How we test.

Every chart follows the same rules. This page is the fine print, kept off the charts on purpose.

The rules

• Every idea is compared with the S&P 500 (SPY) or the Dow (DIA) over the same window, dividends included.
• Daily closes, long only, no leverage. Cash earns nothing.
• Rebalancing happens on the first trading day of January unless the rule says otherwise.
• Trading costs of 0.1% per side are charged on every trade.
• Rankings only use data available before the rebalance date.
• Everything is recomputed automatically every week. Numbers move; the rules don't.

Where the data is weak

Prices come from Yahoo Finance. S&P 500 membership is rebuilt from Wikipedia's add and remove history, so companies that were later dropped are candidates up to the day they left. Yahoo has no history for many delisted names, mostly bankruptcies and buyouts, so those are missing. This makes strategies that pick individual stocks look a little better than they really were. A sanity check, buying every stock in our reconstructed universe equally, lands close to the real equal-weight S&P 500 ETF, which is why we publish the charts at all.

Notes by idea

• Buy Last Year's Losers, Chase Last Year's Winners, Own the 10 Biggest: Includes stocks later dropped from the S&P 500 where price history exists (135 of 345). The rest are missing, which can flatter momentum-style ideas.
• Buy Last Year's Losers: Survivorship works against this idea: a stock that fell so far it was deleted from the index (or went to zero) may not appear here, flattering the results. Treat as illustrative.
• Buy Last Year's Losers: 189 universe tickers could not be loaded and were dropped.
• Chase Last Year's Winners: 189 universe tickers could not be loaded and were dropped (survivorship note).
• Own the 10 Biggest: Shares-outstanding data was missing for many names, so the '10 biggest' were approximated by share PRICE rank for some or all years. This is a rough proxy for market cap.
• Own the 10 Biggest: 189 universe tickers could not be loaded from yfinance and were dropped (survivorship note).
• Missing the Best (and Worst) Days: This is illustrative: nobody can systematically avoid only the worst days or only miss the best. The best and worst days cluster together, usually during volatile periods.
• Sell in May and Go Away: Cash earns 0% here; a real 'cash' position would earn short-term interest, which would help this seasonal strategy in high-rate years.
• Dogs of the Dow: Dividend yield is estimated from trailing 12-month dividends over price; small timing differences vs the exact index-date yield are possible.
• Dogs of the Dow: Dow 30 membership uses the current constituents for all years (no historical Dow reconstruction), so older years carry mild look-ahead in the candidate set.
• The Magnificent 7: This is the definition of hindsight: the “Magnificent 7” label was applied AFTER these names had already won. Picking them in 2015 would have required foresight nobody had.
• The 200-Day Line: Cash earns 0% here; real cash would earn short-term interest, helping the strategy in high-rate stretches. Signals are read at each prior month-end, so there is no look-ahead.
• Dual Momentum: BND (total US bond market) is used as the safe asset; Antonacci's original rule compares against Treasury bills, which behave differently when rates move fast.
• The Turn of the Month: The last three trading days of a month are only known in advance from the exchange calendar, which is exactly how a real trader would use this rule.
• The Turn of the Month: Cash earns 0% for the ~75% of days spent out of the market, and switching in and out twelve times a year costs 0.1% per side each way.
• The Golden Cross: Cash earns 0% here; a real cash position would earn short-term interest, which flatters the rule in high-rate stretches.
• The Golden Cross: Crossovers are slow signals: they typically sell well after a top and buy back well after a bottom, which is exactly what the drawdown numbers below capture.
• The Permanent Portfolio: Browne's original sleeves are stocks, long bonds, gold and cash; SHV (1-3 month Treasuries) stands in for cash, so this line does earn the T-bill yield rather than nothing.
• The 90/10 Crash Hedge: The whole point of the 10% sleeve is the annual rebalance: after a crash it is sold to buy cheap stocks, which is where most of its value comes from.
• The 90/10 Crash Hedge: TLT is long-duration, so the hedge itself can fall hard when rates rise - as 2022 showed.
• A 5% Slice of Bitcoin: The monthly rebalance is doing enormous work here: it repeatedly trims Bitcoin's gains into stocks. Buying and holding 5% instead would give a wildly different - and far more volatile - answer.
• A 5% Slice of Bitcoin: Bitcoin trades every day of the week while SPY does not; this backtest uses only days on which the stock market was open.
• A 5% Slice of Bitcoin: Prices are Yahoo's BTC-USD series and ignore exchange, custody and spread costs, which were substantial in the early years.
• Gold, Only in an Uptrend, Oil, Only in an Uptrend, Long Bonds, Only in an Uptrend: Cash earns 0% while the rule is out of the market; a real T-bill sleeve would add to this line, especially in high-rate years.
• Gold, Only in an Uptrend: Gold has no earnings and no yield, so a trend rule is one of the few systematic ways to hold it; that also means the rule has nothing but price to work with.
• Gold, Only in an Uptrend: Whipsaws are the cost: each false signal pays 0.1% per side and gives up the snap-back that usually follows.
• Oil, Only in an Uptrend: USO holds futures and rolls them monthly. In contango that roll bleeds value continuously, so the fund drifts down even when crude is flat — the buy-and-hold line is a much worse investment than the oil price itself.
• Oil, Only in an Uptrend: In April 2020 crude futures traded below zero and USO changed its holdings mid-crisis; no simple rule fully captures what that did to holders.
• The Copper/Gold Switch: CPER launched in 2011, so this window is shorter than the site's usual fifteen years — and it contains only two real growth scares, which is a small sample for a macro rule.
• The Copper/Gold Switch: The ratio is compared with its own level three months earlier using closes up to the rebalance day only, so there is no look-ahead — but the signal is all-or-nothing, which makes it far more volatile than the 60/40 it is measured against.
• The Copper/Gold Switch: Copper futures funds carry roll costs that a spot copper/gold ratio would not, so the signal here is slightly noisier than the textbook version.
• Long Bonds, Only in an Uptrend: A bond fund's return is mostly coupon income, which you forfeit while sitting in cash at 0% here — the real version would at least earn T-bills, which in 2022-2023 paid more than the bonds being avoided.
• Long Bonds, Only in an Uptrend: TLT is the most rate-sensitive Treasury fund there is; the same rule on intermediate bonds would trade far less.
• Semiconductors vs the Market: Both lines start on the same day — the later of the backtest window and SOXX's own inception — so neither fund gets a head start.
• Semiconductors vs the Market: Semiconductors are a boom-and-bust industry: the drawdown below is the price of admission, and it has repeated in every cycle.
• Semiconductors vs the Market: This is a sector fund, not a stock pick — no hindsight selection of winners is involved.
• Small Biotech vs Big Biotech: Both lines start on the same day — the later of the backtest window and XBI's own inception — so neither fund gets a head start.
• Small Biotech vs Big Biotech: Equal weighting inside biotech is effectively a small-cap, pre- revenue bet: XBI's swings are far larger in both directions.
• Small Biotech vs Big Biotech: Neither fund picks winners with hindsight — both follow published index rules — so this is a clean test of weighting, not stock selection.

Raw numbers for every idea: JSON API. Educational only, not investment advice.