AFL 23-0 — All-Era Team Builder

Status, September 2026: GitHub Actions are currently paused on this repo, so automated data refreshes and deploys are not running. The data and any live site reflect the last build.

▶ Play it: afl23-0.com

Inspired by 23-0.com, rebuilt for Aussie rules. Spin a random club + decade from 130 years of VFL/AFL history, pick a player from that pool, place him on the oval — then run a 23-game season simulated against the real teams of those eras and chase a perfect 23-0.

Every rating, salary and simulation input is derived from real scraped data — no hardcoded players, no invented numbers. Data access patterns follow the open-source fitzRoy project (the package itself is not used).

Game modes

Mode	Squad	Twist
Classic 5	DEF · MID · RUC · FWD · UTL	2 re-rolls
Full 23	back six · forward six · 2 wingers + centre · 2 followers · 1 ruck · 5 bench	5 re-rolls
Salary Cap 23	as Full 23	$21.2M cap, pools listed by price, 3 lifelines to swap a picked player (from his original spin pool)

Every mode has an era multi-select — play all 14 decades (1890s–2020s) or any subset; selecting one decade plays a pure single-era game judged against that decade's real teams. Picks show the top 20 of the rolled club+era (by rating, or by salary in cap mode) with search across the entire pool. Players can be moved around the field between turns — scores recompute for their new role (off-position players are graded against the era's real specialists in that role).

Data sources

• afltables.com — per-season player stats grouped by club (1897→now), every match result, Grand Final lineups, Brownlow tallies 1924–1964
• footywire.com — Brownlow tallies + winners 1965→now, All-Australian teams 1991→now, Rising Star, player profiles/positions (requires browser User-Agent)
• wikidata.org — playing positions for pre-1965 players

How the rating engine, salaries and simulation work: see /about in the app.

Running it

npm install

# 1. scrape everything (cached & resumable; first run ~15 min)
npm run scrape

# 2. crawl footywire profiles for positions (long; resumable; prioritised by career games)
npm run positions

# 3. compute ratings/salaries/strengths and export JSON for the web app
npm run export

# 4. play
npm run dev          # http://localhost:3000

Keeping the 2020s fresh

npm run refresh

Re-scrapes only current-decade stats, results and awards, recomputes and re-exports. Ratings use per-game averages, so established players stay stable mid-season — refreshes mainly add new players, new games and end-of-season honours.

Layout

pipeline/   TypeScript ETL: scrapers (cheerio) -> SQLite -> JSON artifacts
web/        Next.js app; reads web/public/data/*.json; sim runs client-side