Optimize stock portfolios with GPU-accelerated Mean-CVaR and Mean-Variance workflows.
npx skills add https://github.com/NVIDIA-AI-Blueprints/cuFOLIO --skill cufolio
Please help me install this Agent Skill. Skill: cufolio Source: https://github.com/NVIDIA-AI-Blueprints/cuFOLIO/tree/main/skills/cufolio Command: npx skills add https://github.com/NVIDIA-AI-Blueprints/cuFOLIO --skill cufolio
The cufolio skill solves the problem of efficiently building, optimizing, backtesting, rebalancing, or analyzing stock portfolios using GPU-accelerated Mean-CVaR and Mean-Variance optimization.
Use the cufolio skill to build and backtest an optimized portfolio based on the S&P 500 dataset with a risk aversion level of 1.0 and a confidence level of 0.95.
High-intent search queries and answers about installing and using this skill.
Mean-Variance optimization for a stock portfolio is accelerated by solving variance-cap Markowitz allocations as SOCP/QCQP problems using the cuOpt GPU solver. You can compute returns and plot the efficient frontier for your portfolio universe.
Mean-CVaR optimization minimizes Conditional Value at Risk for portfolio allocation. It is handled by generating KDE scenarios for CVaR from returns data and solving the complex optimization problem with the cuOpt GPU solver.
Yes, you need Python, NVIDIA cuOpt, and cuML to run these quantitative portfolio optimization workflows. The environment uses the cufolio package alongside cvxpy, numpy, and pandas for data processing and scenario generation.
Yes, you can backtest trading strategies and rebalance portfolios against benchmarks. The workflow provides insights into strategy fine-tuning and includes rigorous backtesting against benchmarks and rebalancing workflows.
The best way to generate scenarios for CVaR constraints is by computing stock returns and using Kernel Density Estimation to generate KDE scenarios. These scenarios are then fed into the cuOpt GPU solver to solve the optimization problem.
You plot an efficient frontier for a portfolio universe by computing expected returns and solving Mean-Variance or Mean-CVaR optimization problems across varying risk aversion levels. The workflow supports plotting and inspecting the frontier directly.
ladinglogichqTrain and preprocess tabular data with GPU-accelerated cuML models.
ricableDevelop algorithmic trading systems with Rust bindings for Node.js.
ThanhNguyenDatBuild financial models and backtest trading strategies with statistical risk metrics.
sandraschiSolve optimization problems using linear programming, convex optimization, and gradient methods.
