Our Methodology
Every tutorial in the Build Your Own Quant Research System series follows the same honest research methodology:
- Real data, real costs. No fantasy fills. Every backtest uses adjusted prices and honest commission/slippage assumptions.
- Out-of-sample validation. Walk-forward analysis (Part 2) — in-sample is a story, out-of-sample is the truth.
- Selection bias correction. Deflated Sharpe and PBO (Part 4) — the best of 100 trials is mostly luck.
- Robustness over optimality. Plateaus, not peaks (Part 9) — boring beats brilliant.
- Full pipeline. Nine stages, one verdict (Part 10).
When a result is marginal, we say marginal. When the pipeline disagrees with itself, we show the disagreement. Data honesty is not a section — it is the whole method.