Three Clocks Sit Behind One Short Interest Field Open a US equity on almost any broker page and you will find a line labeled short interest: a share count, often a percentage of float beside it, sometimes a days-to-cover figure. The field sits next to the last sale and the session volume, both of which update in seconds, and it quietly inherits their air of currency. It should not. The short interest field is a photograph of a settlement date that has already passed, developed and released on a calendar FINRA publishes a year in advance. On September 22, 2026, the most recent FINRA short interest figure a US screen can be showing comes from the August 31 reporting settlement date. Member firms filed it by 6:00 p.m. Eastern on September 2. FINRA released it on September 10. The next figure, capturing September 15, does not reach the public until September 24. So for twelve calendar days, the field labeled short interest has been describing the last Monday in August. That is not a ...
A One-Cent Gasoline Move Is 28 Cents on a 3-2-1 Crack Spread
EIA’s Cushing WTI spot price rose $4.45 on September 1, 2026, against the previous published day. A widely published U.S. refining spread — the 3-2-1 crack built at New York Harbor — fell $10.90 across that same pair of days. Both figures come out of one table of daily spot assessments, and neither contradicts the other.
A crack spread is not a quote that prints on a screen the way a futures contract does. It is an arithmetic construction with fields: a ratio, a gasoline assessment, a distillate assessment, a crude benchmark, and a price type. Change one field and the answer moves by dollars per barrel. On September 1 the published EIA components support answers from $43.05 to $106.93 per barrel — a spread of $63.88 between two numbers that traders routinely call the same thing.
What follows works the arithmetic through the EIA daily spot table released on September 2, 2026, then separates what a one-session move in the spread says about crude from what it says about products. EIA publishes the component prices; the spreads below are this article’s calculation.
The 42-Gallon Conversion Sets the Leg Weights
EIA’s glossary defines the unit the entire calculation rests on: “A unit of volume equal to 42 U.S. gallons.” Crude is assessed per barrel, U.S. spot products per gallon. A crack spread has to move both into one unit before it subtracts.
The 3-2-1 ratio treats three barrels of crude as yielding two barrels of gasoline and one barrel of distillate. Stated per barrel of crude:
Distribute the constant and the sensitivity profile falls out. Gasoline carries 2 ÷ 3 × 42 = 28. Distillate carries 1 ÷ 3 × 42 = 14. Crude carries 1, with the sign reversed. So one cent per gallon on the gasoline assessment is 28 cents per barrel on the spread, one cent on distillate is 14 cents, and one dollar on crude is one dollar the other way.
Those three coefficients explain most of what looks strange in a crack chart. For the spread to absorb a $4.45 crude move with distillate unchanged, gasoline alone would have to add 15.9 cents per gallon (4.45 ÷ 28). Product assessments do move that far in a session, but rarely, and a trader reading the spread as a crude indicator will keep being surprised by sessions where the crude leg is the smallest of the three contributions.
One Session, Three Legs, and a $10.90 Fall
Running the formula across the six published days in EIA’s daily table gives three parallel series, one per harbor. Gulf Coast and New York Harbor use conventional gasoline and ultra-low-sulfur No. 2 diesel; Los Angeles uses RBOB and ULSD. Each of the three subtracts WTI at Cushing.
Day
NY Harbor
Gulf Coast
Los Angeles
Aug 25
65.82
71.14
83.09
Aug 26
68.87
74.31
88.66
Aug 27
70.77
76.72
90.60
Aug 28
75.10
81.05
97.46
Aug 31
75.24
83.97
98.95
Sep 1
64.34
71.20
86.70
The New York Harbor drop from 75.24 to 64.34 decomposes exactly, because the formula is linear. Gasoline went from $3.550 to $3.203, or −34.7 cents, which is −$9.72 per barrel at a weight of 28. ULSD went from $4.491 to $4.724, or +23.3 cents, worth +$3.26 at a weight of 14. Crude added $4.45, which enters as −$4.45. The three contributions sum to −$10.90.
Los Angeles gives the same shape with different sizes: RBOB −39.5 cents (−$11.06), ULSD +23.3 cents (+$3.26), crude −$4.45, total −$12.25. The crude leg is identical in both because both use the same benchmark. Everything else in the difference is products.
That decomposition is the part a headline cannot carry. Gasoline assessments fell together in every one of the three markets that session — −34.7 cents at New York Harbor, −42.0 on the Gulf Coast, −39.5 in Los Angeles — while distillate rose at each (+23.3, +24.6 and +23.3 cents). A move that hits one product family in every region and leaves the other family moving the opposite way is a product-leg event. Reading the same session as “crude rallied, margins compressed” keeps the arithmetic but loses the cause.
The week also shows how little the harbors converge. Gulf Coast ended the run within six cents of where it started, at 71.20 against 71.14 on August 25. New York Harbor finished $1.48 lower than its own August 25 reading, and Los Angeles finished $3.61 higher than its own. A trader tracking one harbor and quoting it as “the crack” would have described three different weeks.
The Same Day Prices Eight Different Answers
Because the ratio form is linear, a ratio crack is exactly the weighted average of the single-product cracks underneath it. On September 1 at New York Harbor, gasoline alone was $3.203 × 42 − 91.48 = $43.05 per barrel, and ULSD alone was $4.724 × 42 − 91.48 = $106.93. Those two numbers generate the rest.
