The Barrel in 30 Currencies

Why the same barrel is a different price in every country

Crude is quoted and settled in United States dollars almost everywhere. A buyer who does not earn dollars therefore faces a compound price: the dollar price of the barrel, multiplied by the number of local units a dollar costs. Either factor can move on its own, and they frequently move in opposite directions.

The consequences run both ways and are routinely attributed to the wrong cause. A country whose currency is weakening imports the full dollar increase and adds its own depreciation on top, which is how a pump price can set local records during a period when the dollar oil price is flat. A country whose currency is strengthening is partly insulated, and its motorists experience a smaller shock than the headline implies. An oil exporter that sells in dollars and pays wages in local currency gets a revenue boost from depreciation that has nothing to do with the barrel — the mechanism sitting underneath the fiscal breakevens on the breakevens ledger.

So “oil is expensive right now” is an incomplete sentence without a currency attached. The table below makes the incompleteness visible.

One barrel of Brent and one barrel of WTI, priced in each currency

Crude quotes: opening quote, October 23, 2026. Exchange rates: the October 23, 2026 fixing. Listed in the publisher's own order rather than by size, because reordering a reference set invites a reader to rank rows denominated in different units.

Rate source: European Central Bank euro foreign exchange reference rates, eurofxref-daily.xml, published as the 16:00 Central European Time concertation fixing on TARGET business days; the 23 October 2026 fixing, read from the publisher's own daily page on October 24, 2026. Crude source: EIA spot quotes, opening quote, October 23, 2026. The units-per-dollar column is the Institute's computation — each currency's euro rate divided by the dollar's euro rate of 1.1193 — and the two barrel columns are that figure multiplied by $113.12 and $99.36. Decimal places vary by currency because a rupiah figure to four decimals would be false precision and a franc figure to none would discard real information.

Two rows repay close reading, for opposite reasons. The Danish krone at 7.4748 per euro is held inside a narrow band against the euro by policy, so its barrel price moves almost entirely with the euro–dollar rate and hardly at all on its own account. The Turkish lira at 55.6715 per euro is the other extreme: a barrel costs 5,626 lira, and the size of that number is a statement about the currency rather than about oil. Comparing those two barrel figures tells a reader nothing. Comparing each against its own history tells them a great deal.

One publisher, one date, one fixing — and why that beats coverage

Every rate below comes from one publisher, on one date, at one time of day. That is the entire reason to use a single central bank's reference set rather than assembling quotes: a table of thirty currencies drawn from thirty sources is a table with thirty clocks in it. The ECB publishes against the euro, so each dollar-per-barrel figure here is computed as the barrel price in dollars multiplied by the currency's euro rate divided by the dollar's euro rate. 🔴 THE SIGNED GAP CLOSED AND THE STEP MATCH BROKE, WHICH IS THE EXACT INVERSE OF WHAT THE ENTRY BELOW RECORDED, AND TWO CONSECUTIVE SWEEPS MOVING OPPOSITE MEASUREMENTS IS THE STRONGEST AVAILABLE EVIDENCE THAT THEY ARE SEPARATE. The entry below reports legs step-matched at two with the signed gap unchanged at rate-ahead-one-day. This compilation reports the opposite on both counts. The fixing leg advanced THREE TARGET business days, from the 20 October fixing through 21 and 22 October to the 23 October fixing. The barrel leg advanced FOUR market sessions, from the October 19 open through October 20, 21 and 22 to the October 23 open. Three steps against four, so the step match is BROKEN with the barrel leg ahead by one step — while the signed gap, which the entry below found unmoved for two consecutive compilations, has gone from rate-ahead-one-day to an October 23 barrel at a 23 October fixing. Last sweep the steps moved and the gap did not; this sweep the gap moved and the steps did not hold. A reader who had taken either measurement as a standing property of this block has now been corrected on both, in opposite directions, within four days. 🔴 AND THE CLOSED GAP IS AN ARTIFACT OF DAILY GRANULARITY. IT DID NOT CLOSE TO ZERO; IT SHRANK TO ABOUT TWO HOURS AND IS STILL SIGNED THE SAME WAY. Stating the gap as zero would be a statement about the resolution of the clock rather than about the figures. The fixing is the 16:00 Central European Summer Time concertation, which on October 23, 2026 is 14:00 UTC and therefore about 10:00 a.m. Eastern. The barrel figures are a futures OPEN, and the same publisher's page prints an intraday reading at 8:22 a.m. Eastern, so the barrel leg is earlier in the day than the fixing by something on the order of an hour and a half to two hours. The rate leg is therefore STILL AHEAD, in the same direction it has been ahead throughout, and what changed is only that the lead is now too small for a date to show. Same calendar day is not same moment, and this block measures in days. 🔴 THE FROZEN-BARREL ISOLATION IS NOT MERELY UNAVAILABLE — THE TWO BARRELS MOVED IN OPPOSITE DIRECTIONS, SO THERE IS NO SINGLE BARREL DIRECTION ON THIS PAGE AT ALL. The entry below records the isolation discharged because the barrel moved. This compilation is a further remove from it: Brent rose from $109.13 to $113.12, up 3.656 per cent, while WTI fell from $101.44 to $99.36, down 2.050 per cent, over the identical interval. Every currency on this table carries both barrels, so each row now has one column moving up and one moving down underneath a single currency move. A reader cannot say 'the barrel rose' without naming which barrel, and the currency part of any column cannot be separated from the barrel part by inspection of the table. The widening Brent-less-WTI spread, $7.69 to $13.76, is the measurement doing the work, and it is a dollar-denominated spread that no currency in this list affects. 🔴 THE DOLLAR FIRMED AGAIN, A SECOND CONSECUTIVE FIRMING LEG, AND BY ALMOST EXACTLY HALF AS MUCH. The euro bought $1.1215 at the 20 October fixing and $1.1193 at the 23 October fixing, so the DOLLAR FIRMED 0.197 per cent, against 0.338 per cent on the leg the entry below measures. On the currency leg alone — holding the barrel in dollars, which this sweep may NOT do, and now may not do in either direction — the euro price of a dollar barrel rose 0.197 per cent. 