Skip to main content
All CollectionsAnalytics
Risk - Single Factor and Multi Factor
Risk - Single Factor and Multi Factor
Updated over 3 weeks ago

As discussed in the Risk - Beta article, the Novus Platform risk analysis tool provides deep insight into your portfolio’s risk exposures as well as customized stress/scenario analysis available on both a single-factor and multi-factor basis. 

Single Factor 

For the Platform’s single-factor sensitivity analysis, the portfolio as of the selected date is run against multiple single-factor models for each factor.  

R-sq = Fit Score
Alpha = The portion of return unexplained by beta to the benchmark (Y intercept of the regression; based on the fund/portfolio’s return stream)
Alpha P-Val = Tests the significance of the output. - what tests generate this?
Beta =  Sensitivity to a benchmark (the slope of the regression line; based on the fund/portfolio’s return stream)
Beta P-Val = Tests the significance of the output.
Regression Vol = Volatility of the projected returns in the portfolio (the “predicted Y values”).
Residual Vol = Volatility of the regression’s error term. - what are these showing?

Running the analysis updates the portfolio return chart. The portfolio return chart shows the expected run with respect to fixed-length factor moves.  

Multi-Factor Model 

 
For the Platform’s multi-factor sensitivity analysis, the portfolio is run against the multi-factor model over the selected time interval.  

 
The Current Factor Exposures table shows the adjusted R-squared (Adj-R-Sq) fit statistic, alpha and factor sensitivities at the analysis end date. To increase Adj R-Sq, you can add more factors or remove ones that don’t make sense to include in the model. Taking the market and adding one factor at a time is the best way to find the ideal model. 

However, you shouldn’t regress a portfolio’s returns against factors whose performance is in the relatively same direction or close to it as this will often lead to insignificant regression results.  

 
For position-level data, the model will look at the betas of the underlying positions, and weight the betas by the position’s size to calculate the long/short/gross numbers at the portfolio level. 

 

Historical Factor Contribution

Takes into account the portfolio exposure (beta) to each of their factors and multiplies by performance of the factor to calculate the contribution that came from each factor. When you hover over any of these data points, you can conclude a 1% move in the factor caused an X% move in the portfolio’s return for that month.

 

Historical Factor Exposures

 The beta coefficient of each factor in the chosen model (the portfolio’s exposure to each of the factors).

Factor Exposures P-Values 

Tests the significance of the regression output. A p-value below 10% means we can be 90% confident in the beta coefficient results.

Historical Factor Performance 

The compounded factor performance of the factors in the chosen model, independent of the portfolio.

If you're interested in reading about our other risk articles:

Did this answer your question?