Robobroker.com is an intelligent system for trading on the stock markets of the world

The project is aimed at increasing the profitability of legal entities and individuals from trading activities on the Russian and foreign stock markets.
The innovativeness of the approach lies in the fact that the system will allow:
1 To diversify the investor's portfolio as much as possible on optimal terms by forming a portfolio of at least 100 shares (maximum 1000 units), observing the limit per 1 share in the amount of 0.2-1% of the portfolio value, thereby reducing the risks of adverse developments in the market.
2 Form at least 1000 strategies based on historical and current data and substitute the optimal strategy for making transactions at the current moment every minute in the trading session. A deep analysis of the selected strategies will allow the system to learn and form the most effective combinations of strategies in a given period of time, depending on the preferences of users:
o according to the degree of risk:
 Low risk;
 Medium risky;
 Highly risky.
o by expiration date:
 Short-term;
 Medium-term;
 Long-term.
3 Increase the turnover of available funds up to 5 times compared to other systems due to the speed of making new transactions after closing previous transactions. After the transaction, the system is as fast as possible up to 0.5 seconds. selects the most profitable new stock for the current period of time and makes a new deal in the long or short position.
4 Analyze and carry out semantic recognition of at least 1000 news sources (Search, collection, monitoring and parsing of information sources: news (business media, authorities: Central Bank, statistical agencies), social (Twitter, Facebook, etc.) and industry (specialized forums, databases, etc.) with the formation of the largest possible knowledge bases. To solve this problem, we use our own developed algorithms for deep analysis, learning and semantic recognition of news backgrounds, which is carried out using NLP technology (neural networks for natural language processing) to determine potential motives or possible trends in changes in the values of securities).
5 To determine the reliability of news based on the use of our own developed algorithms.
6 Determine the time of purchase or sale of securities using our own developed predictive analytics algorithms based on the results of the analysis of historical data and data obtained in the last second of trading from semantic recognition algorithms and determining the degree of reliability of news.
7 To trade around the clock in 3 exchange zones (Asian, European, American), optimizing cash flows between them.
The scientific novelty of the project consists in:
- development of the RBB1 data array version control technology, which will increase the processing speed of tasks by 10%
- development of news recognition technology, which will reduce the processing time of news by 30% and increase the degree of recognition reliability to 80-85% (defined as recognition of a positive or negative fact for a company and subsequent confirmation of the stock market movement that occurred).
- development of technology for recognizing the reliability of news, which will allow to increase the level of the obtained reliability of estimates to 80-85% to determine the time points of purchase or sale of securities
- development of predictive analytics technologies for determining the trend of stock movement, which will reduce the processing time by 20% to 3 seconds, improve the quality of the forecast to 80-85% by introducing a mechanism for retraining the neural network, taking into account the appearance of new data and enriching the results of news recognition;
- in developing an algorithm for searching, collecting, monitoring and parsing information sources: news (business media, government agencies, Central Bank, statistical agencies), social (Twitter, Facebook, etc.) and industry (specialized forums, databases, etc.) with the formation of the largest possible knowledge base using Bigdata approaches.

No comments yet