Permutable coffee strategy returned 59% while coffee futures lost 6.8%

Coffee rose briefly in early January, then fell through the spring to a low on 9 June, 26.5% below where it started, and finished the period −6.8%. A holder of coffee futures finished −6.8%. The Permutable strategy, which reads the direction of coffee news every hour and can be long, short or flat, earned 58.8% after trading costs: it was short for much of the decline, turned long three days after the low, and finished 6 months out of 11 in profit against 4 for coffee. Over the same window the book's own weighted coffee P&L, before this page's assumed spread, was +5.3 points.

This page sets out what the strategy does and how it decides, the decisions that delivered most of the return, where the outperformance came from, and the basis on which the figures are measured.

The lines separate in February, when the strategy gained 12.7% while coffee lost 13.0%, and again in June and July, when coffee rose 9.4% and 9.9% and the strategy gained 22.7% and 19.9%. Part of the June and July gap is an effect of position sizing, explained under source of outperformance. All figures are after costs.

A decision is a confirmed change in the strategy's view of coffee. Three decisions delivered about 52 of the strategy's 59 points of return after costs. The Sell decisions beat coffee outright; a Buy can at best match coffee over its own days, and the June Buy did, so its value was the timing. Each trade is recorded in the trade log with its entry and exit dates.

The 9 long trades matched coffee, as any long trade does: they earned 30.8% after costs (31.3% before), the same as coffee over the same days on the same basis. The timing edge came from the 8 short trades, which earned 28.0% after costs (28.4% before) while coffee moved −28.4% from the entry prices over those days, a gain of 56.7 points on holding coffee before costs. On the days the strategy had no position, coffee moved −2.9 points. Trading costs took 0.9 points: 17 round trips at 5 bps each. The book's own weighted coffee P&L over this window is +5.3 points, before the spread assumed here.

A further 6.8 points comes from position sizing. Each trade is one unit of notional at its own entry price, while holding coffee is one unit at the start of the year. By mid-June coffee was 28% below its January level, so the trades after that, notably the 12 June Buy, held about 1.4 times as many contracts per unit of notional as a holder from January. This is a sizing effect, not timing.

The result is concentrated. The 12 June Buy alone earned 30.2 of the 58.8 points; without it the strategy made 28.5% against −6.8% for coffee.

The strategy is rules-based. Every step below is fixed in advance and applied the same way throughout the period.

The settings were chosen using data up to 31 December 2025. That window starts in April 2025 and includes 11 trades. The period on this page, 1 January to 2 November 2026, falls after that date and was not used to choose them.