How does Investiva work?

Our trading agent, Investiva, uses the state-of-the-art Deep Reinforcement Learning (DRL) algorithms for autonomous stock trading.

The goal of financial trading is to maximise returns while avoiding risks. Investiva solves this optimisation problem by maximising the expected total reward from future actions over time.

There are multiple advantages of using DRL for autonomous stock trading.

1) Deep reinforcement learning algorithms can outperform human players in many challenging games. For example, in March 2016, DeepMind’s AlphaGo program, a DRL algorithm, beat the Go game world champion Lee Sedol.

2) Return maximisation is the trading goal. By defining the reward function as the portfolio value change, Investiva maximises the portfolio value over time.

3) The stock market provides sequential feedback. Investiva can sequentially increase model performance during the training process.

4) There is no requirement for a skilled human to provide training examples or labelled samples. Investiva uses standard market data (open, close, high, low, and volume) for each minute.

5) Investiva uses multi-dimensional data. By using a continuous action space, Investiva can handle big data with no performance degradation issues.

6) Empowered by neural networks as an efficient function approximator, Investiva can handle extremely large state space and action space.