User guide: How to anayze attribution metrics

Summary

Conduct detailed attribution analysis

About

AI Dragon provides an attribution module that decomposes excess returns (of a portfolio versus a benchmark) into factor returns and stock specific idiosyncratic returns. Factor returns can be decomposed into Style, Industry, Country and Currency. These can be further decomposed, for example Style can be decomposed into common factors (such as Beta, Dividend Yield, Low Vol, etc as well as the AI signal). The decomposition focuses both on the exposure to factors as well as the resulting excess returns. Other aspects considered include risk, Sharpe ratios, and hit rates as a result of the factor exposures.

Sequence of steps

1. Users can click return decomposition to see a breakdown of excess return into:
   1. Factor returns
   2. Stock specific idiosyncratic returns
2. Factor returns can be broken down Style, Industry, Country, Currency and Intercept.
3. Style can be broken down into:
   1. AI signal
   2. Beta
   3. Dividend Yield
   4. ESG
   5. Growth
   6. Liquidity
   7. Momentum
   8. Quality
   9. Residual Volatility
   10. Size
   11. Small Cap
   12. Value
4. Users can click on charts to see the exposure changing over time, along with cumulative excess returns.