Each January, buy the previous calendar year’s best-performing S&P 500 year-end members, ranked by price gain. The Top 2 and Top 3 portfolios begin each year equally weighted; weights then drift. A stock is sold when its regular-session daily low crosses the selected percentage below its entry price. Its allocation stays in cash, earning zero, until the next January. Other holdings keep their original allocations. This is a fixed entry stop, not a trailing stop, so a portfolio’s peak-to-trough drawdown can exceed 35%.

The defaults follow the Gurgavin thread’s conventions: prior December closing entry and a fill exactly at the stop. Those are idealised. Optional first-session opening entry avoids assuming a trade at the same close used to select winners; gap-aware fills sell at the opening price if the stock opens below its stop. Other slippage, fees and taxes are excluded. Dividends use adjusted-close factors, including for stop proceeds. This approximates a dividend-reinvested return series; it is not an exact payment and reinvestment ledger. Benchmark: SPY, a tradable S&P 500 proxy.

Top 1 begins in 1997. Reebok in 2001, Providian in 2003 and original Avaya in 2004 use sourced year-end prices and reported quarterly ranges. No daily prices are inferred for these years: marked dots and dashed connections show annual observations. All other Top 1 years use daily OHLC. Top 2 and Top 3 begin in 2014; earlier runners-up are not reconstructed. Start-year choices follow the selected portfolio’s coverage. Earlier Top 1 picks follow the original thread and are checked against dated index membership. Recent top-three picks come from contemporary performance tables. Newly listed year-end members are ranked from their first available close, matching the thread’s treatment of Sandisk. Its 2025 ranking begins with the 13 February when-issued quote. Cash dividends are excluded from ranking; provider corporate-action adjustments, including spin-offs, are retained.

The three annual records have no common dividends, so their source close-to-close returns also equal total returns. Their reported calendar-year lows determine whether an ideal fixed-entry stop was touched, but the touch date is unknown; any resulting stop is shown at year-end and labelled as occurring during that year. Opening-entry and gap-fill options are available only for periods with daily data. Drawdown for mixed-frequency periods uses observed portfolio values and can understate the true intrayear drawdown. Fully daily drawdown uses closing valuations, rather than intraday portfolio peaks and troughs. Earlier source prices rounded to cents are retained at that precision; results are a reconstruction, not an exact replication of the thread’s endpoint.

Annual sources: Reuters Reebok price history and Reebok quarterly ranges; contemporary 2002 and 2003 Providian closes and Providian ranges; dated Avaya 2003 brokerage holdings, 2004 pension holdings and reported price ranges. Full provenance is in the downloadable dataset.

Older daily prices are recovered from the archived Quandl WIKI dataset, including the original Family Dollar, XL and Cabot Oil & Gas shares. Older WIKI OHLC prices use split adjustments only; adjusted closes include dividends. Recent provider histories also apply corporate-action adjustments such as the WDC/Sandisk separation. Original ticker histories are preserved instead of substituting a relisted company or its acquirer. The default thread return basis uses stock price returns through 2011 and dividend-adjusted stock returns from 2012, with SPY dividend-adjusted throughout. Controls also allow price-only or dividend-adjusted returns across the whole period.

Daily prices: Yahoo Finance; recovered Yahoo archives from Paul Timothy Mooney and Jake Wright preserve acquired stocks. Marathon Oil’s completed December 2022 bars replace an archive’s partial final day after overlapping prices were checked. Hess’s 2023 dividend adjustments are rebuilt from its four declared quarterly dividends. Missing held-day prices stop publication; they are never filled with synthetic returns.

Updates run three times daily and admit completed prior-day sessions. Year-end membership is archived from State Street SPY holdings; subsequent annual selections rank every verified year-end member, with dated iShares IVV holdings as a recovery source. Missing membership or prices prevent an unverified annual rotation. The updater requires all expected completed NYSE sessions and exact year-end anchors, verifies first-listing dates and normalizes share-class ticker names. Failed refreshes retain the last good snapshot and show a notice immediately; observations more than seven days old are also flagged.

The calculation engine matched an independent implementation across 1,800 settings combinations. All 6,738 daily sessions, source-price reconstruction, cash allocations, annual rotations and stop events were checked. The 2026 picks remain SNDK, WDC and MU, selected using 2025 gains.

Through 9 October 2026, the default $100 / Top 1 / 35% stop reconstruction ends at $1,528,104 versus the thread’s $1,514,581, a 0.9% difference that has not been fully reconciled. The same four stop years agree. Three older years remain annual observations only; complete independent historical constituent rankings are not claimed. Daily drawdown uses closing values. Ideal fills and adjusted-close dividends are modelling assumptions.

Updates now check the exchange calendar, exact year-end anchors and first-listing metadata. Failed refreshes retain the last good data and show a notice on this page.