A personal Jekyll site hosted on GitHub Pages. Features spatial biology image galleries, AU ETF tools (compare any ASX fund, backtest portfolios, track your holdings), a frequent flyer credit card tracker, and an astrobiology-themed cosmic horoscope lab.
Live site: felixillion.github.io
Code: All code in this repository is free to use, adapt, and redistribute for any purpose — personal or commercial — with no attribution required. No licence file is attached; treat it as public domain.
Images: All images in assets/gallery/ and elsewhere on the site are original works by the author and may not be reproduced or used without explicit written acknowledgement of authorship. Please reach out before using them.
.
├── _data/
│ ├── ff_cards.json ← FF card offers (edited by hand)
│ ├── ff_history.json ← past offers (grows weekly)
│ ├── ff_checks.json ← weekly automatic check of each bank page
│ └── astrology_daily.json ← Daily horoscope/planetary data
├── _includes/
│ └── nav.html ← Navigation bar (shared across all pages)
├── _layouts/
│ └── default.html ← Base HTML layout
├── assets/
│ ├── css/
│ │ └── style.css ← Global styles + CSS variables
│ ├── js/
│ │ ├── etf_tools.js ← ETF Tools: compare, backtest, my portfolio
│ │ ├── frequent_flyer.js ← FF points card table
│ │ ├── astrobiology.js ← Cosmic lab / horoscope frontend
│ │ ├── gallery.js ← Spatial gallery renderer
│ │ ├── hero.js ← Homepage animated canvas backdrops
│ │ └── home_interactions.js← Homepage misc interactions
│ └── gallery/
│ └── *.jpg ← Gallery images (author's own — see licence above)
├── data/
│ ├── etf_data.json ← Asset-class returns 1901+ (AU Investing) + hand-checked fund sectors/holdings
│ ├── universe.json ← The ~670 ASX codes whose prices are downloaded
│ ├── astrology_daily.json ← Mirror of _data/ for client-side fetch
│ └── prices/
│ ├── index.json ← Search list + headline returns for every code
│ ├── _quality.json ← What the data sanity checks found
│ └── <CODE>.json ← Weekly prices, total-return index and distributions
├── pages/
│ ├── etf-tools.html
│ ├── frequent-flyer.html
│ ├── astrobiology.html
│ ├── gallery.html
│ ├── cv.html
│ └── privacy.html
├── .github/
│ └── workflows/
│ ├── daily-update.yml ← daily data jobs
│ ├── weekly-prices.yml ← weekly ETF/share prices
│ ├── ticker-request.yml ← "Add ticker: ABC" issues → added automatically
│ └── scripts/
│ ├── fetch_prices.py ← yfinance weekly prices + sanity checks
│ ├── update_etf_data.py ← new asset-class years from the price files
│ ├── generate_astrology_data.js ← Daily horoscope generator
│ └── ff_check.py ← weekly FF link/offer check
├── 404.html
├── index.html
└── requirements.txt ← yfinance>=0.2.54
| Page | Path | Description |
|---|---|---|
| Home | / |
Hero with animated canvas backdrops |
| Gallery | /gallery.html |
Spatial biology image showcase |
| CV | /cv.html |
Academic CV with live ORCID publication fetch |
| Cosmic Lab | /astrobiology.html |
Daily science-themed horoscopes |
| FF Points | /frequent-flyer.html |
AU credit card bonus comparison |
| ETF Tools | /etf-tools.html |
Compare funds, backtest a portfolio, track your holdings |
| Privacy | /privacy.html |
Privacy statement |
Workflow file: .github/workflows/daily-update.yml
| Job | Schedule | What it does |
|---|---|---|
Weekly Price Update (weekly-prices.yml) |
Saturdays | Runs fetch_prices.py — full daily history from Yahoo Finance for every code in data/universe.json, cleaned and saved as weekly files in data/prices/; then update_etf_data.py adds any newly completed year to the asset-class returns. |
Ticker Request (ticker-request.yml) |
When an issue titled “Add ticker: ABC” is opened | Checks the code on Yahoo, adds it to universe.json, downloads it, replies and closes the issue. |
| Generate Astrology Data | Daily, 7am AEST | Runs generate_astrology_data.js — computes real planetary positions (VSOP87-approximated), generates date-seeded horoscope readings, writes data/astrology_daily.json and _data/astrology_daily.json. |
FF Offer Check (ff-check.yml) |
Mondays | Runs ff_check.py — tests each card link, compares bonus numbers with the bank's page, records offer history, opens an issue if anything needs re-checking. |
Manual trigger: GitHub → Actions → Daily Data Update → Run workflow.