🔴 THE NEAR-EVEN SIGN SPLIT RETURNS, WHICH SETTLES THE QUESTION THE ENTRY BELOW OPENED WITHOUT ANSWERING. Against the 20 October fixing the euro ROSE against thirteen of the twenty-eight non-euro currencies, FELL against fourteen, and was EXACTLY UNCHANGED against one. The entry below demoted a fourteen-fourteen-one split as 'one day's measurement' after the following day printed six-twenty-one-one. Both demotions were right and neither was the whole claim: the split is not a standing property of the table and it is not a one-off either, because it has now been near-even, lopsided, and near-even again on three consecutive compilations. It is a distribution, and the only honest description of a distribution is a range. Widest euro rises: the Mexican peso 1.207 per cent, the Philippine peso 0.609, the Hungarian forint 0.562, the South African rand 0.546. Widest falls: the Japanese yen 1.032 per cent, the Australian dollar 0.729, the Canadian dollar 0.590, the Norwegian krone 0.551. Smallest movers, stated to three decimals: the Icelandic krona 0.000, the Danish krone 0.003, the Czech koruna 0.008, the Swedish krona 0.072. 🔴 THE ICELANDIC KRONA HAS NOW PRINTED 136.20 AT THREE CONSECUTIVE FIXINGS, AND THAT IS EVIDENCE AGAINST THE COINCIDENCE READING THE ENTRY BELOW GIVES — INCLUDING THE READING IT GAVE OF THIS VERY ROW. The entry below calls the krona's exactly-repeated fixing 'a coincidence of two days' trading landing inside the same decimal'. Two days could be a coincidence. The 16 October, 20 October and 23 October fixings are all 136.20, which requires three consecutive sessions to have moved the rate by less than half of one hundredth on a figure near 136 — under about four thousandths of one per cent each time. The Institute will not convert that into a claim about a mechanism it has not measured: no arrangement holding the krona to the euro has been read in any publisher's release on any sweep, and the ERM II case beside it on the table is the Danish krone and is documented. What is stated is narrower and is the part that is ours: THE COINCIDENCE EXPLANATION IS NOW CARRYING MORE WEIGHT THAN THREE REPETITIONS CAN GIVE IT, and the next sweep should either find the krona moving or find out why it does not. 'Not explained by this reference' and 'unexplained' are different claims and only the first is ours to make. 🔴 THE SCALE WARNING, AND THIS COMPILATION ADDS A SECOND KIND OF ZERO TO THE TABLE. A reader taking the dollar row's 0.197 per cent as the size of the currency contribution to their own column would be wrong in SIGN on NINE of twenty-eight rows — the yen, sterling, the Chinese yuan, the South Korean won, the Canadian dollar, the Australian dollar, the Norwegian krone, the Hong Kong dollar and the New Zealand dollar — because what moves a local-currency barrel is not the dollar's move but the DIFFERENCE between that currency's move and the dollar's. On the currency leg alone the spread runs from the Japanese yen, where a barrel became 0.836 per cent CHEAPER in local terms, to the Mexican peso, where it became 1.403 per cent DEARER — about two and a quarter percentage points of separation from currency alone, against one and ninety-five hundredths on the leg below. And note the SINGAPORE DOLLAR, which moved 0.195 per cent against the euro where the dollar moved 0.196: on the currency leg alone its barrel changed by under one thousandth of one per cent. That is a near zero on the CURRENCY-ONLY leg and the krona's is an exact zero on the EURO leg, and the two rows measure different things while looking alike. Neither figure describes this page's columns, because both barrels moved underneath them and in opposite directions. 🔴 THE HOLES ARE TWO ON THE FIXING LEG THIS SWEEP AND THE THREE CATEGORIES HOLD. On the rate leg the 21 October and 22 October fixings were PUBLISHED AND NOT READ: both exist, this reference skipped both, and both holes are the Institute's. On the barrel leg, whether October 20, 21 or 22 editions were served at the publisher's path is NOT OBSERVED, and the Institute will not convert 'not in the release this reference read' into 'not published'. The 14 October fixing recorded further below remains published and unread from an earlier sweep. So the running total of published-and-unread fixings on this leg is three, and it is stated as a total rather than discovered again each sweep. 🔴 SUPERSEDED AND LABELLED, per the Two-Clock rule, rather than deleted — the entry below was written when this table carried the 20 October fixing against an October 19 barrel, and its findings are correct as of their own measurement dates and are not the figures this block now prints. It is retained because every per-cent figure above is a difference measured against it: 🔴 THE FROZEN-BARREL ISOLATION IS DISCHARGED, AND IT IS NAMED AS DISCHARGED RATHER THAN LEFT TO LAPSE. The entry retained below rests its whole currency reading on a barrel that did not move, which let it attribute the ENTIRE change in every non-dollar column to currency and nothing else. That condition is gone. On this sweep the dollar barrel moved and moved hard: WTI from $95.16 to $101.44, a rise of 6.599 per cent, and Brent from $105.99 to $109.13, a rise of 2.963 per cent. So every non-dollar column on this page now moves for two reasons at once again, and a reader cannot decompose any local-currency figure into its barrel part and its currency part by looking at the table. The isolation was a property of one compilation, not a property of this block, and the sentence below asserting it is superseded by this one. 🔴 THE DOLLAR FIRMED, REVERSING TWO CONSECUTIVE WEAKENING LEGS. The euro bought $1.1253 at the 16 October fixing and $1.1215 at the 20 October fixing, so the DOLLAR FIRMED 0.338 per cent. Every retained entry below this one that describes a weakening dollar describes the two legs this one reverses. On the currency leg alone — holding the barrel in dollars, which this sweep may NOT do, because the barrel moved — the euro price of a dollar barrel rose 0.339 per cent. 🔴 THE LEGS ARE STEP-MATCHED AT TWO, WHICH REVERSES THE PRE-REGISTERED FAILURE CONDITION THE ENTRY BELOW SCORED, WHILE THE SIGNED GAP IS UNCHANGED. The fixing leg advanced TWO TARGET business days, from the 16 October fixing through 19 October to the 20 October fixing. The barrel leg advanced two market sessions, from the October 15 open through October 16 to the October 19 open, across four calendar days. Two steps against two steps, where the entry below recorded one against zero. But the signed gap is the SAME as the entry below carries — the rate leg is ahead by one day, an October 19 barrel at a 20 October fixing — for a second consecutive compilation. Step-match and signed gap are two different measurements and this compilation moved one of them and not the other, which is the clearest available demonstration that the pre-registration below was right to name them separately. 