# Introduction to the Infiniti Capital Four Moment Risk Decomposition

By Peter Urbani and Mitchell Bristow

One way to include the impact of higher moments into the calculation of Value at Risk (VaR) for such portfolios is to use the Cornish Fisher modification to the Normal VaR calculation. The empirical evidence indicates that only about 12% of Hedge Funds have returns that are normally distributed. Consequently the impact of higher moments is of great interest to people building portfolios of hedge funds or managing their risk. The Normal distribution can be full described by its first two moments. 2 .Higher Moments and Risk Classical Investment Theory assumes that investment returns are normally distributed. Hedge Fund returns exhibit significant amounts of skewness and excess ( > 3 ) kurtosis (third and fourth statistical moments). Investors generally have a preference for positive skewness and moderate to low levels of kurtosis. Mean (Mu) and Standard Deviation (Sigma).

Normal versus Modified VaR In the case of a standard normal distribution (Mean=0. Std Dev=1. Note this assumes Raw Kurtosis – Excels formula assumes Excess Kurtosis > 3 and subtracts 3 automatically. This is why Kurtosis is not 0 in the above example. 3 . Skew=0. Kurt=3) both the Normal Var and Cornish Fisher Modified VaR give the same answer.

Std Dev=1. Skew=0.Normal versus Modified VaR In the case of a slightly positive skewness and slightly higher than normal kurtosis (Mean=0. Kurt=4) the Cornish Fisher Modified VaR is lower (less negative) than that given by the Normal VaR calculation.5. 4 .

Std Dev=1. Kurt=4) the Cornish Fisher Modified VaR is higher (more negative) than that given by the Normal VaR calculation. 5 .Normal versus Modified VaR In the case of a slightly negative skewness and slightly higher than normal kurtosis (Mean=0.5. Skew=-0.

The Cornish Fisher Modification In Excel for use with Excess Kurtosis = (Skew*(ZScore^2-1)/6)+(Kurt*(ZScore^33*ZScore)/24)-((Skew^2)*(2*ZScore^3-5*ZScore)/36) for use with Raw Kurtosis =(1/6)*(ZScore^2-1)*Skew+(1/24)*((ZScore^3)3*ZScore)*(Kurt-3)-(1/36)*(2*Zscore^35*ZScore)*Skew^2 6 .

Moving from the Univariate to the Multivariate Weights Weights (normal) Variance Covariance Matrix Std Devs (normal) -> Correlation Matrix (normal) Variance -> Covariance Matrix Std Devs CoSkewn +ess (modified) -> Correlation Matrix Mod Std Devs CoKurtosis -> (normal) VaR -> (modified) VaR 7 .

Infiniti Capital Modified VaR Decomposition 8 .

Normal & Modified VaR for a portfolio 9 .

Basic Portfolio Statistics 10 .

Covariance Matrix 11 .

‘Modified’ Covariance Matrix 12 .

Correlation Matrix 13 .

‘Modified’ Correlation Matrix 14 .

‘Modified’ Volatility 15 .

Normal & Modified VaR from these Matrices 16 .

Infiniti Capital Four Moment Risk Decomposition http://www.com 17 .infiniti-analytics.

4% to the return. Whilst the differences between the Normal and Modified VaR for individual positions may be small in some cases. This is even more evident when we look at the Modified Conditional VaR (Modified CVaR) versus the Normal CVaR of -4.90%. Convertible Arbitrage. 18 .56%.51 contribute to the 95% Modified VaR being higher than the Normal VaR at -2.28% versus -1.0% to the Risk (Modified CVaR). Comparatively Emerging Markets was far riskier as although it contributed 19.a expected shortfall (ES)) tend to be far more pronounced.What can we learn from this Analysis ? First we can see that the portfolio’s negative Skew of -1.3% to return.28 and excess Kurtosis of +4. it also contributed 39. Thus we can see that largest % Contributions to the portfolio’s Modified CVaR came from Emerging Markets.09% versus -2. it also contributed 18. However when we look at ratio of % Contribution to Weight we can see that although Distressed contributed 18.k. those for the Normal and Modified CVaR (average tail expectation beyond the VaR a.6% to the Risk. Event Driven and Distressed.