Risk Level Attribution

 The risk level attribution add-on shows the impact of changing risk levels on performance.

Actual Performance

The first column, Actual, shows the actual realized values for the time period chosen for the screen,

• Mean VaR(t-1): This is the average VaR from the start of the chosen date range, to the day before the end of the date range.

• Daily Standard Deviation: This is the daily standard deviation or realized returns. It is not annualized.

• Worst Day: Worst one day return.

• Sharpe: Sharpe ratio for the chosen period.

• Cumulative Return: This is not annualized.

Constant VaR

The second column, Const, shows what performance would have been, if the VaR of the portfolio had been held constant. To do this, we scale up or down the realized P&L, based on the VaR going into the day (this is equal to the VaR at the end of the previous day). As an example, if the mean VaR was 1.00%, and the VaR was 0.50% on a day when the fund made 0.40%, we would set the hypothical P&L to 0.80%,

[actual P&L] x [adjustment factor] = [hypothetical P&L]

0.40% x (1.00% /0.50%) = 0.80%

We are assuming zero trading costs, and the ability to adjust the portfolio exposures instantaneously. To make the analysis more robust, for portfolios where the VaR has changed significantly over time, the adjustment factor is limited to a range of 0.01 to 10.

1.1x Constant VaR

The third column, 1.1x Const, shows what the P&L would have been if the fund had kept VaR at 1.1x the mean VaR. This could be interesting for two reasons.

• If a fund cares more about its maximum risk, it might be willing to run a higher VaR on average, if VaR varies less from day to day.

• Because cumulative P&L is not additive, applying more or less leverage, might increase or decrease the cumulative return and Sharpe ratio.