We would like to demonstrate FinBrain Technologies’ new Deep Learning Enabled Algorithmic Trading Bot.
In the video, we will show you how our Trading Bot performs over the time, by backtesting it between the time period Nov 1st , 2017 and April, 3rd 2018 for COTY and UAL stocks listed under NYSE.
You can see the stock price movement on the chart at the top-right, together with the 5-day ahead predictions for each period.
And the trade signals can be seen on the chart at the bottom-right, together with the real and algorithm-traded cumulative stock returns.
As you can see on the video, our algorithm has achieved 73.4 percent cumulative return, for the interval that COTY stock has increased 19.2 percent. You can see the generated signals on the left, together with the particular returns for every single trade.
In a 100 day period where UAL stock rose 15.47 percent, our Deep Learning Algorithmic Trading Bot has beaten the market significantly, and generated 60.6 percent return by executing 20 trades.
The Algorithm and Trading Signal Generation
Our algorithm analyzes the time series behavior, “learns” its features and generates future predictions. Then, the trade signals are determined according to the predicted results, and the trades are executed.
The Neural Network first optimizes its hyper parameters, then validates its training and tests itself against the unseen data. The key here is to avoid overfitting the network to the training dataset and to improve the neural network’s generalization capabilities.
Our algorithm perfectly combines various Mathematical approaches together to choose the topology which will yield the best performance on future/unseen data. That’s why FinBrain’s algorithm enables robustness and reliability for predicting the future movements of the time series data(stocks, commodities, indices, Foreign Currencies, ETFs, Cryptocurrencies).
FinBrain has successfully implemented cutting edge Deep Neural Network technologies to the algorithmic trading area, and already started to generate alpha.
You can contact us at [email protected]
FinBrain Technologies, 2018