A good backtest is not a pretty equity curve. It's a process that anyone with the same data could reproduce and arrive at the same conclusion.
If a strategy fails any of these tests, it does not enter paper trading.
任何一項過唔到，都唔會進入 paper test。
A classic 20/10 Donchian channel breakout on daily gold bars. Boring by design. Published here so you can see the whole process, not just the headline number.
經典 20/10 Donchian 通道突破，日線黃金。本意就係沉悶 — 放出嚟係為咗展示完整流程，唔係得個數字。
The Donchian system is deliberately not spectacular. A PF of 1.27 is what a real, cost-honest breakout strategy looks like on a liquid instrument over two decades. What matters is that it survives every robustness test we throw at it — including the chop regime where it underperforms but does not blow up.
呢個策略刻意唔係亮眼。PF 1.27 就係一個真實、誠實計成本嘅突破策略喺 20 年流動品種上應有嘅樣。重點係佢通過每一項穩健性測試 — 包括表現最差嘅震盪市，都只係輸少少，冇爆。
Because a single average R-multiple tells you almost nothing about whether a strategy is real.
因為單一嘅平均 R 值幾乎冇講到呢個策略係唔係真嘅。
We resample the trade ledger with replacement 10,000 times and compute the distribution of possible outcomes. This shows us the range of realities, not just the one we happened to observe.
If the 95% CI on mean R includes zero, the strategy's edge is statistically indistinguishable from luck. We reject it, no matter how pretty the equity curve looks.
If removing the two best trades collapses your profit factor below 1.15, your "system" was one or two lucky trades wearing a trench coat. Real edges are diffuse.
All results on this page are historical simulations of rule-based strategies against past market data. Past performance is not a reliable indicator of future results. Backtesting does not account for all real-world factors such as execution delays, requotes, weekend gaps, broker insolvency, or regulatory change. Strategies shown here are research examples, not trading recommendations. See the full disclaimer for details.
本頁所有結果為規則型策略於歷史市場數據上的模擬。過往表現不代表將來表現。回測無法完整反映執行延遲、重新報價、週末跳空、券商破產或監管變動等現實因素。頁面上的策略為研究範例，非交易建議。詳情見完整免責聲明。