Distributional Refinement Network: Distributional Forecasting via Deep Learning

By Benjamin Avanzi, Eric Dong, Patrick J. Laub, Bernard Wong

Rating

1722
Battle Count: 82

Relevance

7/10
While focused on insurance, the distributional forecasting approach could be adapted for financial time series prediction and risk modeling in quantitative trading.

Implementation Complexity

8/10
Requires deep learning expertise and careful integration with baseline models. Hyperparameter tuning and regularization add complexity.

Reproducibility

4/5
The paper provides detailed methodology, hyperparameter ranges, and evaluation metrics. Code is available on GitHub.

About this paper

Methodology: Distributional Refinement Network (DRN). Problem types: Distributional Regression, Probabilistic Forecasting.

The interactive Everscope explorer (charts, battles, favorites) loads below.