The appraisal project is a thorough review and assessment of the “non-investment” risks present at a hedge fund.
The “non-investment” risks are defined as the risks linked to the implementation of the fund’s investment strategy. Those risks encompass operational risks, fraud risk, business risk and legal risk.
The objective
The objective of the project is to identify areas of vulnerability which may lead to the occurrence of a low probability high impact event.
The process
Our appraisal process is based in two interconnected pillars: the information collection and the information analysis.
The quality and type of information used to appraise a fund is critical (“Garbage In, Garbage Out”).
In designing our process, we have been very careful at strikingthe right balance between information quality versus information quantity. The devil is sometimes in the details but the details may also lead to information over-kill.
We focus on collecting actionable information which will allow us to reach the right conclusions. We have identified a set of proxies which provide us an indirect way for checking and verifying relevant information.
Our data collection is based on four different sources of information:
The information analysis is supported by our proprietary Appraisal database. The database is used to record, organize and analyze the large quantity of information that we collect on the funds (on average 300 fields). In order to control for the information independence and integrity, each piece of information in the database is linked its source (document, call recording, onsite visit report …). The data is organized around the following categories:
Risk management (staff, methodology, oversight, independence, system ….)
Service providers (relationship duration, role confirmation, service provided …)
We perform a quantitative analysis of the fund in order to check statistics like correlation, volatility, returns … for inconsistency with the manager’s peer group. We also look at indicator for return smoothing and price manipulation like the auto-correlation and the bias ratio.