Novosibirsk State Technical University (NSTU) is creating a cryptanalysis platform with machine learning and NLP support for comprehensive market analysis. The tool will allow users, especially novice traders, to competently assess the current situation in the cryptocurrency market and make informed investment decisions.
According to the developers, Mark Starchenko and Ilya Zavgorodny, students of the Faculty of Business, manual analysis of cryptocurrency assets takes a lot of time, the market is mobile, which means prices can change very quickly. There are a large number of cryptocurrency analysis tools, but most of them are professional and may be incomprehensible to novice traders. In addition, the market is constantly evolving, and new trends are emerging that need to be taken into account. The platform will help make the investment decision-making process as simple and expeditious as possible.
"The Telegram bot can be considered as a personal crypto assistant for users. When it becomes difficult to keep track of multiple sources of information on the cryptocurrency market, our solution will be a good option. Instead of monitoring dozens of exchanges, websites, and applications, customers will have access to analytics and risk metrics directly in their messenger. This will save time, help you make informed investment decisions and avoid missing out on profitable deals," said Mark Starchenko.
The project provides a list of key features (specific functions implemented in the product) necessary for complex cryptanalysis: analysis of the user's current portfolio using risk modeling and making recommendations for its optimization; analysis that takes into account on-chain metrics and behavioral data, as well as metrics reflecting public opinion, such as the frequency of search queries for specific assets., and a wide range of technical indicators; semantic analysis of the news background, which makes it possible to assess the impact of information events on the cryptocurrency market. The totality of the collected data and predicted price ranges is fed into the LLM (large language model), and the model interprets the results in an accessible form.
The current version of the project implements basic functionality: price tracking, notifications, risk analysis, forecasting, semantic news processing. In the near future, the developers plan to finalize the semantic analysis module and analyze the user's investment portfolio. "We want to offer the user a more informed and effective portfolio management strategy," adds Mark Starchenko.
The project is currently in beta testing, and there are already the first users in the Telegram bot. It is planned to officially launch the platform in the 1st quarter of 2026 and further switch from the TG bot format to a Telegram mini-application.