- To build time series forecaster onto Streamlit next
- Consider use of other databases (DuckDB, PostgresSQL) for learning
This script aims to perform SARIMA to analyse stock prices and forecast into the future with the following purposes:
- Identify near-term price movement trends
- Determine optimal actions for maximum gain
The web scrapper utilises BeautifulSoup 4 to scrap info from the yahoo finance statistics page.
Two sets of data are utilized: S&P constituents and a personal portfolio.
This script performs a K-means clustering on the S&P constituents to determine different groups of companies. The following features are used in K means clustering:
- Market cap
- Revenue
- Profit Margin
- 52 Week Change
- Quarterly Earnings Growth (yoy)
The identified clusters are compared with one's portfolio for comparison.
20240110:
- Added RMSE or AIC into TS validation
- Optimized for SARIMA parameters, using auto_arima and further analysis
20240108:
- Modularising Forecaster on notebook
- Added functions to
- forecast
- validate forecast
- plot forecast
- determine single buy and sell for maximum profit within forecast
20240107:
- Completed modularisation of YfScrapper
- Completed learning on time series (ARIMA, SARIMA)
20240103:
- Refactored code and directory to modularise scrapper
- Renamed project