(Last updated on 24 Mar 2021)
Dark Pool Gamma Exposure Model
DGM is a machine learning algorithm (statistical model) based on alternative data – dark pool data and options dealer gamma exposure (DIX and GEX) – with the goal to predict the direction of the next day move in the S&P 500 index (SPX).
The model uses data made available courtesy of squeezemetrics and is updated daily. It is re-trained with every update using an extensive optimization and selection process (part design, part randomization, part brute-force), then executed to attempt a prediction and tested against past data.
The final result produces a score that indicates a bullish or bearish leaning stance for the subsequent trading day, along with simulated return statistics (backtest) for a few historical time periods.
More recently (spring 2021) we introduced a second, enhanced version of the model – DGMV ("DGM with VIX Futures", due to lack of imagination). In addition to DIX and GEX, this model consumes VIX futures data (courtesy of vixcentral). We believe VIX futures can be a powerful volatility and sentiment indicator, and a great complement to DIX and GEX. DGMV adds information, complexity, and predictive power; as such we consider it a premium feature and is available to paying subscribers only.
The idea for the models was born out of conversations in online trading rooms. While work remains to further enhance the models, we feel they have reached a level of maturity and performance to be of interest to a wider audience.
The model stats tell the story, sign up and take a look for yourself. We are offering the site for curious minds at this still early (beta) stage. The current model versions are dgm-0.8 and dgmv-0.8 (as of March 2021).
Please understand this website does not offer any investment advice and the usage is entirely at your own risk. By using this site you agree to our Terms of Service.
For any questions, get in touch!