2026). How long do funds and trading signals survive? What the survivors share. Quantex Research, AlgoTrade Pte. Ltd. https://quantex.com.sg/research/how-funds-and-signals-survive/