🔴 THE HOLES ON THE TWO LEGS ARE OF DIFFERENT KINDS AND ONLY ONE OF THEM IS OURS. On the rate leg the 19 October fixing was PUBLISHED AND NOT READ: it exists, this reference skipped over it, and that hole is the Institute's. On the barrel leg the publisher's page was read on this sweep and serves an October 19 open; whether an October 16 edition was served at that path is NOT OBSERVED, and the Institute will not convert 'not in the release this reference read' into 'not published'. Three categories, and this compilation has one of each: published-and-unread on the fixing leg, which is ours; not-observed on the barrel leg, which is nobody's until somebody looks; and the 14 October fixing recorded below, still published and unread from an earlier sweep. 🔴 THE NEAR-EVEN SIGN SPLIT IS DISCHARGED — THE EURO FELL AGAINST TWENTY-ONE OF TWENTY-EIGHT. Against the 16 October fixing the euro ROSE against six of the twenty-eight non-euro currencies, FELL against twenty-one, and was EXACTLY UNCHANGED against one. The entry below reads fourteen, fourteen and one, and a reader who took that near-perfect split as a standing property of the table rather than as one day's measurement is corrected by this row. Widest euro rises: the Hungarian forint 0.441 per cent, the Mexican peso 0.410, the Thai baht 0.362. Widest falls: the Brazilian real 1.500 per cent, the Canadian dollar 1.111, the Norwegian krone 0.937. Smallest movers, stated to three decimals: the Icelandic krona 0.000, the Danish krone 0.003, the Romanian leu 0.004, the Czech koruna 0.025. 🔴 THE EXACT ZERO MOVED TO A DIFFERENT ROW, WHICH STRENGTHENS THE POINT THE ENTRY BELOW MAKES RATHER THAN REPEATING IT. Last compilation the genuine zero was the Polish zloty; this compilation the zloty moved 0.156 per cent and the ICELANDIC KRONA printed 136.20 at both fixings — again an identical published figure rather than a rounding artifact of a small move. The krona is NOT held to the euro by any arrangement, so an exactly repeated fixing on it is a coincidence of two days' trading landing inside the same decimal, exactly as the zloty's was. The Danish krone sitting beside it at 0.003 is the structural case — ERM II — and the two rows look alike on the page for entirely unrelated reasons. That the zero jumped from one floating currency to another in a single sweep is the evidence that no row on this table owns it. 🔴 THE SCALE WARNING, AND ON THIS COMPILATION IT IS WORSE THAN USUAL. A reader taking the dollar row's 0.338 per cent as the size of the currency contribution to their own column would be wrong in SIGN on TWELVE of twenty-eight rows — sterling, the Swiss franc, the Chinese yuan, the Indian rupee, the South Korean won, the Brazilian real, the Canadian dollar, the Indonesian rupiah, the Australian dollar, the Norwegian krone, the Singapore dollar and the New Zealand dollar — because what moves a local-currency barrel is not the dollar's move but the DIFFERENCE between that currency's move and the dollar's. On the currency leg alone the spread runs from the Brazilian real, where a barrel became 1.166 per cent CHEAPER in local terms, to the Hungarian forint, where it became 0.781 per cent DEARER — about one and ninety-five hundredths percentage points of separation from currency alone. Those two figures are currency-only and the page's own columns are not, because the barrel moved underneath them. 🔴 SUPERSEDED AND LABELLED, per the Two-Clock rule, rather than deleted — the entry below was written when this table carried the 16 October fixing against an October 15 barrel, and its finding, its refusal to hold a published fixing back, and its own pre-registration all stand as written: 🔴 THE PRE-REGISTERED FAILURE CONDITION FIRED, AND IT IS STATED RATHER THAN ABSORBED. The entry retained below closes by naming what to watch for — legs advancing DIFFERENT numbers of steps, or skipping different dates — and this compilation is the first instance of it on this block's record. The fixing leg advanced ONE step, from the 15 October fixing to the 16 October fixing: two consecutive TARGET business days with nothing between them, so there is no unread interior fixing on this leg at all. The barrel leg advanced ZERO steps. The price publisher was read on this same sweep and still serves the October 15 opens, byte for byte what this reference already carried. So the table is NOT step-matched, the date gap is ONE DAY, and the rate leg is the one that is ahead. 🔴 THE TABLE WAS ADVANCED ANYWAY, AND THE REASONING IS RECORDED BECAUSE THE ALTERNATIVE WAS AVAILABLE AND WAS REFUSED. Holding the 16 October fixing back to sit level with another publisher's October 15 barrel would have subordinated one measurement clock to another, which is the precise opposite of what this block exists to do: each leg carries the date its own publisher measured it on, and where those dates differ the difference is the finding. The retained history below also shows the rate leg running ahead four separate times with no intervention on any of them, so a one-day lead is a shape this record has carried before rather than a new condition requiring a new remedy. What is new is only that the shape is now scored against a pre-registration instead of noticed in passing. 🔴 AND THE FROZEN BARREL ISOLATES THE CURRENCY LEG, WHICH NO OTHER COMPILATION ON THIS RECORD HAS DONE. On every previous sweep the local-currency barrel figures moved for two reasons at once and the two could not be separated by a reader. Here the dollar barrel did not move, so the ENTIRE change in every non-dollar column is currency and nothing else. Measured: the euro bought $1.1208 at the 15 October fixing and $1.1253 at the 16 October, so the DOLLAR WEAKENED 0.401 per cent — a second consecutive weakening leg, where every entry retained below this one describes a firming dollar. With the barrel held in dollars, the euro price of a barrel therefore fell 0.400 per cent on currency alone. The spread across the table runs from the Brazilian real, where the barrel became 0.510 per cent DEARER in local terms, to the New Zealand dollar, where it became 1.118 per cent CHEAPER — about one and six-tenths percentage points of separation generated by a single day of currency movement against a barrel that did not move at all. The dollar column is the only one on the page that is unchanged this sweep, and it is unchanged by construction rather than by observation. 