Clustering of Market Scenarios
Brain Clustering Engine performs a clustering of the market history in order to provide to two types of information:
- A descriptive model: “days like today” - to which days in the past the current day is most similar. This can be done using features divided by topic (e.g. financial stress, jobs, sentiment etc.) or by using all features altogether. Dimensionality reduction can be applied in both cases.
- An input to an asset selection strategy (sector ETFs, stocks, asset classes): how the financial asset has performed in average in the current cluster. A customized strategy can be built to be tested and validated.
Using Unsupervised Machine Learning techniques the objective is to identify which days in market history are most similar to current day according to variables corresponding to a certain topic, e.g. financial stress, sentiment surveys, jobs.