The financial service industry has a long history of quickly adapting to new technologies, and deep learning is no exception. Today’s service processes, financial products, and even investment decisions already incorporate advanced degrees of automation, but often in form of rule-based systems. The emergence of deep learning enables recognizing sophisticated patterns in high dimensional spaces and non-trivial activities can be automated with impressive results. The field of application ranges from back-office automation, to credit scoring, to asset management and high-frequency-trading.
AI-powered Hedge Fund
Today’s players in the HFT space rely primarily on traditional models, that are fast to execute but limited by their theoretical assumptions. Machine Learning methods are able to capture complex market patterns with enormous value potential hidden under financial micro-structure noise. Merantix leverages it’s cutting-edge deep learning research and technology stack to exploit these patterns and outperform incumbent players. Merantix has very successfully deployed deep learning based trading algorithms on FX and Crypto future contracts with an astonishingly competitive wire-to-wire execution time. With the active involvement of HFT experts, access to the latest trading infrastructure and a strong industry network, Merantix is currently setting up a “Fund of the future” with 1000+ deep learning enabled trading strategies on a growing number of products.
In partnership with Europe’s leading credit and collection agency Merantix has developed a technology to significantly improve the distinction between good and bad payers. In comparison to current scoring methods, our deep learning algorithm was able to reduce the amount of rejected purchases by 20%-30% while the general risk of default remained unchanged. This is ultimately not only reducing the default rate and bad debt losses but also increasing the number of positive decisions and direct revenue.
Wire-to-wire execution time
We have created deep-learning based HFT strategies with less than 3 µs execution time.
Daily level-3 financial data input
Our HFT trading models are trained and executed on 120GB level-3 financial data each data.
Accuracy of credit-scoring
We were able to reduce the amount of rejected purchases by 30%, while the general risk of default remained the same.
Data points considered each day
Our model evaluates more than 2 M times a day to find the optimal trading moment.
Your data and knowledge,
Schedule an appointment
Skalitzer Str. 104