eiten

Why we included this project

Individual investors and small quant teams who want to move past a single index fund get a concrete starting point here. Point Eiten at a plain text file of tickers and it builds several candidate portfolios, drawing on eigen decomposition, minimum variance, maximum Sharpe ratio, and genetic-algorithm based weight selection. The part that matters most is the testing loop: you hold back out-of-sample bars, fit portfolios on the rest, forward-test them, and compare results with a benchmark like QQQ or SPY. There is also optional noise filtering on the covariance matrix built on random matrix theory, which helps when your stock universe is small. Expect a command-line Python tool rather than a hosted service, so you run it locally and read its output instead of calling an API.