Development of Algorithmic Investment Systems

Key Insight

Following a financial setback, the founder recognized the profound value of computers in thinking and analysis, which proved essential for the company's success. Initially, early microcomputers were used for econometric analysis, applying statistics to economic data with the theoretical belief that a perfectly programmed computer holding all global facts could perfectly foretell the future. However, early systems, while providing insights into price equilibrium, failed to develop robust trading strategies, often missing crucial price drops before a rally and the optimal timing for trades, leading to significant losses.

A shift in understanding occurred, realizing that effective investment relied on reacting appropriately to information at each point in time, rather than predicting the future. This led to a systematic approach: recording decision criteria for market positions, then evaluating their performance upon trade closure. These criteria were translated into formulas, now called algorithms, and back-tested using historical data, often spanning over 100 years and multiple countries. This rigorous process refined the rules to be 'timeless and universal,' enabling real-time data processing for decision-making mirroring human thought.

The firm's original interest rate, stock, currencies, and precious metals systems were integrated into a comprehensive portfolio management system. This system operated in parallel with human analysis; discrepancies prompted investigation. The computer often educated the human by revealing overlooked factors, while the human taught the computer new criteria. The computer's superior processing power allowed it to 'think' about many things at once, more precisely, rapidly, and unemotionally, ultimately becoming more effective, though human imagination, understanding, and logic remained crucial. Overrides of the system occurred less than 2 percent of the time, typically for extraordinary, unprogrammed events.