Home +44 (0) 20 3795 2330 info@promethion.co.uk
Home Financial Strategy Fintech & Emerging Technologies Business Processes Team Case Studies

Promethion Limited

20 St. Dunstan's Hill

London EC3R 8HL

+44 (0) 20 3795 2330

© Promethion Ltd 2024

Creation of the risk management framework for a credit hedge fund: expertise in risk management and asset allocation models.

Solution

Promethion worked with a leading academic to create an optimisation and Conditional Value-at-Risk (CVaR) model to measure fund exposures and integrated it into the client's risk procedures.

CVaR is the most suitable risk measure for this type of skewed return distribution.

The Optimiser minimises the CVaR of the portfolio return distribution subject to various constraints, including minimum expected return, portfolio weights and leverage.

The CVaR Optimiser allows the manager to quantify expected portfolio losses with a given confidence level; assess the risk in combining different types of asset; and reach conclusions on appropriate levels of leverage.

It provides a state-of-the-art solution which can be used to evaluate the fund's risk profile and gives comfort to bank counterparties, regulators and investors.  

Financial Impact

The model allowed the manager to estimate extreme tail risk in the portfolio.

Although the manager employed a wide range of strategies at any one time, the variable leverage factor means that there was exposure to negative spread movements and hence significant portfolio downside.

Despite extraordinary market conditions since launch, the fund was successful in minimising portfolio risk and performed in the top quartile of managers in this strategy.

Problem

The fund included a range of complex credit assets and strategies.

The client needed to measure the risk of the portfolio and examine the efficiency of combining different asset classes through an optimisation model.

Credit assets have non-symmetric return distributions (large probability of small earnings and a small probability of very large losses) and conventional VaR models and variance of portfolio return as used in the traditional Markowitz mean-variance framework are inappropriate in this environment.