The goal of this work is to build a comprehensive end-to-end alpha agent that has the capability to iteratively generate alpha strategies, backtest them on historical datasets, perform evaluations and consequently trade the alpha signals live if they exceed expected performance threshold.
- Create a RAG over
101 alpha paper
to generate information likealpha number, alpha expression and alpha explanation
.
from utils import save_output, SavePathType from data_sources import FMPUtils