Research
Fazen Portfolio v3.0
XAUUSD · NAS100 · BTCUSD · ETHUSD — multi-asset momentum & breakout · 2024-01-02 → 2026-08-03 (2.58 years)
Total return
+248.7%
CAGR
+62.1%
Max drawdown
-13.27%
Calmar
4.68
Sharpe
2.26
Sortino
2.78
Positive months
74.2%
Monthly vol
6.17%
Quarters
| Period | Return | Index |
|---|---|---|
| 2024-Q1 | +4.28% | 104.3 |
| 2024-Q2 | -0.71% | 103.5 |
| 2024-Q3 | +17.65% | 121.8 |
| 2024-Q4 | +23.53% | 150.5 |
| 2025-Q1 | +13.16% | 170.3 |
| 2025-Q2 | +11.47% | 189.8 |
| 2025-Q3 | +16.30% | 220.8 |
| 2025-Q4 | +5.43% | 232.8 |
| 2026-Q1 | +29.50% | 301.4 |
| 2026-Q2 | +29.54% | 390.4 |
| 2026-Q3 | -10.69% | 348.7 |
Monthly (%)
| Year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec | Year |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2024 | · | -2.0s | -1.9s | +3.0s | -4.0s | +0.4s | +10.2s | +2.2s | +4.5s | +3.7s | +2.9s | +15.8s | +38.7 |
| 2025 | +8.0s | -0.5s | +5.3s | +8.3s | -4.4s | +7.6s | +2.3s | +5.2s | +8.1s | +10.2s | -3.7s | -0.7s | +54.7 |
| 2026 | +8.5s | +4.2s | +14.5s | +10.3s | +13.9s | +3.1s | -11.9s | +1.4 | · | · | · | · | +49.8 |
Out of sample, it does not hold
The same default parameters, never tuned, run over four years the series above does not cover. If the result were an edge rather than a fit, it would show up here too.
| Year | Sharpe | On deposit |
|---|---|---|
| 2020 | +0.16 | +3.6% |
| 2021 | -1.72 | -31.5% |
| 2022 | +1.39 | +37.9% |
| 2023 | -0.36 | -6.6% |
| 4 years | -0.13 avg | +3.5% |
Out-of-sample Sharpe ÷ in-sample Sharpe = -0.06. The accepted threshold is 0.7; below 0.3 is curve fitting. This ratio is negative — two of four years lose money, and 2021 gives back nearly a third of the capital, against the 13.27% maximum drawdown measured over the good period shown above. Four years add up to +3.5%: flat, with a -31.5% inside it.
What it does not mean: that the strategy is useless. 2022 does Sharpe 1.39 and +37.9% — in some conditions it works, and works well. What is missing is the part that decides when to stay out. And it is not parameter overfitting: these are the defaults, never searched.
The problem is drawdown, not regime
The obvious defence would be that the strategy only works after some structural change in 2024. Eight quarters of 2022-2023, same defaults, say otherwise: six of the eight are positive and all of them predate 2024. 2023Q1 is the best quarter of the eight and beats the 2024-2026 in-sample average.
| Quarter | Sharpe | Trades |
|---|---|---|
| 2022Q1 | +1.74 | 111 |
| 2022Q2 | +0.50 | 114 |
| 2022Q3 | +1.90 | 125 |
| 2022Q4 | +1.27 | 121 |
| 2023Q1 | +2.60 | 106 |
| 2023Q2 | -3.94 | 116 |
| 2023Q3 | -1.88 | 117 |
| 2023Q4 | +1.88 | 125 |
There is no step up in 2024, there is a hole before it. Expectancy is positive almost everywhere, interrupted by rare and severe episodes: 2023Q2-Q3 together give back about 86% of everything the four quarters of 2022 earned, and 2021 costs a third of the capital.
The trade count stays flat — 106 to 125 per quarter — straight through the losing stretches. The strategy does not stop when it stops working. It keeps trading, and loses.
So the open question is not which window to show. It is whether those episodes are detectable while they happen rather than afterwards. If they are, a brake on drawdown turns an episodic strategy into a product, and the useful sample becomes 2022-2026 rather than 2024-2026. If they are not, it does not become one — and this page will say so.
Metrics are computed on the compounded monthly returns of the index; Sharpe and Sortino are annualised (×√12) with the risk-free rate assumed to be zero, therefore overstated against a calculation with a real risk-free rate.
Maximum drawdown is measured on the daily end-of-day index, not on intraday equity: the real intraday drawdown is deeper. The opening month of the series is partial and is excluded from every statistic and from the monthly table; the closing month and quarter are partial too and are shown as measured, so the last cell of each table covers fewer days than the ones before it.
No monetary value is reported, by choice: a percentage is the only measure comparable across accounts of different size, and the only one coherent with a proportional allocation.
Source: MT5 Strategy Tester report over 2024-01-02 → 2026-08-03 — 1,249 trades, 2,498 deals, 4 instruments. Out-of-sample and quarterly runs use the same defaults on a 100,000 deposit.
Trading involves risk of loss. Nothing on this page is investment advice, an offer, or a solicitation, and none of it is a promise of future results.
How an allocation would work
If the drawdown brake holds up, the strategy is not something you download and install — you connect an account and it trades in proportion to your capital. That mechanism, and what it would take before it is offered, is described in full.
How the allocation works