Data

The data was collected from Wharton Research Data Services(WRDS).
contains information for publicly traded companies in the United States.
- 1. Quarterly financial statements
- 1. Daily prices.

Task

As a method of defining asset classes for asset allocation, we use k-means clustering.
After data preprocessing (e.g., removing outliers, scaling features), clustering was conducted and clusters are evaluated on Silhoueete analysis.
Finally, various asset classes defined in different ways were compared.
- 1. k-means clustering with financial statements
- 1. k-means clustering with style factors
- 1. Style classification, a traditional method
The three methods above were compared based on the following criteria.
- 1. Do they have distinct characteristics?
  - Distribution of financial features (e.g., scatter plot, boxplot)
- 1. Do they move differently in the market?
  - Return correlations (intra-correlations and inter-correlations)
  - Return and risk characteristics

A key determinant of portfolio returns is asset allocation, which requires proper asset classification.
AI models allow you to classify assets in a different way than traditional methods.
K-means clustering is one of the simplest and popular unsupervised machine learning algorithms.
As a result of classifying stocks via clustering with financial properties, stocks are grouped differently from the existing method.