• machine learning
• long short term memory
• portfolio trading
• cryptocurrencies
• futures contracts

With the advent of automated trading, both the academy and the industry have exploited machine learning models to forecast financial markets.
This work implements a framework for the automated trading of 8 Binance Futures contracts. Long Short Term Memory networks are exploited to predict the future prices. A portfolio strategy, grounded on these forecasts, is implemented with the aim to trade all the contracts simultaneously while hedging risk.
This work also introduces the Hyperelliptical Frontier as a mathematically optimal limit to the admissible trading positions. This redefinition allows to expand the accessible subspace without exceeding the intended risk.