Historical financial market data is the time series data from the past. It is one of the most important and the most valuable components for speculating about future prices. With more data, hence more information is available, it is possible to make better conclusions about what will happen in the next time period. ANN/Deep Learning algorithms can "learn" the complex relations between the financial time series by analyzing large amounts of market data. At the AI in Finance Summit in New York this September 06-07, Ahmet Salim Bilgin, founder of FinBrain Technologies will share his work in deep learning for modelling the future price movements of assets. In advance of the summit, we caught up with Ahmet to hear about his work at FinBrain and his career in AI and Finance.

How did you begin your work in AI and Finance, what came first?As an Electrical-Electronics Engineer, I have a strong background in Signal Processing, Optimization and Controls. These fields of Electronics require a strong understanding of Matrix Theory, Time-Frequency Domain Transformations, Probability and Stochastic Processes as well as the Signals and Systems Analysis Methods. Participating in Life Rescue Radar Development has provided me the opportunity to apply different mathematical approaches into a real-life problem. Constant learning and development is a necessity when it comes to Artificial Intelligence Technologies, and this challenge is the driving factor for my passion about AI. Combining my interest in the Stock Market, passion for emerging AI Technologies and Mathematical background has enabled me to find the niche. I have started working on Recurrent Neural Networks for developing Time Series Prediction models after noticing the sector's need for new models to analyze and predict the price behavior of the financial assets.
How are you using AI to create models to predict the future movements of assets?Deep Learning algorithms yield great performance on analyzing multi dimensional highly non-linear datasets. Deep Neural Networks “learn” from the data by adjusting every single neuron's weight and bias values using backpropagation method to minimize errors. FinBrain’s algorithm is capable of “learning” from all kinds of Financial data including Stock Open, High, Low, Close values, Technical Indicators and News/Market Sentiment. Our algorithms analyze large datasets for every single asset, assign more weight to the most important features that determine the asset prices, extract the time series correlations for the historical data and generate future projections from what the Neural Networks have “learned”. Our Deep Neural Network Model optimizes its hyper parameters for the best performance on the unseen data, and utilizes a number of Mathematical approaches to prevent overfitting.
What have recent advances in AI helped you progress with your work?Adjusting the Neural Network parameters is computationally expensive. The computation time might grow exponentially for larger datasets and for more complex models, depending on the type of the optimizer used. Parallel computing technologies such as cloud computing and GPU computing have shortened the time spent on the “learning” process, while sacrificing no accuracy. Some of the widely known Machine Learning Frameworks have helped us with decreasing the programming complexity and saved us resources.
Privacy and security is always a concern when handling sensitive data, how do FinBrain ensure data is kept safe?FinBrain collects Financial Data from various sources, combines and organizes them into large datasets for every single financial asset of our interest. Our datasets are encrypted and backed up periodically. Firewall and anti-malware protection measures have been taken to keep the data and algorithms safe. We constantly monitor the performance of our hardware equipment and servers to minimize downtime and ensure system security.
What's the next big goal for FinBrain?FinTech industry is FinBrain’s main focus, we keep developing products for different markets to reach out more amateur/professional traders as well as the large financial institutions. FinBrain aims to create a platform that will transform the way people and institutions fundamentally trade. A trader spends years and thousands of dollars to build the necessary skills to generate profits from trading. Becoming an experienced trader costs time, money and huge effort and even so, only 5-10% of all traders are able to make money from trading. Which means the remaining 90-95% lose money, time and hope.
The traditional Fundamental Analysis and Technical Analysis techniques are the Financial Prediction methods of the 20th Century. These old-school techniques fall short to capture the non-linear characteristics of the large amounts of Financial Data, whereas Artificial Neural Networks are non-linear by their nature and are capable of capturing these relations with a huge success rate. Even the ordinary people, who have no trading background, can easily use FinBrain to decide on which asset to invest in and start generating alpha.
FinBrain’s services will help the ordinary people who want to invest in financial assets who have no background and courage, as well as the traders and investors who spend most of their time analyzing charts, reading news, and putting so much effort. Also, we would like large financial institutions who manage and risk billions of dollars, to utilize our special solutions for maximizing their returns. FinBrain’s goal is to be a tool for everyone, who would love to adopt the newest technologies to generate higher profits.
What are you most looking forward to about the AI in Finance Summit?

As FinBrain, we would like to share the technological background of our services and the challenging parts of our work with sector professionals. AI will disrupt many industries and we believe that, Finance will adopt these new technologies faster and better than the other sectors. RE•WORK brings the professionals from large-scale institutions, startups, engineers and scientists together in AI in Finance Summit, where all participants will share their experiences and ideas. We see this as a great opportunity to connect and collaborate with the professionals in FinTech. We also look forward to raise an awareness for our technologies and capabilities in applying Deep Learning models to Financial problems.