Predict movement direction of a share
High level componenets:
- Data Fetcher: Data can be fetched via API and dumped in tabular structure. Output will be in stock_market/stock_name/daterange_currentdate
- Data Transformer/ Featurizer : It will do the filtering , interpolation, other data enrichment, calculation of derived fields, reading multiple data and joins and date ranges. Test Train split
- ML model, BackTesting, and other offline testing : Run the suitable ML algorithm
Low level details :
- Data Fetcher Support fetch data from API and write to a systematic folder structure ( at path stock_market/stock_name/daterange_currentdate , data should be in tabular format CSV) Download data from internet sources and store at correct location
- Data Transformer/ Featurizer Read multiple data, stock data and other indexes data Join multiple data on date column Fill in the missing values Calculate derived features Select only required columns Split the data into train and test
- ML model Read the training and test files EDA of features, feature importance , feature correlation Calculate precision /accuracy on test data Tune the hyperparameters to achieve maximum accuracy Measure more offline parameters like profit booked, risks associated