All calculation code is in assets/js/etf_tools.js; prices come from data/prices/ (weekly).
Data
fetch_prices.pydownloads full daily history (split-adjusted closes and cash distributions) and builds its own total-return index: distributions reinvested at the ex-date close. Yahoo's "Adj Close" isn't used because it is wrong for some funds and LICs.- Sanity checks: one-day jumps over 40% that reverse within 3 days are treated as bad ticks; a single distribution over 50% of the price is dropped; over 15% is kept but flagged. Findings go to
data/prices/_quality.jsonand the Data Audit. - Saved weekly (last close of each week) to keep files small; dates are implicit (start Friday + 7 days × index).
Compare — buys at the weekly close; DRP on = units grow with the total-return index; DRP off = distributions paid as cash to units held before each ex-date. "Your annual return" is money-weighted (XIRR); "fund total return" is time-weighted. Today's-dollars mode uses ABS CPI (annual average, interpolated).
Backtest — weights, rebalancing at the first week of each year/quarter/month, contributions split by target weight, distributions reinvested. Metrics: CAGR, volatility (weekly, annualised), max drawdown, best/worst year, worst 12 months, Sharpe vs the average RBA cash rate, correlations of weekly returns. Look-through uses the hand-checked sectors/holdings in etf_data.json.
My portfolio — trades saved in the browser (localStorage, key felixillion-etf-portfolio-v1); export/import JSON; CSV import for generic broker exports and CommSec. Average-cost gains; distributions estimated from units held before each ex-date; XIRR includes distributions; the benchmark gets the same cash flows.
Retirement — moved to the AU Investing page (one maintained withdrawal-rate engine).
Franking credits
- Formula:
dividendYield × frankingPct × (0.30/0.70) × (1 − marginalRate) - This is the ATO-correct calculation of the net benefit of imputation credits to an investor at a given marginal rate
- At marginal rate = 30% (= corporate tax rate), benefit is exactly zero (credits offset the tax owed)
- Verified against ATO franking credit worked examples ✓
Cents-per-point (FF Tracker)
- Formula:
(annualFee × 100) / bonusPoints - Two-year version:
(fee × 2 × 100) / (signupBonus + year1Bonus) - Lower = better value ✓
Open pages/cv.html and fill in the CV_CONFIG block at the top:
window.CV_CONFIG = {
orcid: '0000-0000-0000-0000', // ← your ORCID iD
scholarUrl: 'https://scholar.google.com/citations?user=XXXXX',
// ...
};Once orcid is set, publications are fetched live from https://pub.orcid.org/v3.0/{orcid}/works on every page load. No GitHub Action needed. Falls back to fallbackPublications if ORCID is unreachable.
- Place image files in
assets/gallery/ - Edit
_data/gallery.json:
[
{
"title": "My Dataset",
"type": "Imaging Mass Cytometry",
"color": "var(--dapi-blue)",
"image": "assets/gallery/my_image.jpg",
"description": "One-sentence description."
}
]No code changes needed — gallery.js reads from window.GALLERY_DATA injected by Jekyll.
