Bootcamp_repo_portfolio

Bootcamp project portfolio optimization

See jupyter notebook attached for full project

https://github.com/Lucas-BLP/Portfolio-optimization-a-K-mean-hierarchical-clustering-approach/blob/master/Portfolio%20optimization%2C%20a%20K-mean%20%26%20hierarchical%20clustering%20approach.ipynb

Project design and created in 3 days during data science course. Objective : Portfolio optimization using clustering solutions (Hierarchical Clustering, K-Mean) and a variance optimization process (Markovitz 1952). Limitations: Variance optimization process are hit by exponentially increasing bias.

Step 1: Extract S&P500 market data (Open High Low Close, OHLC) from API (survival bias here)

Step 2: Hiearchical clustering of companies (5 clusters retained)

Step 3: Creation of equally weighted portfolio among clusters (1/N, DeMiguel, Garlappi)

Step 4: Markovitz optimization process among those 5 clusters to reduce estimatino bias (Maximum sharpe ratio retained)

Step 5: Results analysis, benchmark comparison

Sources: https://medium.com/@saadahmed387/machine-learning-for-stock-clustering-using-k-means-algorithm-126bc1ace4e1 https://pythonforfinance.net/2018/02/08/stock-clusters-using-k-means-algorithm-in-python/