I construct a series of time-series features from the literature and apply a novel XGBoost model to predict the next days price of a number of assets. The concept is simple and can be expanded to many variables, incorporate many assets and be applied to different Machine Learning models.
A simple backtested Bollinger-Band strategy and Directional Movement Index (ADX) strategy.
I break down some popular and fundamental models from quantitative finance & engineering, construct some factor analysis on a series of Assets and EFTs along with a randomly generated portfolio constructed from scraping tickers from the SPY500 wikipedia page.
I analyse the difference between two time series and obtain a 67% accuracy (on anonymous data)
I show my solutions to a few Hackerrank competitions/problems I had completed and discuss the models and solutions.
I use the na_kalman function from the imputeTS package to impute randomly generate missing stock prices.
I compute the money left on the table for companies listed on the yahoo finance IPO calendar.