Loading Events

Open lecture on Meet on cryptocurrency price forecasting using AI/ML/DS

On Friday, January 21, 2022, I gave a lecture on the topic of cryptocurrency exchange rate forecasting on Google Meet and invited those who wanted to join the Kaggle competition. As an option, I invited them to my team. The invitation to join was in the Telegram group “Department of SAIT VNTU”. There were mostly VNTU teachers, but there was also a guest from Uzhhorod who followed my announcements on this topic, as he is actively involved in it. I also briefly covered:
basic concepts related to cryptocurrencies (principles of operation, blockchain, mining, cryptocurrency exchanges)
advantages and disadvantages of cryptocurrencies that are absent in traditional currencies;
cryptocurrency indicators on exchanges, important for predicting their rate;
briefly described each of the 14 cryptocurrencies that must be predicted in the competition,
typical problems that complicate forecasting (non-stationarity, heteroscedasticity of series, the “Elon Musk tweets” problem, etc.);
described the conditions of the competition and how to submit a decision;
described an example of a solution I had already submitted, which I posted as a public notebook (the rules prohibit sharing information about private notebooks outside the team, so I had to make a program and share it with the world - it's interesting that at the moment there have already been about 80 forks of this notebook and more than 1000 views, which Kaggle just congratulated me on - I'm attaching their email message).
I gave the main part of this presentation earlier during a lecture to master's students of specialty 126 of VNTU, where I teach the discipline “Information Technologies of Monitoring and Data Analysis.” During this lecture, I took into account the wishes of the students regarding which aspects should be discussed in more detail.
Later, two teachers from our university expressed a desire to form a team and we started working together (when the competition ends, we will write a joint post). There is still room for 2 (teams of up to 5 people are allowed). Or you can form your own team. The deadline for forming teams and joining the competition is tomorrow! The deadline for submitting a solution is 02/01/2022. Then the laptops will live “their own life” – for 3 months, they will analyze how they predicted the future – the 15-minute profitability of each of the 14 cryptocurrencies.
Given the strong noise and stochasticity of the processes, all participants have a real chance of taking 1st place or one of the first 10 prize winners, where the prizes are from 5 to 50 thousand USD. But it is interesting and simple to strengthen your skills on a real task. And then it can be students' diploma theses, scientific publications, etc.
I plan to repeat my lecture again in the spring for those who want to, but I will also add a review of the winning laptops. In general, this will be a lecture (rather, material for independent work) for master's students of specialty 126 and undergraduate and graduate students of specialties 124 and 126, on the topic "Time Series Forecasting."

  • Start Time
    January 21 @ 14:30
  • End Time
    January 21 @ 16:00
  • Organizer
    SITE VNTU