Treffer: Implementation of an artificial intelligence-based algorithmic stock trading system ; Tekoälyyn perustuvan algoritmisen arvopaperien salkunhoitojärjestelmän käyttöönotto

The objective of this Master’s study was to construct a fully automated trading system using Artificial Intelligence and Machine Learning methods. This was accomplished through an extensive literature review supplemented with individual analysis and deduction. Based on the literature examined, it was determined that a fully automated trading system for stocks can be built by integrating algorithmic trading with machine learning techniques. This conclusion was reached by identifying the main steps of a portfolio management process and applying algorithms to automate predictions and stock selections. For the trading system, a framework of 10 steps was created in order to help individuals or enterprises create their own automated systems. The final process consisted of data sourcing and analysis, machine learning implementation, prediction development, asset selection, portfolio optimization, order execution and performance evaluation. For all the aforementioned steps, methods and technologies were provided in order to automate their execution. Finally, a detailed implementation of the trading system was provided which involved sourcing data from Yahoo Finance, conducting data analysis, applying feature engineering, selecting and training ML models using cross-validation, and deploying the model to generate real-time trading signals. Portfolio optimization was conducted through available Python libraries and executed via an application programming interface to the market. ; Tämän maisterin tutkimuksen tavoitteena oli rakentaa täysin automatisoitu kaupankäyntijärjestelmä käyttämällä tekoälyn ja koneoppimisen menetelmiä. Tämä saavutettiin kattavalla kirjallisuuskatsauksella, jota täydennettiin yksilöllisellä analyysillä ja päätelmillä. Tutkitun kirjallisuuden perusteella todettiin, että täysin automatisoitu kaupankäyntijärjestelmä osakkeille voidaan rakentaa integroimalla algoritminen kaupankäynti koneoppimistekniikoihin. Tähän johtopäätökseen päädyttiin tunnistamalla salkunhoitoprosessin päävaiheet ja soveltamalla ...