01 / PROBLEM STATEMENT
Why this work exists
Classical allocation relies on historical moments, while forecast-assisted allocation adds model risk. The project compares both ideas within one educational workflow.
Data and source material
- Historical market data retrieved with yfinance
- Processed return series for portfolio construction
- Generated portfolio weights and risk/return analytics
02 / SOLUTION OVERVIEW
From input to usable output
Compare efficient-frontier allocation with ML-assisted strategies using common analytics.
- 01
Load and preprocess historical prices
- 02
Construct mean–variance and efficient-frontier portfolios
- 03
Train return-forecasting models
- 04
Generate ML-assisted allocations
- 05
Compare return, volatility, and Sharpe analytics
03 / TECHNICAL ARCHITECTURE
Implemented components
Market data → preprocessing → classical/ML branches → common evaluation
04 / METHODOLOGY & IMPLEMENTATION
Decisions that shape the system
- SciPy supports constrained mean–variance optimization.
- scikit-learn models supply experimental return forecasts.
- A common analytics layer compares classical and forecast-assisted allocations.
Implemented capabilities
- Mean-variance
- Efficient frontier
- Return forecasting
- Sharpe and volatility
05 / RESULTS & EVIDENCE
What the source supports
- The repository documents implemented classical and ML-assisted branches.
- It does not publish a tracked, date-bounded out-of-sample result table, so no performance advantage is claimed.
Results preserve the repository’s units, evaluation framing, and stated limitations. Missing evidence is not converted into a claim.
06 / CHALLENGES & LIMITATIONS
Where the evidence stops
- The README does not establish chronological evaluation or trading-cost treatment.
- Future-work items such as Black–Litterman, reinforcement learning, deep learning, and regime switching are not represented as implemented.
07 / WHAT THIS DEMONSTRATES
Engineering signal
A comparative implementation of allocation paradigms and the discipline to withhold performance conclusions when tracked evidence is absent.
08 / SOURCE & RELATED WORK
Trace the evidence
3b56869995b057d4600395285026c8d989ad5774 ↗README.md · scripts/ · visualisation/