🔴 THE TABLE SPLIT ALMOST EXACTLY IN HALF BY SIGN, AND ONE ROW PRINTED A GENUINE ZERO. Against the 15 October fixing the euro ROSE against fourteen of the twenty-eight non-euro currencies, FELL against fourteen, and was EXACTLY UNCHANGED against one. Widest rises: the Brazilian real 0.914 per cent, the Indonesian rupiah 0.885, the Turkish lira 0.624. Widest falls: the New Zealand dollar 0.720 per cent, the South Korean won 0.628, the Swiss franc 0.355. Smallest movers, stated to three decimals because two of them are below a hundredth of a per cent: the Polish zloty 0.000, the Danish krone 0.003, the Thai baht 0.027, the Romanian leu 0.051. The zloty printed 4.3700 at both fixings — the identical published figure, not a rounding artifact of a small move — and that is worth distinguishing from the krone beside it. The Danish krone's near-zero is STRUCTURAL: it is held to the euro inside ERM II and a reader should expect it to print a figure like this on most days. The zloty FLOATS, and an exactly repeated fixing on a floating currency is a coincidence of two days' trading landing inside the fourth decimal place. A reader who infers a peg from the zloty row would be inferring it from one day. 🔴 AND THE SCALE WARNING THIS BLOCK CARRIES ON EVERY COMPILATION APPLIES WITH UNUSUAL FORCE HERE. A reader taking the dollar row's 0.401 per cent as the size of the currency contribution to their own column would be wrong by more than twice on the widest row and wrong in SIGN on fourteen of twenty-eight, because what moves a local-currency barrel is not the dollar's move but the DIFFERENCE between that currency's move and the dollar's. The zloty is the cleanest demonstration available: it did not move against the euro by any amount at all, and the barrel in zloty still fell 0.400 per cent, entirely because the dollar did. A row that did not move is not a row whose barrel did not move. 🔴 THE HOLES ON EACH LEG ARE OF DIFFERENT KINDS AND ARE NAMED SEPARATELY. On the rate leg there is no hole: the step is consecutive, and because October 17 is a Saturday there is no TARGET business day between the carried fixing and the date of this reading, so the 16 October fixing is the newest that can exist. The next is due Monday 19 October. On the barrel leg the publisher's own page was read and still carries October 15; whether an October 16 edition exists anywhere else is NOT OBSERVED, and the Institute will not convert 'not in the release this reference read' into 'not published'. The 14 October fixing remains published and unread from the preceding sweep, and that absence stays recorded below as ours. 🔴 SUPERSEDED AND LABELLED, per the Two-Clock rule, rather than deleted — the entry below was written when this table carried the 15 October fixing against an October 15 barrel, and its finding, its withdrawal of the weekend mechanism, and the pre-registration this sweep has now scored all stand as written: 🔴 STEP-MATCHED AT TWO FOR THE THIRD TIME, BOTH LEGS SKIPPING OCTOBER 14 — AND THE MECHANISM THE ENTRY BELOW OFFERED FOR THE RECURRENCE IS REFUTED BY THIS INSTANCE, NOT MERELY UNCONFIRMED BY IT. The barrel went from an October 13 open to an October 15 open: two trading sessions, October 14 and October 15, across two calendar days. The fixing went from the 13 October fixing to the 15 October fixing: two TARGET business days, with the 14 October fixing SKIPPED AND NEVER READ. One interior observation is missing on each leg, it is OCTOBER 14 on both, and nothing else is absent — so the compilation is step-matched at two and the date gap is zero, which is the condition this block holds to be readable. 🔴 BUT THE WEEKEND EXPLANATION DOES NOT SURVIVE, AND IT IS WITHDRAWN RATHER THAN SOFTENED. The entry retained below states that the shape recurs because each interval runs FRIDAY TO TUESDAY, and that a Friday-to-Tuesday sweep 'puts a weekend inside the gap, which removes two calendar days without removing any observation, and leaves exactly one unread Monday on each leg.' It then pre-registered the prediction: 'the next Friday-to-Tuesday sweep should be expected to reproduce it.' This interval runs TUESDAY TO THURSDAY. There is no weekend in it, the calendar gap is two days rather than four, the unread day is a WEDNESDAY, and the shape is reproduced exactly. So the weekend was never the mechanism. The headline claim — that the shape is produced by the sweep cadence rather than by luck — SURVIVES and is strengthened, because it now holds across two different weekday arrangements; what fails is the explanation offered for it. The actual mechanism is simply that the sweep interval is two steps wide on both legs, and the weekday composition of those steps is irrelevant. This is the second time this block has mistaken the particulars of the sweeps that happened to occur for the principle behind them — the first was calling 'one step, nothing skipped' the clean case rather than step-matching — and both corrections run the same way: a sufficient condition was promoted to a necessary one. 🔴 AND THE TWO ABSENCES ARE NOT ABSENT FOR THE SAME REASON — NOR FOR EITHER OF THE TWO REASONS THIS BLOCK HAS NAMED BEFORE. On the FIXING leg October 14 is PUBLISHED AND UNREAD, as it was on the previous sweep: October 14, 2026 was a Wednesday and a TARGET business day, so a 14 October fixing exists, and this sweep did not go to the historical series for the one behind the latest. That gap is ours. On the BARREL leg October 14 is NEITHER ours nor the publisher's, because WHETHER IT EXISTS IS NOT OBSERVED. The daily quote page serves one edition at one URL and overwrites it, so the question of whether an October 14 edition was ever posted cannot be answered from two readings taken on October 13 and October 15, and the price ledger of this same sweep records a refusal to answer it. The exchange leg is no help: no bulletin for trade date October 14 is posted at the path this reference reads. So this block now recognises a THIRD category of hole — published and unread, not yet published, and ownership unassignable — and the entry below, which says 'one gap is ours and one is the publisher's', is complete only for the interval it describes. 🔴 THE DOLLAR LEG RAN WITH ONE BENCHMARK AND AGAINST THE OTHER, WHICH NO PARAGRAPH IN THIS BLOCK HAS HAD TO DESCRIBE BEFORE. The euro bought $1.1188 at the 13 October fixing and $1.1208 at the 15 October, so the DOLLAR WEAKENED 0.179 per cent — the first weakening leg this block has carried in the run; every retained entry below describes a firming dollar. Over the same interval WTI rose 1.917 per cent in dollars and Brent FELL 0.170 per cent, a sign disagreement the price ledger confirms off the exchange's own change column rather than off arithmetic. A weakening dollar damps a dollar-price rise and amplifies a dollar-price fall when the barrel is restated in a currency that gained, so the currency leg pushes the two columns in OPPOSITE directions: restated at the fixing, WTI rises 1.735 per cent against 1.917 in dollars, and Brent falls 0.348 per cent against 0.170. The euro barrel moved less than the dollar barrel on one benchmark and twice as much on the other. Every retained paragraph below assumes both benchmarks move the same way and is silent on this case. 