Quality Money Management : Process Engineering and Best Practices for Systematic Trading and Investment

CHAPTER ◆ 2 5 STAGE 4: Overview

Portfolios of securities and derivatives require constant monitoring and so successful

implementation of a trading system necessitates that periodic reports be generated to deter-

mine if the system is performing to specifications. These reports will present the portfo-

lio performance statistics, risk calculations, attribution analysis, and Value at Risk (VaR).

Furthermore, quality control should determine if the system is in or out of control. If the

process is out of control, a determination of the cause of the variation from the expected

results and the risk managers should create an action plan to deal with those causes. The

three risk methodologies represent three different sets of tools for monitoring the system.

Processes in a stochastic world change regularly. They drift within a range of values

that are normally considered the upper and lower control limits. Sometimes these stochastic

processes change due to a real shift in the real world. For example, the reformulating of a

plastic ’ s chemistry due to an EPA ruling would change all parts made with the plastic. In

the same manner, changes in securities laws may cause real shifts, the way Sarbanes-Oxley

shifted volatility around earnings. The real question in systematic trading and investment is

“ Is the variation normal and expected? Or, is the process out of control due to a fundamental

shift in the world? ” More simply, “ Is this the same process the product team backtested? ”

Some risk managers may not catch credit. Interest sensitive stocks, bonds, and many derivatives all con-

tain a credit forecast embedded in their values. As a result, these instruments may become highly cor-

related during a credit shock. Algorithmic trading/investment requires careful real-time performance

monitoring as well as pre- and posttrade analysis to ensure the algorithms are properly applied. 1

Risk has always been at the heart of trading/investment systems. There is a difference

between calculating market risks for capital reserves and using risk to control or improve

a process. In the K | V methodology we propose that we use outputs of the risk calcula-

tions to improve the process instead of setting capital reserves only. Our methodology

differs from the classical use of risk calculation that are used to protect the trading firm

from rogue traders since in systematic trading/investment, the machine controls the proc-

ess, not human traders. We believe that a firm needs both views to be successful in the

competitive world of trading due to the continual drive to benchmark funds.

25.1. History of Risk and Return

Traditionally, a risk department calculates risks and returns separately from a trading

group to ensure a trading group stays within its limits. The risk department defines limits

through stochastic mathematics with the goal of ensuring that the firm is optimizing its

risk/return ratio. We group the measures of risk into three basic categories:

● Single performance. These classical measures of risk and return compare a portfo-

lio relative to itself. Examples of single performance calculations include: returns,

drawdowns, Sharpe ratio, Sortino ratio. These calculations do not forecast, only

document the current model outputs. We believe these outputs can be viewed as sto-

chastic, contrary to popular beliefs. We base this view on the low ability of current

systems to explain variation.

● Performance attribution. In the not so recent past, attribution has been introduced into

the world of risk and has now become a standard for most pension funds. Attribution

allows a fund manager to determine how he is performing relative to a benchmark—are

the individual security bets adding value? Or is it the over-/underweighting of sectors?

230