# ZhengyaoJiang/PGPortfolio

PGPortfolio: Policy Gradient Portfolio, the source code of "A Deep Reinforcement Learning Framework for the Financial Portfolio Management Problem"(https://arxiv.org/pdf/1706.10059.pdf).

Repository: https://github.com/ZhengyaoJiang/PGPortfolio
Canonical: https://ross.abutalabs.com/products/pgportfolio
Language: Python
License: GPL-3.0
License Family: copyleft
Last push: 2021-10-09T19:16:40+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 3216, "days_push": 1789, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1849, forks 757 (observed 2026-08-28T04:05:44.163294+00:00)

## What it is
PGPortfolio is a Python library implementing a deep reinforcement learning framework for financial portfolio management, based on the paper 'A Deep Reinforcement Learning Framework for the Financial Portfolio Management Problem'. It includes a configurable training toolkit with TensorBoard visualization, parallel training, and embedded financial-model-based portfolio algorithms for comparison.

## Use cases
- train a deep reinforcement learning agent to manage a cryptocurrency portfolio
- backtest portfolio management strategies against financial-model-based baselines
- research policy gradient methods for immediate-reward portfolio optimization
- run hyperparameter optimization experiments for trading policies
- compare RL-based portfolio selection with online portfolio selection algorithms
- visualize and log training runs with tensorboard

## When to choose
- you are doing academic research on RL for portfolio management and want a reproducible paper implementation
- you need a configurable framework to experiment with policy network topologies and training data
- you want built-in classical portfolio strategies (from OLPS) as baselines for comparison

## When to avoid
- you need a production trading system with live broker integrations and low-latency execution
- you require modern maintained TensorFlow 2.x or PyTorch support, since the code targets older TensorFlow 1.x
- you want plug-and-play stock or options trading out of the box without writing market adapters yourself

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, reinforcement-learning, trading, data-science, benchmarking
- domain: machine-learning, fintech, data-science
- platform: windows, python
- tags: portfolio-management, deep-reinforcement-learning, policy-gradient, cryptocurrency-trading, backtesting, tensorboard, research-toolkit, cryptocurrency, algorithms, linux

## Member repositories
- ZhengyaoJiang/PGPortfolio (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:44.163294+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 2026-08-30T03:17:16.206119+00:00, confidence not recorded.
  - readme: https://github.com/ZhengyaoJiang/PGPortfolio (fetched 2026-08-28T04:05:44.163294+00:00, sha 26ac0eaf6469)
- Data as of 2026-08-30T08:39:29.467469+00:00.