TideAlphaSPY buy and hold
The market row: one purchase of the SPDR S&P 500 ETF, held for the whole test window.
What it holds
SPY holds the 500 companies of the S&P 500 index at their weight by market value. One purchase buys the whole index. In the TideAlpha tables SPY is what an investor gets for no work and no decisions, so it sets the level every theme book must beat. Returns use the dividend and split adjusted close, so the row is a total return.
The numbers
CAGR and worst loss on month-end values, so the worst loss is a drawdown and not one month. Sharpe on monthly returns at 0% risk-free. One basis for every row: members holdable after 126 sessions and $1M of daily dollar volume, decisions at the close, fills at the next close, 10 bps a side, idle cash at the 3-month bill.
| Book | CAGR | Sharpe | Worst loss | Window | Basis |
|---|---|---|---|---|---|
| SPY buy and hold this book | 11.1% | 0.80 | -51% | 2003-12 to 2026-08 | SPY, one position |
| Equal-weight theme basket | 17.5% | 0.89 | -51% | 2003-12 to 2026-08 | 150 themes, every holdable member |
- Annual volatility 14.6% and Calmar 0.22 on the same basis, the lowest volatility of the five rows in the landing table.
- Calendar years: -37% in 2008 and -18% in 2022, the two worst (monthly curve of results/themestrat_more_a.json).
- Source: results/themestrat.json, read through crashdash/results/oracle_audit.json (rule_books, generated 2026-08-27), which carries the row of each source file named below.
- Every row is in sample. The rules were chosen with the full history in view, and the theme list is the 2026 list, so themes that died without a surviving member are missing. Live results from 2026-08-31 are the only holdout.
What it means for an investor
- In a bad year it falls with the market. It lost 37% in 2008 and 18% in 2022, and the worst loss on month-end values is -51%, from the 2007 peak to the 2009 trough.
- Turnover is zero. One trade, then nothing, so no cost and no work.
- It does not select. It holds the losers with the winners and it cannot leave a falling market.
- It is the only row in the table with no selection bias. No rule was fitted to it, so it is the honest yardstick for every other row.
Live record
This is a benchmark, not a TideAlpha book. It has no rule set to run and no ledger. It is recomputed at every weekly rule-book run and every Oracle run, so every live book is scored against it from the first week.
Read more
- Forward test design. The shared basis, the benchmark set and the three rule books.
- Oracle audit. The backtest of the Oracle rule set against SPY and the basket.
- The Oracle: rules, evidence and research. The technical page: the stage model, the rules and the papers.
The research reports need an account. The Oracle page at /oracle is public.