The goal of the Path Integral team has been to minimize back calibration. Instead of building a model entirely calibrated to perform well over past sample data, we have developed a system that is constantly adapting and rebuilding itself according to current conditions, with no human intervention. Our dynamically adjusting framework does well over a wide range of past conditions and, because it is dynamic, it is much more likely to perform better in the future than back-calibrated models.

Put another way, the Integral system runs “blind”. Imagine that you are a model developer and are evaluating your system. All models have parameters and it is very tempting to modify them so that particular market conditions are handled a little bit better. That’s cheating. After all, how will you know how to twiddle parameters so the model works right in the unknown future? The Integral system figures things out for itself and does so only by looking at current market conditions and interpolating future conditions. All tests involving past market conditions were conducted in this way, with absolutely blind knowledge of the future and no human intervention.

The Path Integral team is made up of very cautious people and the Integral system reflects this. Early on, we developed models that could predict movements in the stock market. Unfortunately, they also were quite volatile – our own money rides on the Integral system and so its risk tolerance is consequently lower. The latest versions perform well under most past market conditions, regularly making money and with few significant downdrafts. One way this is done is to stay out of the market during periods of uncertainty.

The Path Integral team is composed of individuals with backgrounds in diverse fields: physics, economics, software engineering, and statistics. All these fields have played a role in our work. Over and again, the multi-disciplinary approach has been proven to achieve better results.