# jankrepl/deepdow

Portfolio optimization with deep learning.

Repository: https://github.com/jankrepl/deepdow
Canonical: https://ross.abutalabs.com/products/deepdow
Homepage: https://deepdow.readthedocs.io
Language: Python
License: Apache-2.0
License Family: permissive
Topics: deep-learning, portfolio-optimization, finance, machine-learning, pytorch, timeseries, markowitz, convex-optimization, stock-price-prediction, wealth-management, trading, allocation
Last push: 2024-01-24T15:56:49+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2404, "days_push": 952, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1181, forks 169 (observed 2026-08-28T04:03:53.995515+00:00)

## What it is
deepdow is a Python library built on PyTorch that connects deep learning with portfolio optimization, allowing networks that output asset weight allocations in a single forward pass. It merges market forecasting and allocation layers into a fully differentiable pipeline trainable with gradient descent.

## Use cases
- train neural networks to allocate portfolio weights end-to-end
- research differentiable portfolio optimization layers
- optimize networks with sharpe ratio or maximum drawdown losses
- experiment with clustering-based allocation algorithms
- backtest buy-and-hold allocation strategies learned from time series

## When to choose
- you want to jointly learn forecasting and allocation with gradient descent
- you need differentiable convex optimization inside a PyTorch model
- you are researching deep learning approaches to asset allocation

## When to avoid
- you need active trading strategies with transaction cost modeling
- you want a reinforcement learning trading framework
- you need a production-ready trading system rather than a research framework

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, trading, math
- domain: fintech, machine-learning, deep-learning, time-series
- platform: python
- tags: portfolio-optimization, pytorch, markowitz, asset-allocation, differentiable-optimization, finance, gpu

## Member repositories
- jankrepl/deepdow (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:53.995515+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T06:25:11.618704+00:00, confidence not recorded.
  - readme: https://github.com/jankrepl/deepdow (fetched 2026-08-28T04:03:53.995515+00:00, sha 501fbc3d87d1)
  - registry_pypi: https://pypi.org/pypi/deepdow/json (fetched 2026-08-29T12:32:44.583764+00:00, sha 3701d7272b0e)
- Data as of 2026-08-30T08:39:29.467469+00:00.