Practical notes on strategy validation, automation and risk management.
HaoQuantX Research · Method note · Updated 2026-08-27
Scope: a decision framework, not an empirical performance claim. No dataset or return estimate is used in this note.
Separate evidence from execution
Automation repeats a rule; it does not prove that the rule has an edge. Define the market, time frame, costs and failure conditions before evaluating a result.
Use an explicit promotion gate
Move from research to Demo only after the backtest configuration is reproducible. Move from Demo to Live only after exposure, protection orders, daily limits and operational recovery steps are understood.
Record the reason for the decision
Keep the parameter version, cost assumptions, test window and approval reason together. A later change to any of them should trigger a new review rather than inheriting the old decision.
HaoQuantX Research · Reproducibility checklist · Updated 2026-08-27
Scope: a reusable checklist. Results still depend on the selected data, instrument, parameters and execution assumptions.
Identify the test
Record exchange, instrument, interval, data start and end, strategy version, parameters, initial capital and leverage. Without this context, a result cannot be reproduced or compared.
Model executable costs
Include maker and taker fees, slippage, funding and position-sizing rules. State how missing candles, partial fills and order rejections are handled.
Test outside the fitted sample
Keep a separate out-of-sample period, compare behavior across market regimes and years, and review trade count, net expectancy, drawdown and liquidation events alongside return.
HaoQuantX Research · Interpretation guide · Updated 2026-08-27
Scope: metric interpretation, not a recommendation or forecast. ROI alone does not establish repeatability.
Start with the denominator
Confirm whether ROI uses initial capital, deployed margin or peak capital. Compare strategies only when the capital and leverage definitions match.
Read return with risk and sample size
Review net costs, maximum drawdown, trade count, net expectancy and liquidation history. A high return from a small sample or concentrated exposure is weak evidence.
Look for stability, not one best run
Compare out-of-sample and year-by-year behavior, then test nearby parameters. Large deterioration under small changes is a warning that the headline ROI may be fitted to one period.
HaoQuantX