Value at Risk (VaR)
Introduction

We provide Value at risk, plus stress testing and back testing. This also includes other risk measures such as DV01, CS01 etc.

For an overview of the different types of VaR, stress testing and back testing and for e-learning on the regulatory requirements, click here.

Whilst there are limitations to VaR, it has become accepted as a standard risk measure for regulatory and reporting purposes. For examples of where VaR is required for regulatory purposes and the regulatory requirements click here.

For VaR terminology and definitions click here.

Data Flows

We utilise third party risk vendors for VaR calculations and stress testing. This involves automated efficient processes. Confidentiality of client and client data is preserved throughout the process. Click here for an overview of data flows and system architecture.

Calculation Details

Confidence levels: All the VaR calculations are based on one day VaR at the following levels: 99%, 97.5% and 90%

Holding Periods: All the VaR analysis is based on one day VaR. the exception is that that for regulatory purposes we also calculate the 20-Day VaR. The calculated 20-day VaR is also used for comparing to the internal expected level of VaR, maximum VaR and regulatory limits. The 20 day VaR is calculated based on the 1 day VaR. For the calculation methodology click here.

Look back period:

Client data Flows

For this purpose, the data flows are

  • Anonymised holdings data is sent to the VaR provider
  • The VaR provider sends back 1-day VaR results plus stress test / scenario test results.

For further information on the data flows, click here.

Funds-Axis then perform

  • Data analysis
  • Calculation of 5 day and 20 day VaR, extrapolated from the 1 day VaR calculations – click here for details.
  • Produce the required reporting
  • Monitoring of actual VaR vs. portfolio expected and maximum VaR
  • Back-testing
Benchmark VaR

For relative VaR, you need to compare the VaR of the portfolio to the VaR of the benchmark, e.g. to the VaR of the FTSE 100.

The VaR of leading benchmarks ….

Stress testing / scenario testing

Typically we receive back a consistent set of stress test result for all funds from the VaR provider.

However, within the Funds-Axis reporting only the results relevant to the particular portfolio will be shown.

We receive data back for all portfolios for all of a series of stress tests. The stress tests are listed below. Whether or not a particular stress test is relevant for a particular portfolio can be recorded in the Portfolio Master. This has the impact of limiting the range of stress test results that will show against any portfolio.

Typically we receive back a consistent set of stress test result for all funds from the VaR provider.

However, within the Funds-Axis reporting only the results relevant to the particular portfolio will be shown.

For details of the stress tests applied as standard to different portfolio types, click here

Back testing

Basically VaR backtesting is comparing the 1 day VaR (not the 20 day VaR) to what actually happened on the portfolio from one day to next.

The VaR Backtesting compares:

  • The Portfolio 1 day VaR, 99%, 250 day history
  • The portfolio actual return for that 1 day (e.g. from one day to next)

For more on back-testing, including the reporting and management of overshoots, click here.

Risk Limit Testing

The regulations require monitoring and reporting of the consistency of the risk profile of the portfolio with the portfolios risk limit system. This can include monitoring against the expected level of VaR or the internal / regulatory maximum level’s of VaR.

This risk limit monitoring is done within the Funds-Axis application.

In the Portfolio Master, we record:

  • the expected level of VaR
  • the internal maximum level’s of VaR
  • the regulatory maximum level’s of VaR
Reports

There are a range of reports available In Zoho. The Sisense reports will replicate these.

There will also be a histogram report in Sisense.

For now Zoho Reports has been based on 3 uploaded tables, but the reports should all be reproducible from.