Research Methodology

Rigorous Historical Validation Methodology

Backtesting is not proof of a working strategy — it is one step in a process of disconfirmation. Our methodology is designed to find and expose weaknesses, not to produce flattering metrics.

Seek to Disprove

Every backtest is designed with the goal of disproving the hypothesis. If the strategy survives rigorous scrutiny, the results become more meaningful — not less.

Out-of-Sample First

In-sample results are used for development only. All reported performance figures are validated on held-out data the strategy was never optimised against.

Realistic Assumptions

Slippage, commission, and bid-ask spread are modelled explicitly. Performance figures assume realistic fill conditions, not theoretical best-case executions.

Performance Metrics Explained

Every metric reported has a precise definition and an explicit interpretation context.

Total Return

(Final Equity − Initial Equity) / Initial Equity × 100

Overall percentage gain or loss over the backtest period. Must be evaluated alongside risk metrics, not in isolation.

Profit Factor

Gross Profits / Gross Losses

A ratio above 1.0 means the strategy generates more gross profit than gross loss. Values above 1.5 are generally considered meaningful; above 2.0 is strong but requires out-of-sample confirmation.

Sharpe Ratio

(Mean Return − Risk-Free Rate) / Std Dev of Returns

Risk-adjusted return per unit of volatility. Above 1.0 is acceptable; above 1.5 is good; above 2.0 is excellent. Computed on monthly equity returns.

Maximum Drawdown

Max (Peak − Trough) / Peak × 100

The largest peak-to-trough equity decline. A critical risk metric — a strategy with high returns but severe drawdown may not be psychologically or financially sustainable.

Win Rate

Winning Trades / Total Trades × 100

Percentage of trades that closed profitably. Must be considered alongside average win/loss ratio. A 40% win rate with a 3:1 R:R can be more profitable than a 65% win rate with 1:1 R:R.

Average Win / Loss Ratio

Avg Winning Trade P&L / |Avg Losing Trade P&L|

How much the average winner earns relative to the average loser. Combined with win rate, this determines long-run expectancy.

Four-Phase Validation Framework

In-Sample Development

Jan 2021 – Dec 2023

Strategy parameters are developed and selected using this window. Results here represent the best-case and are always subject to overfitting scrutiny.

In-sample only — used for development, not evaluation.

Out-of-Sample Validation

Jan 2024 – Dec 2025

The strategy is run unchanged on this held-out period to validate that results are not purely artefacts of curve-fitting to historical data.

Critical validation window — determines if the strategy generalises.

Walk-Forward Analysis

Rolling 12-month windows

Sequential optimisation and forward-test cycles to assess parameter stability across different market regimes.

Robustness check — looks for parameter sensitivity and regime dependence.

Monte Carlo Simulation

10,000+ random trade sequences

Random resampling of the trade sequence to estimate the distribution of possible outcomes, max drawdown risk, and probability of ruin.

Stress test — reveals tail risk hidden in sequential backtest results.

Backtest Results — All Strategies

Full backtest: Jan 2021 – Dec 2025 · Silver Futures (SI) · 1H bars · $100,000 initial equity

StrategyReturnProfit FactorSharpeMax DrawdownWin RateAvg W:LTradesAvg Duration

ST-Base

Base Supertrend System

+45.2%1.350.85-18.4%42.5%1.802454.5 days

ST-Multi

Multiple Confirmation System

+62.8%1.681.25-12.1%48.2%1.901686.2 days

ST-TSL

Target/Stop-loss Integration

+55.4%2.101.65-8.5%52.4%2.001852.1 days
Important caveat: These results are historical backtest simulations only. They incorporate modelled slippage (0.01% per side) and commission ($4 per round turn). Results would differ under live market conditions. This data does not constitute a solicitation to trade.

Explore the Full Dashboard

See all strategies, equity curves, and trade logs in the interactive research dashboard.