$10.65 separates the 3-2-1 and 2-1-1 answers on one day, from one set of prices, with no disagreement about any input. The ratio is a yield assumption, and when the distillate crack sits $63.88 above the gasoline crack, the assumption dominates the result.
Location does the same work. The identical 3-2-1 ratio on September 1 gives $64.34 at New York Harbor, $71.20 on the Gulf Coast and $86.70 in Los Angeles — a $22.36 gap between the first and the last, larger than any single-session move in the six days above. A Gulf Coast jet fuel crack that day, taken alone, was $84.50. Eight defensible constructions, one calendar day, $43.05 to $106.93.
What Breaks When the Series Is Stitched
Four failure modes show up when these numbers become a time series rather than a single reading.
Grade mismatch inside one column. EIA’s New York Harbor and Gulf Coast gasoline series are conventional regular; the Los Angeles series is RBOB. Both are gasoline, and they are not the same specification. Pasting them into one “U.S. gasoline” column produces a spread series whose jumps are partly definitional.
Days that exist on one leg only. The same daily table carries a WTI price for August 31, 2026, and no Brent price for that date. Any Brent-minus-WTI spread on that day has to be either skipped or invented. Forward-filling the missing leg manufactures a spread print that no assessment supports, and a backtest that trades off spread thresholds will happily take that print as a signal.
Averaging before subtracting. EIA states its method plainly: “Weekly, monthly, and annual prices are calculated by EIA from daily data by taking an unweighted average of the daily closing spot prices for a given product over the specified time period.” When both legs cover the same days, the spread of the averages equals the average of the spreads. When one leg is missing a day, that identity fails, and a weekly crack silently stops matching the daily series it is supposed to summarize.
Rounding before weighting. The 42-gallon conversion magnifies whatever precision is lost at the leg. A gasoline assessment rounded to the nearest cent carries up to half a cent of error, which the 3-2-1 weight of 28 turns into 14 cents per barrel on the spread; the distillate leg adds up to another 7 cents at a weight of 14. That is small against a $10.90 session move, and larger than the six-cent round trip the Gulf Coast series made across the same week.
One more distinction sits underneath those four failure modes. EIA defines a spot price as “The price for a one-time open market transaction for immediate delivery of a specific quantity of product at a specific location where the commodity is purchased ‘on the spot’ at current market rates.” That is an assessment of physical trade, not a futures settlement. A crack built from spot assessments and a crack built from exchange settlements are two different series with the same name.
What Would Invalidate This
A crack spread is not refinery profit, and treating it as one is where this framing breaks first. The calculation leaves out operating and energy costs, renewable fuel obligations, the value of secondary products, and the discount or premium of the crude a refiner actually runs against the benchmark in the formula. A refiner buying a heavier, cheaper grade than WTI at Cushing earns a different margin than the spread shows, in either direction.
The crude leg is a proxy, not a purchase record. The Cushing benchmark is used here because EIA publishes it daily alongside the products, not because it matches any specific plant’s crude slate.
The assessments are third-party judgments. EIA lists Refinitiv, an LSEG business, as the source of these spot prices. Two vendors assessing the same barrel on the same day can differ, which puts a floor under how precisely any of these spreads can be compared across sources.
The September 1 decomposition also shows what moved, not why. Seasonal specification boundaries are one live candidate for a same-direction gasoline move: EPA states that “For all regulated parties except retailers and wholesale purchaser-consumers, the summer season is generally May 1 to September 15; for retailers and wholesale purchaser consumers, the summer season is generally June 1 to September 15.” A single session cannot separate a specification or contract-roll artifact from a genuine supply event, and this article does not claim to.
Finally, the whole argument shrinks when the product cracks converge. The ratio choice mattered by $10.65 on September 1 because distillate stood $63.88 above gasoline. In a period where the two sit within a dollar or two of each other, 3-2-1 and 2-1-1 answer nearly the same, and the fields that dominate here become rounding.
Concrete Framework
A checklist for anyone building or reading a crack spread series.
Write the five fields down before plotting. Ratio, gasoline series, distillate series, crude benchmark, price type. A series without those five recorded is not reproducible, and a chart labeled only “crack spread” cannot be compared with another one.
Convert once, at the leg. Multiply gallon prices by 42 and apply the ratio weights — 28 for gasoline and 14 for distillate in a 3-2-1 — then subtract crude. Rounding the legs to cents before weighting introduces error the weights then multiply.
Decompose every move larger than a threshold set in advance. Split the change into gasoline, distillate and crude contributions, as done above for the $10.90 fall. If the crude contribution is not the largest, a crude-driven narrative is not supported.
Check whether product legs moved together across regions. Three harbors falling together on gasoline while distillate rises points at the product side, not at the barrel.
Keep the daily grid intact. Mark missing legs as missing rather than filling them, and recompute weekly figures from days present on both legs.
Match price types end to end. Do not mix a spot-assessed product leg with an exchange-settled crude leg, and do not compare a spot-built history to a futures-built one without relabeling.
Size positions off the leg weights, not the headline. In a 3-2-1, a one-cent gasoline error is 28 cents per barrel of exposure; anyone hedging or trading a refining spread should be sizing against that multiplier rather than against the spread level.
None of the above is investment advice, and the spreads shown are calculations from public data rather than tradable quotes. Anyone acting on refining spreads should verify current assessments and contract specifications from the primary sources before committing capital.
Comments
Post a Comment