Project type: Non-academic research
Current status / development stage: Proof of concept / Working prototype
Technologies: AI, cloud
FinancialMarkets.cloud is a system that aims to create fundamental market context of the US stock market based on the insights from various data sources. Most of the data processed by the system are free and publicly available. FinancialMarkets.cloud uses AI to support its analytics. The entire system workload takes place in a secure cloud ecosystem.
Project type: Non-academic research
Current status / development stage: Working prototype
Technologies: convolutional neural network, cloud
In the following Nasdaq scan example, Kolejarz checked 3727 tickers/stocks. Entire scan took approximately 4 minutes. As a result, potential high volatility was detected on the 47 stocks.
while scanning Kolejarz browses every stock's chart
to find potential Railroad Track (RRT) pattern that indicates potential unusual volatility
Kolejarz is an AI powered cloud based scanner that detects high/unusual volatility on financial instruments. It mainly focuses on the US stock market (NYSE + NASDAQ). Although it was successfully tested with Asian and some European stock markets (Xetra, GPW,WSE) as well.
Kolejarz utilizes a deep learning model / convolutional neural network (CNN) that was trained with approximately 30k samples of the Railroad Track (RRT) patterns (DiNapoli, 1998)
manually identified on the charts from the following markets:
The training process took place in April 2020. Since June 2020 Kolejarz has been residing in the cloud. Currently, work on the 2nd training attempt has been postponed.
Kolejarz delivers market intelligence as well. It understands the Global Industry Classification Standard (GICS®), hence is able to detect unusual volatility on a series of stocks that belong to a particular sector or industry. For every stock Kolejarz provides multiple charts in different time intervals (W,M,Q), fundamental data, detailed stock’s description and recent news (with the sentiment) associated with that particular stock/ticker.
Project type: Non-academic research
Current project status: Reserach, working prototype
Technologies: data analytics, ML, DL
SuperCOT is sub-system of the FinancialMarkets.cloud that aims to hunt anomalies (with a little help of artificial intelligence) in the traders behaviour that are encoded in the publicly available Commitments of Traders (COT) Reports published by the Commodity Futures Trading Commission (CFTC). The outcome from the SuperCOT is then subsequently used to determine the possible impact of the detected anomalies on the future price behaviour.
Project type: Non-academic research
Current project status: Reserach, working prototype
Technologies: AI, cloud
Radio Sowa as a stand-alone sub-system of the FinancialMarkets.cloud processes US market and stock news to provide additional insights which might be useful while building the market or industry context.
Radio Sowa creates an outlook on how a particular industry is seen in the electronic media. Weekly observations of the news sentiment might help in spotting turnarounds on the stocks within the industry group or mark industries with highly speculative traders behaviour.
In addition, Radio Sowa delivers to the public free daily (4am UTC time) US Market News headlines service with most relevant news from the last 24h.
Project type: Security as a service
Current project status: Working prototype (under development)
Technologies: cloud, SIEM
Live projects in the cloud ecosystem need constant access to data sources and other 3rd party services. It is much easier with the cloud to process the workload from a number of projects and manage authentication and authorization for users. Residing in cloud ecosystems are inevitable for now and the future.
Nevertheless, cyber-security for cloud environments is a challenge. Cloud security must be transparent and not invasive. Although strong enough to identify explicit threat actor activity against most critical intangible assets.
That is why Cyber Zuzanna comes into play. Monitoring normal and suspicious network activity, investigating all unauthorized access to the cloud and threat intelligence, those are some of the tasks that Cyber Zuzanna does. Many of the processes in this project are automated, some of the solutions and tools are well known industry standards.
In a nutshell, Cyber Zuzanna helps to understand from the cyber security perspective what is happening now and what has happened in the (recent) past.
Investiagting unautorized access attempt to the project data, that took place on 6th Feb 2022. Source Poland, according to the threat intelligence offender's IP address was not invlolved at that time to any malicous activity
6th Feb 2022, source Poland, unautorized access attempt to one of servers in the cloud. According to the threat intelligence data offender's IP was classified as an exploited host.