[Task] Quantitative trading

[e10nMa2k]

Task Description

You are given a Qlib-based repository. Your task is to automatically optimize an existing algorithm (model, training pipeline, or strategy) to achieve better performance than the current baseline.

Objectives

• Improve key metrics (e.g., IC, Sharpe Ratio, Annualized Return, Max Drawdown).
• Maintain compatibility with Qlib’s full pipeline (data → training → backtest → evaluation).
• Avoid data leakage and overfitting.
• Ensure fair comparison using the same dataset and evaluation settings.

Optimization Scope

You may improve:

• Model architecture
• Hyperparameters
• Training strategy
• Feature engineering
• Portfolio construction or risk modeling
• RL reward design (if applicable)

Expected Output

• Summary of changes and rationale
• Modified code/config (minimal diffs preferred)
• Baseline vs. optimized results
• Clear performance comparison

Goal: Achieve robust and reproducible improvement over the baseline within Qlib’s framework.

Baseline Repository Link (Must be Public)

https://github.com/microsoft/qlib/tree/main/examples/benchmarks

Baseline reproduction (minimal)

cd examples # Avoid running program under the directory contains qlib
qrun benchmarks/LightGBM/workflow_config_lightgbm_Alpha158.yaml

Dataset (Must be Public)

https://github.com/chenditc/investment_data/releases

Results and Evaluation Metrics

Rank IC (Information Coefficient)
or
Sharpe Ratio

Preconditions (required)

• The baseline I provide comes from public code hosted on an open-source platform such as GitHub.
• The task I use/optimize is based on a public dataset.
• If my request involves private code/data, I will contact interndiscovery@pjlab.org.cn by email instead.