Edit _data/ff_cards.json (one object per card; the _readme at the top explains each field). Example:
{ "id": "anz-ff-black", "program": "qantas", "issuer": "ANZ", "name": "Frequent Flyer Black",
"stages": [{"pts": 50000, "year": 1, "need": "Spend $6,000 in 3 months"}, {"pts": 50000, "year": 2, "need": "Keep the card over 12 months"}],
"minSpend": 6000, "spendDays": 90, "fees": {"y1": 425, "ongoing": 425}, "cashback": 0, "credits": 0,
"link": "https://www.anz.com.au/personal/credit-cards/frequent-flyer-black/", "affiliateLink": "",
"source": "anz.com.au product page", "verified": "2026-11-11" }When you re-check a card, update its figures and its verified date. .github/workflows/ff-check.yml runs weekly (Mondays): it tests every link, checks the bonus numbers still appear on the bank's page, adds changed offers to _data/ff_history.json (the page's "Past offers"), and opens a GitHub issue “FF Points: offers to re-check” when something needs a human. Some banks block automated visitors — those cards just show no automatic tick.
Affiliate links: Set affiliateLink to your referral URL. The JS uses card.affiliateLink || card.link, so it falls back to the official link when empty.
gem install bundler jekyll
bundle install
bundle exec jekyll serveVisit http://localhost:4000.
Cache busting: The layout adds ?v={{ site.time | date: '%Y%m%d%H%M' }} to all JS imports so browsers pick up changes on each Jekyll build.
Push to main — GitHub Pages builds and deploys automatically within ~60 seconds.
The Actions workflow runs separately on its own cron schedule and commits data updates back to main, which triggers a new Pages build.
- FF card offers are edited by hand in
_data/ff_cards.json(banks have no feed). The weekly check tells you when to look. - Horoscope readings: 7 per sign × 12 signs = 84 unique readings. With a daily seed they cycle slowly but will repeat. Expanding the reading pool in
generate_astrology_data.jswould reduce repetition. - Pre-1970 asset class returns are estimated from academic sources and carry ±2–5% uncertainty. Treat retirement simulations starting before 1970 as illustrative only.
- ETF Tools only show real history: a comparison or backtest starts when the youngest fund listed (no stand-in index data). Prices are weekly and update weekly.
- ETF/share prices and distributions: Yahoo Finance via
yfinance, total returns rebuilt from prices + distributions - Asset class history (1970–2024): RBA Statistics, ASX 200 total return index, MSCI data
- Pre-1970 estimates: Dimson, Marsh & Staunton — Global Investment Returns Yearbook (UBS, annual)
- Australian CPI: ABS Cat. 6401.0
- Tax rates: ATO 2025-26 resident individual rates (Stage 3 tax cuts)
- Planetary positions: VSOP87 truncated series (generate_astrology_data.js)
- FF card data: Manually sourced from bank websites
tests/audit_checks.py (run automatically by .github/workflows/data-audit.yml after every data update, and on every push) checks the site against independent references in data/ground_truth/reference_series.json:
| Check | Reference | Fails when |
|---|---|---|
| Freshness | — | Cosmic Lab data > 1 day old, ETF data or weekly prices > 8–9 days old |
| CPI tables (ETF tools, AU investing) | RBA Table G1 | any year differs by > 0.15 pts |
| AU shares asset class | All Ords Accumulation (Market Index) | any year out of tolerance |
| Cash asset class | RBA Table F1 | any year differs by > 0.3 pts |
| VAS weekly total returns | All Ords Accumulation | any year differs by > 2.5 pts (checks the price pipeline) |
| Income tax second bracket | ATO rates by financial year | rate doesn't match the current FY |
| Status Credits | Qantas earning tables | table or tier thresholds differ |
| Planet positions | PyEphem (VSOP87/ELP) | any body off by > 0.5° |
Warnings (don't fail the build): FF offers not re-verified in 60 days, Rich List edition out of date, scraped "proposed" values awaiting review.
If the audit fails, an issue labelled data-audit is opened automatically.
Yearly checklist (Jan–Jun): add last year's CPI and cash rate from the RBA tables, last year's All Ords Accumulation return, next FY's tax rates (by 1 July) and the new AFR Rich List figures (late May) to reference_series.json / _data/billionaires.json; re-verify the FF card offers every ~6 weeks (update verified per card in _data/ff_cards.json).