🔴 AND THE DOLLAR COLUMN IS NOT A PROXY FOR THE CURRENCY LEG — THE TABLE SPLIT IN SIGN. The euro ROSE against twenty of the twenty-nine and FELL against the other nine on the same fixing, so there is no single 'currency move' to read off the dollar row. The widest rise is the Mexican peso at 1.204 per cent, then the Australian dollar at 1.154 and the Israeli shekel at 1.061; the widest fall is the Hungarian forint at 0.195, with the Norwegian krone and the Polish zloty tied at 0.155. A reader taking the dollar row's 0.179 per cent as the scale of the currency contribution would be wrong by about seven times on the peso column and wrong in SIGN on nine columns. Note also that the INDIAN RUPEE, which the entry below singles out as the table's outlier at 1.66 per cent, is this fixing's fourth smallest mover at 0.056 per cent — behind the euro-pegged Danish krone at 0.004, the Thai baht at 0.032 and the Romanian leu at 0.049 — and it moved in the opposite direction. An outlier observed once is a property of that fixing and not of the currency, and this block should not have named it without saying so. Every per-cent figure in this paragraph is computed from the two FX_RATES compilations and is stated to three decimals because several of them are smaller than a tenth of a per cent, where one decimal would round a real move to zero. 🔴 SUPERSEDED AND LABELLED, per the Two-Clock rule, rather than re-dated — the entry below was measured on the October 9-to-13 barrel against the 9-to-13 fixing, and its weekend mechanism is withdrawn above while its step-matching finding stands: 🔴 BOTH LEGS ADVANCED TWO STEPS, BOTH SKIPPED THE SAME DATE, AND THE DATE GAP IS ZERO — SO THIS COMPILATION IS STEP-MATCHED AT TWO, AND IT IS THE SECOND INSTANCE OF A SHAPE THIS BLOCK HAS NOW SEEN TWICE. The barrel went from an October 9 open to an October 13 open: two trading sessions, October 12 and October 13, across four calendar days. The fixing went from the 9 October fixing to the 13 October fixing: two TARGET business days, with the 12 October fixing SKIPPED AND NEVER READ. One interior observation is missing on each leg, it is OCTOBER 12 on both, and nothing else is absent. That is the cleanest condition this block recognises short of a single step, and a reader differencing any currency column against the figure it carried last sweep is comparing a two-session price move against a two-fixing rate move with the same date unobserved on both sides. 🔴 THE SHAPE RECURS BECAUSE OF THE SWEEP INTERVAL, NOT BECAUSE OF LUCK, AND SAYING SO IS WORTH MORE THAN THE FIGURES. The October 2-to-6 entry retained below has the identical structure — two steps, one skipped interior date, the same date on both legs — and the two intervals have one thing in common: each runs FRIDAY TO TUESDAY. A Friday-to-Tuesday sweep puts a weekend inside the gap, which removes two calendar days without removing any observation, and leaves exactly one unread Monday on each leg. So the property is produced by the cadence rather than discovered in the data, and the next Friday-to-Tuesday sweep should be expected to reproduce it. The condition worth watching is not this shape but its failure: legs advancing DIFFERENT numbers of steps, or skipping different dates. 🔴 AND THE TWO ABSENCES ARE NOT ABSENT FOR THE SAME REASON, WHICH THE PARAGRAPHS ABOVE WOULD OTHERWISE FLATTEN. October 12, 2026 was Columbus Day, a United States federal bank holiday — and it was NOT a TARGET closing day and NOT a CME energy holiday, so both a 12 October fixing and a 12 October trading session exist. On the FIXING leg that observation is PUBLISHED AND UNREAD: the ECB's daily page shows only the latest fixing, and this sweep did not go to the historical series for the one behind it. On the BARREL leg it is NOT YET POSTED AT THE PATH THIS REFERENCE READS: on October 13 the exchange's /current/ Section 61 bulletin still served #195 for trade date Friday October 9, so no October 12 settle was available to read at all. One gap is ours and one is the publisher's, and they are not the same kind of hole. 🔴 THE DOLLAR LEG RAN AGAINST THE BARREL AGAIN, BY LESS THAN A FIFTH OF A PER CENT. The euro bought $1.1206 at the 9 October fixing and $1.1188 at the 13 October, so the dollar firmed 0.161 per cent while the barrel rose about 2.4 per cent on both benchmarks. Every column here therefore carries two movements at once, and the non-dollar columns move by more than the dollar column does by construction. Note one outlier read rather than computed: the Indian rupee went 108.3965 to 106.6000 per euro, a 1.66 per cent move that is ten times the dollar's and the largest in the table. It is printed as the publisher prints it and is not reconciled against anything here, because this page takes one publisher's set whole or not at all. 🔴 SUPERSEDED AND LABELLED, per the Two-Clock rule, rather than re-dated — the entry below was measured on the October 6-to-9 barrel against the 6-to-9 fixing, a three-step interval on both legs with two interior observations missing on each and the absences coinciding on one date only: 🔴 BOTH LEGS ADVANCED THREE STEPS AND LANDED ON THE SAME DATE, SO THE DATE GAP IS ZERO AND THE COMPARISON IS STEP-MATCHED AT THREE — BUT TWO INTERIOR OBSERVATIONS ARE MISSING ON EACH SIDE AND THEY ARE NOT THE SAME DATES. The barrel went from an October 6 open to an October 9 open: three trading sessions, October 7, 8 and 9, across three calendar days. The fixing went from the 6 October fixing to the 9 October fixing: three TARGET business days, with the 07 October and 08 October fixings SKIPPED AND NEVER READ. On the barrel leg the absent marks are the October 6 and October 8 settles, because the bulletins carrying them were never read at all. So both legs are three steps wide with two interior observations missing, the absences coincide on October 8 and nowhere else, and a reader comparing any currency column against the figure it carried last sweep is comparing a three-session price move against a three-fixing rate move with neither path observable in between. The dollar leg ran against the barrel: the euro bought $1.1269 at the 6 October fixing and $1.1206 at the 9 October, so the dollar strengthened while the barrel rose, and every column here carries both movements at once. 