OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Evaluation & Observability

AuctionGym

AuctionGym is an open-source simulation framework for reproducible offline evaluation of bandit and reinforcement-learning approaches to ad allocation and bidding in online advertising auctions.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

AuctionGym — OpenViST project profile in AI Evaluation & Observability.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use AuctionGym when you need repeatable, controlled offline experiments to compare bidding or ad-allocation approaches in simulated auctions.

Who it is for

  • Researchers benchmarking learning-to-bid methods
  • Practitioners comparing bidder behavior and auction outcomes in controlled simulations

Documented uses

  • Compare advertiser welfare, surplus, and auctioneer revenue as competition changes.
  • Study truthful bidding against value-based bid shading in first-price auctions.
  • Run repeated auction rounds from JSON-defined environment and bidder configurations and review CSV metrics and PDF visualizations.
Before you choose

Requirements & limits

The source repository is public and licensed under Apache-2.0.

  • Experiments are simulated and offline, rather than measurements from live auctions.
  • The provided examples cover competition in second-price auctions and bid shading in first-price auctions.
  • Environment and bidder behavior are specified through JSON configuration files.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Reproduce the truthful-bid second-price example

Run the repository's documented example configuration and locate its generated metrics and visualizations.

Before you start

  • AuctionGym repository source
  • Python environment for the repository
  • The provided config/SP_Oracle.json configuration file

Steps

  1. Use the repository's documented command: python src/main.py config/SP_Oracle.json.

  2. Open the generated results subdirectory for that configuration.

  3. Review the CSV files for raw metrics and PDF files for visualizations.

Expected outcome

A results directory containing configuration-specific CSV metrics and PDF visualizations for the documented truthful-bid second-price auction example.

Guide limitations

  • The README documents the example command and output types but does not specify dependency installation steps.
Documentation 1 ↗
Source-based practical guide

Explore the introductory auction notebooks

Open the included notebooks to examine competition effects and bid shading in simulated auctions.

Before you start

  • AuctionGym repository
  • Jupyter Notebook

Steps

  1. Run jupyter notebook from the repository's main directory.

  2. Navigate to the src directory.

  3. Open the competition-effects or bid-shading introductory notebook.

Expected outcome

The notebooks present simulations of second-price auctions at varying competition levels and first-price auctions comparing truthful bids with value-based bid shading.

Guide limitations

  • The README describes the notebook topics but does not specify further setup or execution details.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
15 October 2026
Original website title
amazon-science/auction-gym · Official homepage title · retained in its source language
How we review sourcesSuggest a correction