Legend Advisory Corporation (LAC) created its proprietary neural network–the Asset Allocation Neural Network (AANN)–to make predictions regarding the relative strength of individual asset classes.
Neural networks and financial modeling
Artificial neural networks are adept at detecting patterns that may appear random to the human brain or a linear statistical model. In addition, unlike human beings, neural networks operate without emotion, bias, distractions or fatigue. As a result, they are not influenced by the twin emotions of fear and greed that may compel many investors to act imprudently.
Artificial neural networks are rapidly replacing traditional, rules-based computer modeling systems that seek to establish a series of rules and then compare data inputs to those rules to draw conclusions. Rules-based systems are extremely limited; in particular, they are inflexible and cannot adapt to changing input and interrelationships without defining new rules. Legend believes such rudimentary techniques are often inadequate for financial modeling and forecasting, which typically involves large numbers of variables that can interact with one another in an endless variety of ways. We believe neural networks excel at making predictions in situations like these, where the range of possible outcomes is infinite.
A machine with the ability to "think" and "learn"
The process by which AANN assesses the relative strength of individual asset classes is complex and constantly evolving. AANN reviews thousands of historical data inputs to identify and interpret patterns and causal relationships as they relate to the major domestic and foreign debt and equity sectors. This input consists of market and economic data, including stock market indices, global interest rates, earnings and a myriad of other relevant data.
By its very nature, AANN is an adaptive system with the innate ability to adjust to an evolving global market place. Unlike traditional modeling which is static, AANN continuously tests its hypotheses regarding the relationships between various economic and market variables and hones its predictive abilities. Assumptions that have proven invalid over time are discarded while expectations that continue to be corroborated gain greater influence.
Over time, AANN deftly identifies the key variables and examines the impact that changing relationships among these variables may have on world financial markets in order to provide predictions of asset class performance.
Learn more:
|