234 CHAPTER ◆ 2 5 STAGE 4: Overview
25.6. LOOP 3: Assess Value at Risk
STEP 1: Choose VaR Methodology
STEP 2: Benchmark VaR Calculations and Software
STEP 3: Perform SPC on VaR Metrics
STEP 4: Determine Causes of Variation in VaRSingle performance and attribution analysis quantifies only what has happened in the
past. Risk is the potential for loss in the future, so a risk manager needs a forward look-
ing methodology. Value at Risk is just that; it helps predict how a portfolio will behave
in potential shocks. However, without attribution a manager ’ s ability to control risk is
limited since there can be little definition of which risks to hedge (credit risk for equities,
sector risk for bonds, etc.). A key point here is that a firm may or may not hedge an item
that is underweighted relative to the benchmark weight since underweighting is itself a
form of hedging. With VaR, simulations can be done to determine where the stress points
in a portfolio are so they can either modify the portfolio or hedge the risk. Also, risk
managers use VaR to determine risk-based margins (which is why traders are generally
hesitant to help the risk managers build better systems).25.7. Deliverables
● Periodic SPC charts and analysis reports.
● Return attribution and risk attribution report.
● Root cause analysis reports.
● Identification of future risk distributions.25.8. Summary
The primary difference between the classical view of risk and our view of risk in the
systematic trading/investment world is that the former attempts to manage a person (i.e.,
a trader or portfolio manager), the latter a machine. A machine does not change its behav-
ior after a stern conversation with a risk manager. A machine does not fear losses, nor
does it celebrate winning. A machine can effectively work forever if the inputs and the
economic environment were to stay constant. A machine simply changes raw data into a
working product.
In our view of risk, the machine has predetermined performance and predetermined
risk/reward ratios, which are stochastic outputs that should be monitored using SPC. The
purpose of SPC is to notify risk managers of a problem in the machine, to make them
aware of special, unassigned but assignable, variation in the system. The task of the risk
managers and the product team, then, is to determine the root cause of the special varia-
tion in the performance of the machine.
Our process is more complex than the traditional risk overview process due to the
complexities of continuously monitoring a process. In manufacturing, variation in proc-
esses is usually quite small, sometimes measured in thousandths of an inch. In trading,
variation is most often quite large. The demands on a risk manager should naturally be
more complex, since they should be part of a continuous improvement team. Management