How the commodity strategies were tested
- Rules fixed first: every rule comes from published research and was written down, with the checks below and the pick rule, before its first run. Nothing was tuned after seeing results.
- Data: Yahoo Finance daily bars from 2005-01-03 to 2026-10-16. Front-month futures with the jump at each contract roll removed; days at or below zero (WTI on 20 Apr 2020) are excluded. Round 2 adds the CFTC Commitments of Traders report and charges gold, silver and copper longs the financing rate.
- Costs: 0.05% a side for spread and slippage on every entry and exit. A second run with double costs is kept for each rule.
- Checks (all five to pass; the first four to count as profitable): Enough trades (and, for monthly rules, at least 200 rebalances); 2005-2015: made money with a profit factor of at least 1.2; 2016-today: made money with a profit factor of at least 1.2; Made money in at least 65% of the years; Sharpe ratio since 2016 at least the benchmark's. Benchmark: equal-weight long basket of the 17 commodities.
- What was deleted: every rule that failed any of the first four checks, which was all of them. Their signals, live boards and trade lists are gone from the site; this page keeps their results so nobody is shown only survivors.
How 10-month trend filter (long only) trades
Holds each commodity long only while its price is above its 10-month average; otherwise stays in cash.
- Entry: On the first trading day of each month, buy any commodity whose close is above the average of its last 10 month-end closes.
- Exit: Sold at a monthly check when the close is below that average. Long only, no shorts, no price stop.
- Source: Faber (2007), "A quantitative approach to tactical asset allocation".
Why these signals come with a warning
A signal is only worth following if the rule behind it made money across good and bad years before anyone looked. None of the 12 commodity rules did; 10-month trend filter (long only) came closest and is shown at the owner's request, so its track record will show whether the test was too harsh or right. Publishing a failed rule as if it had passed would most likely cost followers money, so the commodity section shows prices, charts, tools and this honest record alongside it. The full numbers are on the commodity backtest page; the same method picks the live crypto and stock signals.