In this project i implemented naive genetic algorithm that learns from the available Google Trends data i scraped from the internet. I used also Yahoo to fetch stock prices.
The model tracks some features like the brand popularity, stock popularity and then applies this knowledge for predicting what stock to buy by what amount.
Each agent has bias towards trends, companies, sectors as well as many more biases (aprox. 60 weights)
The report.txt file contains results from the bot from 06-2010 to 06-2023
The project was made for educational purposes as a project for Machine Learning Methods class in Gdańsk University of Technology.
The result of the training was this agent:
AgentReport(profit=(396.7624472728585,),
number_of_transactions=9,
average_transaction_amount=2259.6402719192065,
multiplier=0.30722326771436814,
courage=0.5518224612738823,
invested_companies=3,
sector_biases={'AUTOMOTIVE': 0.4945335353859816,
'BANKING': 0.5160124180318092,
'ENERGY': 0.7505727618903173,
'FOOD': 0.5001275936823805,
'HEALTHCARE': 0.6685896409295591,
'MEDIA': 0.5532138072973746,
'RETAIL': 0.4889761815866155,
'SOFTWARE': 0.5232449605079104,
'TECHNOLOGY': 0.5500394773036379,
'TELECOMUNICATION': 0.5305580345668408},
bought_stocks={'AAPL': 2611, 'MSFT': 3494, 'PFE': 3835})