In-sample vs. out-of-sample evaluation
Introduction to Portfolio Analysis in R
Kris Boudt
Professor, Free University Brussels & Amsterdam
Bad news: estimation error
- Limitation to data-driven portfolio allocation:
Bad news: estimation error
- Limitation to data-driven portfolio allocation:
Bad news: estimation error
- Limitation to data-driven portfolio allocation:
Good news: opportunities
- Do not ignore estimation error
- Use split-sample analysis to do a realistic evaluation of portfolio performance
Good news: opportunities
- Do not ignore estimation error
- Use split-sample analysis to do a realistic evaluation of portfolio performance
Good news: opportunities
- Do not ignore estimation error
- Use split-sample analysis to do a realistic evaluation of portfolio performance
Good news: opportunities
- Do not ignore estimation error
- Use split-sample analysis to do a realistic evaluation of portfolio performance
No look-ahead bias in optimized weights
- Split-sample design matches with the investor who:
No look-ahead bias in optimized weights
- Split-sample design matches with the investor who:
- Function
window()to do split-sample analysis in R