🔴 SUPERSEDED AND LABELLED, per the Two-Clock rule, rather than re-dated — the entry below was measured on the October 2-to-6 barrel against the 2-to-6 fixing, a two-step interval on both legs with one skipped date shared between them: 🔴 BOTH LEGS ADVANCED TWO STEPS AND BOTH SKIPPED THE SAME DATE, SO THE DATE GAP IS ZERO AND THE COMPARISON IS STEP-MATCHED — AND THAT CORRECTION IS WORTH MORE THAN THE FIGURES. The barrel went from an October 2 open to an October 6 open: two trading sessions, October 5 and October 6, across four calendar days. The fixing went from the 2 October fixing to the 6 October fixing: two TARGET business days, with the 05 October fixing SKIPPED AND NEVER READ. So a reader comparing any currency column against the figure it carried last sweep is comparing a two-session price move against a two-fixing rate move, and the intermediate observation is missing on BOTH sides, on the same date. SUPERSEDED AND LABELLED, per the Two-Clock rule, rather than re-dated: the previous compilation stood here as 'BOTH LEGS ADVANCED ONE STEP, NOTHING WAS SKIPPED, AND THE DATE GAP IS ZERO — WHICH IS THE CLEANEST COMPILATION IN THE RUN', and went on to say that the asymmetry this block had disclosed on every prior sweep was absent and that its absence was being recorded with the same prominence the asymmetry had been. Every figure in it was correctly measured for the step it described. The CLAIM does not survive, and it does not survive by being re-dated, because the condition it named was the wrong condition. 🔴 WHAT MAKES A COMPILATION READABLE IS STEP-MATCHING, NOT ONE STEP WITH NOTHING SKIPPED. 'One step, nothing skipped' was a SPECIAL CASE of the property that actually matters, and this compilation is the instance that separates them: two steps on each leg, one intermediate observation skipped on each leg, the same calendar date skipped on both — and a reader differencing the columns is in exactly the position the previous note described as uniquely clean. What would make a column unreadable is the legs advancing DIFFERENT numbers of steps, or skipping different dates, and neither happened here. This is the same shape of correction the dispersion block below had to make when it stopped calling dispersion a fifth case and recognised it as what the table does whenever both legs move: a taxonomy built out of the sweeps that happened to occur, mistaking the particulars of those sweeps for the principle. 🔴 AND ONE RESIDUAL ASYMMETRY INSIDE THE STEP MATCH, STATED RATHER THAN GLOSSED, BECAUSE IT IS NOT SYMMETRIC AND THE PARAGRAPH ABOVE WOULD READ AS THOUGH IT WERE. The barrel's skipped October 5 is PARTLY observed and the fixing's is WHOLLY unobserved. The price ledger read the October 5 SETTLEMENTS from CME Daily Bulletin #191 — BZ DEC26 at $100.32, CL NOV26 at $89.43 — so the dollar leg decomposes: Brent went 102.23 to 102.25 to 100.32 to 100.25, and essentially the whole of the $1.98 fall is the October 5 session. No equivalent decomposition is available on the rate leg, because the 05 October fixing was never read and the ECB's daily sheet carries one date at a time. A reader can therefore say WHEN the dollar barrel moved and cannot say when the currencies did. 🔴 A DEFECT IN THIS NOTE'S OWN PREVIOUS GENERATION, FOUND ON THIS SWEEP AND CORRECTED RATHER THAN QUIETLY OVERWRITTEN. Two paragraphs of this note disagreed with each other. One read 'THE RATE LEG IS AHEAD FOR A SECOND SWEEP RUNNING' and described the table as converting a September 30 barrel at an October 1 rate — a signed gap of one day — while a later paragraph in the same note read 'ON THIS COMPILATION THE DATE GAP IS ZERO' and stated correctly that both legs carried October 1. The second was right and the first was a stale generation left in place without a superseded label, which is the one thing the Two-Clock rule forbids: retaining superseded text is required, and retaining it UNLABELLED turns provenance into contradiction. The same stale generation also carried a skipped-fixing paragraph — 'the fixing carried here is 1 October; the fixing carried last sweep was 29 September' — which was accurate for the sweep that wrote it and describes an interval two steps behind this one. Both are superseded here and labelled, and the mechanism that produced the fault is worth more than the fault: a note edited by appending a new finding at the top, without re-reading what sits below it, accumulates paragraphs that each were true once and now disagree. Superseded and labelled rather than deleted, per the Two-Clock rule: this note previously read that the gap had reopened with the rate leg ahead and that the skipped fixing was new; before that, that the clocks agreed and that both legs had moved to get there; before that, that the gap had closed because one leg moved and the other stood still; before that, that the rate ran one day AHEAD of the barrel — a September 23 barrel at a September 24 fixing; before that, that the two agreed; and before that, that the barrel ran ahead. Every version was true when it was written. 🔴 ON THE COUNT ITSELF, AND THE CORRECTION MADE LAST SWEEP IS RE-DERIVED RATHER THAN CARRIED: this note used for several compilations to say 'four distinct alignments', which conflated two different things. The signed date gap has taken exactly THREE distinct values across the run — rate ahead by one day, barrel ahead by one day, and zero. The count of four came from also counting the MECHANISM by which a zero was reached (one leg moving versus both legs moving) as an alignment, which it is not. The run now stands at TEN compilations, still three distinct signed gaps — rate ahead by one, barrel ahead by one, and zero — and this is a further occurrence of the zero value, reached by a route the run has not taken before: both legs advancing TWO steps rather than one leg moving or both moving once. Per the correction two paragraphs above that is not a fourth signed gap and must not be counted as one — it is a mechanism, and conflating the two is the exact error that produced the 'four distinct alignments' figure this note carried for several compilations before withdrawing it. No intervention on any of the ten. The only durable statement is the one below: two publishers, two calendars, a small signed gap that is sometimes zero, labelled every time. Neither publisher is late. They keep different calendars, and which one is ahead on any given morning — or whether either is — depends on the day of the week and on whether a fixing has yet been struck. The Institute carries the newest figure each publisher has issued and labels the gap whenever there is one, rather than holding a superseded number on either side to make the table look tidy. 