How financial history becomes wood

Annual rings

One ring

One calendar year. The first year is at the center, the newest at the edge. Partial first and current years use only the observations available and are marked during inspection.

Width

Annual log growth, normalized within asset class, so a bond’s best year is as wide as a crypto asset’s best year. Better years are always wider. The narrowest ring is 0.60 of the widest, so a severe down year is still a ring, not a line.

Color

Annual return, sign and size. Loss and gain use separate color ramps, so a mild loss and a mild gain never share a tone. Wood palette: losses are red-brown heartwood, gains pale sapwood. Market palette: losses red, gains green, with lightness rising from dark red to light green so the sign reads without hue. Magnitude is normalized within asset class; the strongest tones are reserved for exceptional years.

Grain and waviness

Annualized realized volatility on one scale for every tree: a log scale from about 6% to 100%, the 5th to 95th percentile of all featured years. So a Bitcoin year carries more grain than an Nvidia year, which carries more than a bond year. Higher volatility draws denser, heavier grain fibres and a wavier ring boundary; a calm year is near-circular. Volatility never changes a ring’s color or lightness; color is return only. Forest cards are too small for grain, so there the waviness is the only volatility signal.

Drawdown scars

A scar is a whole-history drawdown episode. It qualifies when its depth exceeds a gate set in the asset’s own units — a fall of 1.25 times its median annual volatility, in log terms, never shallower than 10% — and it remains underwater for at least 120 days after its trough or is still open. That is about a 12% fall for a currency fund, 20% for the S&P 500 and 57% for Bitcoin, so a scar means “a major fall for this kind of asset.” Episodes are measured from the all-time high; a second crash from an interim high inside a still-open episode (Bitcoin in March 2020, the U.S. market in 1937–38) is a nested scar, measured from that interim high. On the tree a scar starts in the ring of its low. Its angle is the calendar date of that low, January at the top and time clockwise, so a market-wide event sits at the same angle on every tree. It runs outward until the peak it fell from is reclaimed; a scar that reaches the edge has not recovered. Deeper drawdowns draw wider scars.

Families

Every tree is assigned to exactly one family, the first it qualifies for in this order:

Storm growers

Annualized growth above the collection median with volatility above the collection median — the pioneer species, whatever their age.

Old growth

A record spanning at least fifty calendar years, among the trees that are not storm growers.

Quiet compounders

Positive annualized growth with mean realized volatility below the collection median.

Scarred survivors

At least two qualifying drawdown scars, or one still unrecovered.

Thresholds are the median growth and the median volatility over the 100 featured trees, so a family says how a tree compares with this collection, not with the whole market. A tree that qualifies for several families carries only the first.

Data provenance

The published data contains derived annual returns, realized volatility, maximum drawdowns, recovery dates, and summary statistics—not the underlying daily price series. Coverage and latest observation dates vary by asset. Published annual and monthly returns are rounded to a whole percent, so the record cannot be inverted back into a provider's underlying closing prices; volatility, drawdown, and growth-rate statistics keep their full precision.

The interface reports the latest included price date globally and on each tree. The committed artifact is validated before deployment.

Equity, fund, bond, commodity, and currency price history is provided by Twelve Data, whose daily record begins in 1970. Crypto market data is data provided by CoinGecko. The century-long market and industry series are computed from the Kenneth R. French Data Library (daily Fama/French research factors and 12 Industry Portfolios, 1926 onward), compounded here into total-return indexes.

Five trees are built from public-domain U.S. government data and one from the IMF. The 10-year and 30-year Treasury trees are computed from the Federal Reserve Board's constant-maturity yields via FRED (1962 and 1977 onward): each day the holder owns a par bond issued at the previous day's yield, earns its coupon, and reprices it at the new yield, so the series is the total return of a bond rolled daily, not a vendor index. WTI, Brent and Henry Hub natural gas are daily spot prices from the U.S. Energy Information Administration (1986, 1987 and 1997 onward). Copper is the monthly price from the IMF Primary Commodity Prices (1992 onward), transformed here into annual rings; its volatility is annualized from twelve observations a year and is not comparable with a daily series. Gold is the spot price (XAU/USD) from Twelve Data, from December 1979.

The U.K. stock market tree is the oldest on the site: the Bank of England's spliced monthly share price index from A millennium of macroeconomic data for the UK (Open Government Licence v3.0), April 1709 to April 2017, continued to the present by the OECD Main Economic Indicators share price index for the United Kingdom via FRED, scaled to match at the join. The two agree closely over their sixty years of overlap. It is a price index, so dividends are not included and its growth reads low beside the total-return trees; before 1825 it averages three companies (the Bank of England, the East India Company and the South Sea Company), and it is monthly throughout, so its volatility is annualized from twelve observations a year.

Important limitations

• Equity, fund, bond, commodity and currency series are dividend- and split-adjusted closes (total return), but the vendor’s dividend adjustment is thinner before 2013, so early growth for long-lived equities reads a little low (Coca-Cola since 1970 compounds about 1.5 points a year below a fully adjusted series). The century series are total-return indexes by construction. Crypto is a spot price with no yield. Each tree states its basis under Data notes.
• Partial calendar years are not directly comparable with complete years. A year counts as complete when it holds at least 60% of the asset’s own trading days (about 150 for market-hours instruments, 219 for crypto).
• Annual returns run from the prior year’s last close to this year’s last close. Because published returns are rounded to a whole percent, accumulated-growth multiples on tree pages are shown to two significant figures and are derived from the full-precision annualized growth rate over calendar time.
• Featured crypto assets are those with the longest and most liquid records, which are also the ones that survived; the crypto long tail in the lake has a far weaker record than what the collection shows.
• The universe is curated and survivor-biased; absent or delisted assets cannot appear.
• Width and color are graded within each asset class, so a bond’s best year is as wide and as pale as a crypto asset’s best year; they compare a year with its peers, not across classes. Grain and waviness are the exception: volatility sits on one scale for all trees.
• Historical performance is not predictive. Money Tree Rings is educational and is not investment advice.

Open methodology

The complete encoding contract is maintained with the source code. Visual encodings should change only when their corresponding data contract and documentation change together.

Data terms and trademarks

Data is provided as is. It may be delayed, incomplete or inaccurate, and it is shown without warranty of any kind. The site is educational and is not investment advice. Company, fund and index names are trademarks of their owners and appear here only to identify the asset a tree describes; no affiliation with or endorsement by any of them is implied.

Privacy

Money Tree Rings does not use accounts, advertising trackers, or application cookies, and it does not collect user-submitted personal information. Page views are counted with Cloudflare Web Analytics, which sets no cookies and does not identify individual visitors. The hosting provider may process routine request metadata for security and delivery.