Quantitative research course¶
The goal is to read strategy, backtest and risk results critically and identify their evidence and limits. Basic arithmetic is sufficient to start; learn statistics and programming alongside the lessons. This is an educational course, not investment advice. All numerical examples are synthetic.
Learning path¶
Begin with research questions, instruments and data. Continue with returns, statistics and factors, then signals, execution, allocation and models. Finish with rotation research and paper-trading review. The application's Learning Space provides chapter navigation and previous/next controls; the same lessons are available in MkDocs.
| Chapter | Question to answer |
|---|---|
| 01 Research | How do hypotheses, validation and execution differ? |
| 02 Instruments | Why are stocks, indexes and ETFs different? |
| 03 Data | How do adjustments and historical membership affect a result? |
| 04 Returns | Why should returns be compounded? |
| 05 Statistics | What can correlation and drawdown establish? |
| 06 Factors | How does IC differ from trading returns? |
| 07 Signals | How are trend, reversal and rotation rules defined? |
| 08 Execution | How are fills, costs and biases checked? |
| 09 Portfolio risk | How do target weights become actual positions? |
| 10 Models | What makes a result out of sample? |
| 11 Rotation | What historical evidence does a rotation universe require? |
| 12 Operations | Does successful paper trading imply live readiness? |
A minimal synthetic study¶
At time t, compare the previous 20 trading days of returns for three assets. Hold equal weights in assets with positive returns and attempt execution at t+1. If none qualifies, hold cash. Retain the universe available at the decision time, input bars, signals, target weights, orders, fills and costs. Check selection and actual purchases separately.
Pitfalls and exercise¶
A strong return does not establish a hypothesis. A completed task does not establish historical coverage, and mock mode is not real market data. Check definitions before increasing algorithmic complexity.
Exercise: Can a NAV screenshot without instruments, dates or trade records reproduce a study?
Answer: No. At minimum retain data source, availability times, adjustments, parameters and versions, requested and actual dates, benchmark, orders, fills and costs. The next chapter organizes these into a research process.