The chart below captures CONCERT AI historical performance relative to the S&P 500 Index over an 11-year period. Using the original selection process designed in 2004, the model produced an annualized return of 10.5% vs. S&P 500 6.3% over the past 11 years. All trades were machine generated and recorded in real time without human intervention since 1/1/07. The system was free to pick from 9000 ticker symbols in its database. Algorithms for this layer of model were frozen in 2004. This early iteration of the model is the foundation upon which subsequent layers of the model were built and has not been modified since 2004.
The next layer of model development was the addition of secondary neural networks, Monte Carlo iterations and investment rules. Inputs for this layer of the model were frozen in 2007 but the neural networks continued to learn from daily market data.
To test the robustness of the system, we restricted the inputs to a more limited set of securities, looking first at the components of widely followed indices, and then individual stocks. The model ‘knowing’ what it ‘knows’ today made historical selections looking back over time. These results are considered pro-forma.
We reviewed the S&P 500 using the 9 Sector SPDRS. The results can be seen in the following chart.
The CONCERT AI sector selection process produced a pro-forma annualized return of 14.2% vs. 6.3% for the S&P 500.