Conservative
£110,125
final equity (no haircut)
£106,058
with 40% haircut
−6.19%
max drawdown
92.4%
days underwater
Best month 2024-03 (+3.73%) · worst 2026-05 (-3.45%) · flat-quarter odds 44.2% · worst 21-day start -4.09%
Balanced
£120,448
final equity (no haircut)
£112,232
with 40% haircut
−12.09%
max drawdown
92.6%
days underwater
Best month 2024-03 (+7.59%) · worst 2026-05 (-6.81%) · flat-quarter odds 44.2% · worst 21-day start -8.03%
Aggressive
£161,028
final equity (no haircut)
£137,670
with 40% haircut
−33.59%
max drawdown
92.8%
days underwater
Best month 2024-03 (+24.27%) · worst 2026-05 (-19.33%) · flat-quarter odds 44.9% · worst 21-day start -22.43%
The haircut table — the realistic read
Our methodology assumes live execution lands 20–40% below simulation (multiple-testing bias, spreads, slippage, regime change). The 40% rows are the conservative planning case.
| Profile | Haircut | Final | Total | Annualized | Max DD | Longest UW |
|---|---|---|---|---|---|---|
| Conservative | 0% | £110,125 | 10.12% | 1.37% | −6.19% | 522d |
| Conservative | 20% | £108,089 | 8.09% | 1.1% | −4.98% | 522d |
| Conservative | 30% | £107,073 | 7.07% | 0.97% | −4.37% | 522d |
| Conservative | 40% | £106,058 | 6.06% | 0.83% | −3.75% | 522d |
| Balanced | 0% | £120,448 | 20.45% | 2.66% | −12.09% | 525d |
| Balanced | 20% | £116,341 | 16.34% | 2.16% | −9.76% | 525d |
| Balanced | 30% | £114,286 | 14.29% | 1.9% | −8.58% | 525d |
| Balanced | 40% | £112,232 | 12.23% | 1.64% | −7.39% | 522d |
| Aggressive | 0% | £161,028 | 61.03% | 6.95% | −33.59% | 525d |
| Aggressive | 20% | £149,734 | 49.73% | 5.86% | −27.07% | 525d |
| Aggressive | 30% | £143,778 | 43.78% | 5.25% | −23.97% | 525d |
| Aggressive | 40% | £137,670 | 37.67% | 4.61% | −20.83% | 525d |
Stress scenarios
- Anchor (USDJPY ADX-Regime) underperforms. The anchor contributes the largest trade count (N=510 broker-validated). Its OOS decay of 1.04 is the best evidence in the book, but a regime break would push the portfolio toward the Conservative profile's return shape.
- NAS100 stack pauses. Four of nine systems express NAS100 exposure under one cluster cap (2 positions). Pausing the stack removes the cluster's contribution and roughly halves trade frequency.
- Spreads double. Approximated by the 20-40% haircut variants below — spread costs scale with trade count, hitting the H4/H1 systems hardest.
- BTC disabled. BTCUSD VWTSM is probation, half-risk and regime-dependent (full-window PF 1.51). Removing it barely changes the Conservative/Balanced shapes.
- First 30 days are unlucky. Computed per profile as the worst contiguous 21-trading-day start in the series (see worstStart21dPct).
- Probation strategies removed. Conservative IS that scenario — it runs core-only by definition.
What this is — and is not
Engine-based historical simulation (V20-era stack) on institutional data, 1,787 trading days 2021-07 to 2026-06. Equity rescaled from the £10k research run to £100k (percentages identical). Haircut variants scale daily returns by 20/30/40% to discount multiple-testing bias — our documented live expectation. This is research, not live performance and not a forecast.
Source artifacts: results/portfolio_5yr/portfolio_5yr_results.json · results/portfolio_5yr/daily_*.csv · results/REMEDIATION_V21_RESULTS.md. Full study: QUANTCOREX_100K_SIMULATION_STUDY.md. Read the risk disclosure before drawing conclusions — leveraged trading is high-risk and simulations are not guarantees.