🔴 WHAT THE ONE-DAY GAP COSTS A READER IS BOUNDED HERE RATHER THAN STATED, AND THE PREVIOUS WORDING OF THIS PARAGRAPH IS CORRECTED RATHER THAN CARRIED. It said that between the two fixings the dollar firmed 0.42 per cent 'so every non-dollar column sits roughly that much high relative to a table struck wholly on the barrel's own date'. That inference does not follow from what was measured. The move between the two fixings READ spans two business days; the gap the reader carries is ONE day — the 30 September fixing against the 1 October fixing — and the 30 September fixing was skipped and never observed. Measured on what was read: the dollar firmed from 1.1403 to 1.1355 and then from 1.1355 to 1.1298 across those compilations, about 0.92 per cent in total. That figure is an UPPER BOUND on the gap's cost, not a measurement of it, and it would equal the cost only if the skipped day happened to be flat. The correction is to the description, not to the arithmetic, which was right about the two fixings it compared. The paragraph is retained because the gap will return — it has taken three distinct values across ten compilations and zero is one of three — and because the reasoning in it is about how to bound a gap's cost rather than about any particular day. The cost of a gap remains a function of what the currency did inside it, not of the gap's size in days — a one-day gap across a quiet fixing costs less than a zero-day gap would if the intraday move were violent. And a separate mismatch sits inside the day regardless, since a 16:00 Central European Time fixing and a morning New York open are hours apart even when they fall on the same date. 🔴 ON THIS COMPILATION THE DATE GAP IS ZERO AND THE INTRADAY MISMATCH IS NOT QUITE THE WHOLE OF WHAT REMAINS. The barrel quotes and the fixing both carry October 6, 2026. That is the clean case the paragraph above describes as better than a one-day gap, and it arrived because BOTH legs advanced — two steps each this time — rather than because one leg sat still while the other caught up, a distinction this note has had to make before, since a zero reached by two moving legs and a zero reached by a frozen one are the same number and not the same condition. SUPERSEDED AND LABELLED rather than re-dated: this paragraph previously read that both legs carried October 2, 2026 and that both had advanced ONE step. The reader still carries hours — the fixing is struck at 16:00 Central European Time and the quote is a New York morning open — and on a one-step compilation that was the whole of the remaining mismatch. On this one it is not: the unread 05 October fixing sits inside the step as well, so a reader carries the hours AND an unobserved intermediate on the rate leg. Neither is a defect in the table and both are stated, because a mismatch described as the only one when it is not is worse than a larger mismatch described accurately. 🔴 THE DOLLAR WEAKENED FOR THE FIRST TIME IN FOUR READINGS, WHICH REVERSES THE DIRECTION THIS BLOCK HAS RECORDED ALL RUN — AND NOTHING HAS TO BE WITHDRAWN, WHICH IS THE POINT. The dollar went from 1.1225 to 1.1269 per euro: it weakened 0.3905 per cent, measured as the change in euros per dollar so that the sign is the dollar's own. The three readings before it ran 1.1403, 1.1355, 1.1298, 1.1225 — the dollar firming on each of three consecutive steps, 1.5610 per cent in total. This block never called that a trend. It recorded each step and, when a repeat first occurred, said in terms that two consecutive observations of the same direction are not a regime — the SPR row's reasoning applied before the fact rather than after it. Because the firming was reported as a sequence of measurements and not as a tendency, the reversal is simply the next measurement and there is no retraction to make. Net across the whole run the euro is still 1.1751 per cent cheaper in dollars than at the first reading, so one reversal does not undo three advances, and that figure is stated so a reader does not read the reversal as a round trip. 🔴 AND BOTH LEGS MOVED, SO THE DISPERSION BELOW IS A MEASUREMENT RATHER THAN AN ARTIFACT. SUPERSEDED AND LABELLED, and RE-DERIVED rather than re-dated per the ninth rule in CLAUDE.md, because every figure in it is a stated consequence: the previous compilation reported the dollar firming from 1.1298 to 1.1225 per euro at 0.6461 per cent, the dollar barrel UP 4.4335 per cent, the twenty-nine columns spread across 3.0690 percentage points from the Swiss franc at +3.3529 per cent to the Romanian leu at +6.4219 per cent, and EIGHTEEN columns rising by more than the dollar column against TEN by less. Every one of those was correctly measured for the step it described and not one of them describes this step; the signs alone have all flipped. Measured here: the dollar barrel is DOWN 1.9368 per cent, Brent $102.23 to $100.25, and the twenty-nine columns spread across 5.2068 percentage points, from the Swiss franc at −1.4775 per cent to the Brazilian real at −6.6843 per cent. NINETEEN columns fell by more than the dollar column and EIGHT by less, so for most of the world the barrel got cheaper by less than the dollar price alone reports — and one column, the Chinese yuan, coincides with the dollar column to four decimal places, which is treated below as an audit rather than as a coincidence. 🔴 THAT 5.2068-POINT SPREAD IS THE WIDEST THIS BLOCK HAS RECORDED AND IT IS ALMOST WHOLLY ONE COLUMN, SO IT IS REPORTED BOTH WAYS. The prior readings were 3.0690 points and, before that, 1.90 points. Strike the Brazilian real and the remaining twenty-eight columns span 2.5458 points, from the Swiss franc at −1.4775 to the Mexican peso at −4.0233 — narrower than the previous compilation's. So the headline figure is a single-column artifact and a reader told only the headline would conclude the table had become twice as dispersed when on twenty-eight of twenty-nine columns it became less so. Where one input dominates a summary statistic, print the statistic with and without it; the pair is the measurement and neither number alone is. 🔴 A TRANSCRIPTION CHECK WORTH PRINTING, BECAUSE A PEGGED CURRENCY IS A FREE AUDIT OF THE WHOLE TABLE. The Danish krone is held in a narrow band against the euro, so its cross against the dollar must reproduce the euro's own cross against the dollar to within the band's drift or the transcription is wrong somewhere. Re-derived on this fixing rather than carried: the krone gained 0.3759 per cent against the dollar and the euro gained 0.3905 per cent, differing by 0.0146 of a percentage point — against the 0.0147 per cent the krone weakened against the euro inside the band, 7.4736 to 7.4747. The residual is a ten-thousandth of a point and is the rounding in the published rates, not a defect. The table reconciles. Superseded and labelled: the previous compilation's figures were 0.6207, 0.6503, 0.0296 and 0.0294, all correctly measured for that fixing. This is cheaper than re-reading twenty-nine numbers and it tests the arithmetic as well as the keystrokes. 