Illuminating the last dark corners of the investment landscape
Experiments in Financial Data Science and Machine Learning
Market Forecasts
Track machine learning models designed to detect the closest historical matches to the current environment, and then forecast based on those most similar historical cases.
For the Machine Learning experts in the room, these are K-Nearest Neighbors models on historical prices data. Track a simple model that boils the essence of technical analysis into an extremely simple to interpret model.
Track the popularity of over 50 million unique words and phrases used in filings with the SEC for every public company since 1995. Essentially a financial reporting version of Google's Ngram Project.
