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

181

regression testing to ensure that errors are detected and fixed. Our methodology assumes

that regression testing is built from the successful white, gray, and black box test cases

used during testing of prototypes (K | V 1.4).

While we have devoted only one loop to this section, we fully expect and advocate

that the tasks be broken down into several iterations using an agile framework. In the end,

the goal is to create a stable, fast running code that correctly implements the business

logic. In this stage the development team performs unit tests on all software components.

19.4. LOOP 2: Build and Test Interfaces

STEP 1: Document Interface Requirements

STEP 2: Design Interface Packages

STEP 3: Program Interface Packages

STEP 4: Test Interfaces

Loop 2 integrates the trading/investment algorithms (i.e., the business logic tier) with

external software components and the user interfaces. Most of these external software

components have been purchased from vendors in Stage 2 and the product team is respon-

sible for ensuring that COTS products conform to the specified requirements. External

components include databases, potentially an optimization engine, and execution, risk,

and accounting software. Integration into the existing technological environment should

already be well understood; the data dictionary, data maps, and data flow maps created in

Stage 2 should facilitate the integration process.

For integration with external components, the development team will also design, code,

and test exception handling mechanisms in this step, including recovery procedures in the

event of external errors. Where recovery is possible, the trading/investment system should

be put into a fail-safe mode (such a mode of operation may reduce response times, but

service only good elements of the system). This offers a sort of graceful shutdown, or fault

tolerant operation, where full and safe performance is maintained in the presence of faults.

Trading/investment systems that use real-time data and/or automate execution must be

planned and built with safety in mind, concerned with the consequences of failure. Fail-

safe systems make no effort to meet normal operational requirements. Instead they aim to

limit the danger or damage caused by fault.

19.5. LOOP 3: Create Network Architecture

STEP 1: Document Hardware and Network Requirements

STEP 2: Design Network Architecture

STEP 3: Buy/Build Network Infrastructure Components

STEP 4: Perform Acceptance Test

Loop 3 assembles and connects the hardware and network components that will house the

trading/investment system. Now, hardware changes quickly. In fact, computer processing

power doubles every 18 months, according to Moore ’ s law. Given that trading/investment

system development may take several months to complete, hardware available in Stage

1, or even in Loop 1 of this stage, may be outdated by the time development is complete.

For this reason, hardware development should come last.

19.5. LOOP 3: CREATE NETWORK ARCHITECTURE