SELF-LEARNING, ADAPTIVE TECHNOLOGIES

TITIAN systems are the result of 5 years of scientific R&D in advanced machine learning methods and of the lessons learned from their successful deployment in real-world applications (e.g. weather forecasting, network traffic load management, object detection, voice & text recognition, critical illness diagnosis, surveillance / security applications, etc). The software that we have developed has been specifically designed to address the significant challenges involved in forecasting, with an acceptable level of reliability and consistency, complex non stationary phenomena such as financial time series.

Each trading system has built-in mechanisms that every day:

• Incorporate the most recent data from the evolving market
• Map the relationships between these data and the output (future price or market direction)
• Determine the predictive algorithm that best characterizes these relationships at that specific point in time
• Use this (new) algorithm to predict the price or market direction for the following 24 hours
  

Patterns, similarities and relationships between observable variables (continuously updated to incorporate the most recent data) and future price are re-assessed by the systems daily.