Wall St Snacks Analytics
We are a core group of PhDs who believe that all data have a story it is up to us to find this information, store it, process it, and learn. We combined have spent > 10 years doing so and here are just some of our findings.
Every day our servers take in 100s of gigabytes of new information and layer that upon terabytes of market and option data dating back >10 years. Building on this we generate a unified single machine learning (ML) model that can predict future positive returns. This is signified by certain stocks having a higher weight.
The signals we give our unified ML model are ones we’ve identified to be statistically significant on their own in predicting future returns. By combining all signals into a unified model we can leverage the power of all signals to yield a single less volatile prediction (confirmed w/ backtests).
In all, our machine learning model takes these novel predictive signals and yields a single predictive model Below is how our model performs on a backtest when rebalancing every 8 days (1st column), 1 day (2nd column), SP500, Nasdaq each row is the yearly return. Overall on backtest, we outperform the sp500 index at a statistically significant rate year over year. In fact, annually we return 5x the SP500. At an annual average return of 55% compared to the sp500 11%.