Finding
The historical q90 reserve covered 20 of 26 holdout quarters (76.9%), compared with 24 of 26 (92.3%) for a fixed 10% buffer. Its average reserve was 7.2% above the quote. The worst shortfall was $22,176 at an assumed 100,000 gallons. The historical rule missed its nominal coverage target.
Why it matters
Compare protection against budget capacity tied up in reserve. A simpler, larger buffer covered more historical quarters. Use the workbench to inspect price and volume assumptions before applying the idea to actual fleet invoices.
Who could use this
Operations, procurement and FP&A teams reviewing cash reserves for a thirteen-week fuel budget.
Data and scope
1,035 weekly EIA U.S. retail diesel observations through October 26, 2026, including taxes. Historical prices are observed. The 100,000-gallon quarterly exposure and equal weekly purchases are hypothetical.
How the analysis works
79 complete non-overlapping thirteen-week blocks. Each rule uses a prior-week quote and completed historical targets. A 2016 to 2019 validation period selects the reserve window before a 2020 onward holdout. SQL independently verifies realized quarter means and shortfalls.
Limits of the evidence
Retail national prices differ from local fleet contracts and wholesale hedges. Fuel specifications changed over the historical series. There are only 26 holdout quarters, and serial dependence remains possible. Current revised history is not a point-in-time vintage backtest. Historical coverage is not a future probability guarantee. Unused reserve is not savings.