Can you trust your risk number?

In risk management a lot of focus and attention is (rightly) put on models and methodologies used to compute ex-ante risk measures. And in the context of a multi-asset class universe which is vast by nature, perfect data (market data, terms and conditions provided by the user) and bug-free algorithms are not always possible. Therefore, one of the key challenges for risk managers is to ensure that the portfolio risk analytics produced are sound and reliable. In other words: can we trust the number being produced?

In the case of simulation-based portfolio risk analytics (historical and Monte-Carlo) one can visually inspect the distribution the tails ensuring no spurious PnL is hidden there. This can be done when there is only a handful of portfolios but when there are many, it’s a Herculean Task.

In this white paper you will:

• Learn how to approach this ‘Herculean Task’
• Get guidance on a framework that helps to detect outliers in a high dimension space
• Explore two machine learning tree-based techniques to tackle the challenge

Extracting portfolio risk analytics insights 

The data used can be easily extracted from our investment risk management system, Axioma Risk. With it, the end user can implement a workflow that can automate detections.