Out of Sample Backtesting

AiVi was trained on 10 years of quantitative S&P 500 stock fundamentals and 10-K statements filed between September 2012 and September 2018. 80% of the data was used for actual learning, the remaining 20% was reserved for fine-tuning of AiVi’s learning models. Backtested data was not part of the learning sample simulating a real world scenario.

Using the out of sample backtesting method, AiVi achieved an AAR of 4% and 6% above the S&P 500 index for the periods 2017/2018 and 2018/2019 respectively.

Future Development

AiVi is a constant learner. Each year, AiVi optimizes its predictions with the help of new historical data and AI model refinements. The next step in AiVi’s development is rebuilding the learning model using optimized and cleaned quantitative and qualitative data. Owing to AiVi’s flexible design, it's AI model is planned to utilize additional data sources in the future such as quarterly data, non-S&P 500 stocks, news articles and technical analyses