Deep Reinforcement Learning combines deep Neural Network and RL algorithms, turning every sequential task into a Markov Decision Process: an agent interacts with environment via action, getting rewards, and improve upon its future actions to reach better environment.
Trading financial markets is such a task to optimize.
Data in financial markets are endless, yet price is the single information that reflects everything that has happened so far. The study of price movement is an ancient art that few has mastered, yet our Deep Learning A.I. is the perfect student to surpass our human best, learning from 5 million live trades executions in 5 minutes, more than any human can trade in his or her entire life.
Our focus is on the world's biggest financial markets, the Foreign Exchange Markets. With more than $5 trillion dollar transacted every day, it is largely dominated by technical analysis in the 1 minutes to 4 hours term. Forex Major USD Pairs are the perfect playgrounds for our agile small-scaled mid-frequency A.I. traders, with low cost and latency combined with ample liquidity.
When Quantitative Trading Meets
Deep Reinforcement Learning
Google's AlphaGo has solved the ancient game of Go. Now with the same algorithm, we are here to solve the game of Quantitative Trading.
When A.I. Makes the Trades for Us
Watch our A.I. trader trading 2018 EURUSD Spot market before and after 1 millions steps of training.
The first video shows the agent's trading before any training is done. It's randomly placing trades.
The second video shows its trading after it's trained on 2017 EURUSD data.
It learns when to buy and sell, and how to manage open positions.
Cumulative return until late June is 277%, and we achieved similar or better results among all G7 pairs.
The Solid Line ─── : Buy Action
The Dashed Line ------- : Sell Action
Background Chart : Market Price Chart
As Seen On
Awards and Recognitions
We are finalist of MassChallenge Boston's 2018 Cohort, selected from over 1600 startups across industries and regions around the world, as a top Fintech venture with potential to make great impact in finance. MassChallenge is the world's biggest and most startup friendly accelerator, a global network of zero equity startup committed to strengthening the global innovation ecosystem by supporting high potential early stage startups across all industries.
We are part of NVIDIA Inception Program, a virtual accelerator that nurtures dedicated and exceptional startups who are revolutionizing industries with advances in AI and data science. We have access to NVIDIA's global ecosystem—a massive network of deep learning experts and thought leaders, try out the latest software, get early access to state-of-the-art technology and training at The NVIDIA Deep Learning Institute (DLI) with the latest techniques in designing, training, and deploying neural network-powered machine learning.
We are part of National Science Foundation's Innovation Corps Program (NSF I-Corps) Brandeis University local cohort, which seeks to develop and nurture a national innovation ecosystem that builds upon research to guide the output of scientific discoveries closer to the development of technologies, products and processes that benefit society.
We are the Grant Prize Winner for Brandeis University's SPARKTank 2018 Challenge. Our A.I. Capital Management stood out among 14 amazing startups, winning prize of $10,000 in funding. The challenge also grants us fast track to this year's MassChallenge semi-final. You may see us on the biggest venture stage in Massachusetts later this year.