Optuna - Hyperparameter Optimization
Optuna is a Python library for automated parameter optimization using adaptive search instead of manual tuning.
It shines whenever:
- You have multiple knobs to tune.
- The system returns one final numeric score.
This pattern shows up everywhere: machine learning metrics, simulation outcomes, and trading strategy ROI.
What Is a Hyperparameter?
A hyperparameter is a setting you choose before running an evaluation.
Examples:
- A model’s learning rate or tree depth.
- A trading rule’s lookback window or threshold.
- A simulator’s step size or penalty weight.
Unlike learned parameters (like model weights), hyperparameters are not fit directly by gradient descent. You set them, run the system, and observe a final score.