🔴 AND THIS FIXING SUPPLIES TWO FURTHER FREE AUDITS, ON THE DOLLAR SIDE RATHER THAN THE EURO SIDE, WHICH IS THE STRONGER DIRECTION BECAUSE THE BARREL IS QUOTED IN DOLLARS. The Chinese yuan is managed against the dollar, and its dollar cross is 7.5259 ÷ 1.1225 = 6.704588 on the 2 October fixing and 7.5554 ÷ 1.1269 = 6.704588 on the 6 October fixing — identical to seven significant figures, a drift of about two millionths of one per cent. That is why the yuan column coincides with the dollar column at −1.9368 per cent and the coincidence is arithmetic rather than a finding about China. The Hong Kong dollar is held inside a 7.75–7.85 convertibility band and crosses at 7.84713 then 7.84791, both inside it, weakening 0.0098 per cent — which is the 0.0097 of a point by which its column sits above the dollar column. Three pegs, two conventions, and all three reconcile. The reason to print them is the first rule in CLAUDE.md: a table that agrees with itself and a table nobody checked emit the identical silence, and a managed cross is the cheapest discriminator between the two that this page has available. 🔴 AND ONE CROSS IS LARGE ENOUGH TO FLAG RATHER THAN TO CORRECT — A DIFFERENT ONE FROM LAST COMPILATION'S, WHICH IS ITSELF THE FINDING. The Brazilian real strengthened 4.4685 per cent against the euro across these two fixings, 5.8610 to 5.5991, and 4.8416 per cent against the dollar. That is far the largest move in the set — the second largest is the Mexican peso at 1.7439 per cent, then the Hungarian forint at 1.1458 and the South African rand at 1.0702 — and it is the whole of the reason the barrel column spread doubled. It was read directly off the ECB's published table for 06 October 2026 and is reported as the publisher gave it; the Institute does not smooth a figure because it looks big. 🔴 AND HERE IS WHAT THE INSTITUTE CANNOT CHECK, SAID PLAINLY RATHER THAN LEFT TO BE ASSUMED. A move that large over two fixings is also the shape a transcription error makes, and the two cannot be distinguished from this sweep's reading, because the error could sit on EITHER END and only one end is re-readable. The ECB's daily sheet, eurofxref, carries ONE DATE AT A TIME: the 5.5991 was read from the 06 October sheet on this sweep and the 5.8610 was read from the 02 October sheet on the last one, and that sheet no longer exists at that address. So the new figure is verified against the publisher and the old one is verified only against the Institute's own record of having read it. The remedy is the ECB's historical series, which is a different fetch on a different path and was NOT made on this sweep; saying that is cheaper than implying a check that did not happen. A reader for whom the real column matters should take both ends to that series rather than to this page. 🔴 SUPERSEDED AND LABELLED: this paragraph last carried the Romanian leu, at 1.9039 per cent against the dollar across a single fixing, 5.2830 to 5.3488 per euro, described as the largest move in the set and large for a managed currency. Correctly measured, and on this fixing the leu is one of the quietest columns in the table — 5.3488 to 5.3510, 0.0411 per cent against the euro. That is worth keeping rather than overwriting for a reason beyond provenance: a flag raised on one compilation and absent on the next is evidence about the currency only if the reader can see that the flag was looked for. The flag moves because the outlier moved, not because the test changed. 🔴 THE WTI CONTRACT BREAK IS NOW SIX STEPS BEHIND THIS COMPILATION, AND IT IS STATED RATHER THAN DROPPED. The warning this note carried six compilations ago was that the publisher's front-month WTI series had changed contract between the September 22 and September 23 sessions without saying so — the CME Daily Bulletin for September 22 settles the October 2026 crude contract at $94.59 and November at $90.52, October's open interest having collapsed to 1,337 lots in that session — so that roughly $4.07 of a $5.67 fall was the contract changing underneath the series rather than a price move, and every currency column inherited it intact. The October contract has since left the bulletin entirely: its open interest reached zero on Bulletin #184 for the September 24 session, and in every bulletin observed from #186 for September 28 onward — #187, #188, #190 and now #191 for October 5, FIVE bulletins and not five consecutive ones — the crude section opens at November and October does not appear at all. The WTI column is still a November contract on both ends of this step, so it may be read as a price move: $92.83 to $89.28, DOWN $3.55. 🔴 AND SO MAY THE BRENT COLUMN, FOR A SECOND CONSECUTIVE COMPILATION — BOTH ENDS OF THIS STEP ARE THE DECEMBER CONTRACT. $102.23 to $100.25, DOWN $1.98. So every one of the twenty-nine columns on both benchmarks is a price move with no contract change buried in it, which is now true twice running and is said explicitly because the compilations before those two each had to warn that one column or the other was not readable. A reader differencing the two benchmarks still cannot: WTI is November, Brent is December, and the crude rows on the hub page set out why two legs on two months are not a spread. 🔴 AND THIS IS THE FIRST COMPILATION WHERE THAT READABILITY RESTS ON A DIRECT READING RATHER THAN ON AN INFERENCE, WHICH IS A BETTER CLASS OF EVIDENCE AND IS WORTH SAYING SO. On every prior step this block concluded 'no contract change' from the contract's open interest and from the absence of a row — both of which are statements about what the bulletin does not contain. On this one the price ledger's roll test resolved each October 6 open against Bulletin #191's FINAL settlements directly: the WTI open of $89.28 sits 15 cents from CL NOV26's $89.43 and $1.08 from CL DEC26's $88.20, and the Brent open of $100.25 sits 7 cents from BZ DEC26's $100.32 and $2.67 from BZ JAN27's $97.58. Each open is nearest the month the series is supposed to be on, by a margin that identifies it. That is a positive identification of the contract rather than the absence of evidence for a roll, and the distinction matters here specifically because this block established last compilation that open interest is the LAGGING indicator and not the leading one. The honest qualifier: WTI's $1.08 alternative-month margin is the narrowest this test has recorded, so the identification is sound on this step and the test's own headroom is shrinking as the November and December contracts converge. 🔴 RETAINED AND LABELLED RATHER THAN DELETED, per the Two-Clock rule — the roll finding from the PREVIOUS compilation, which describes the September 30-to-October 1 step and not this one. Its arithmetic is unchanged and its subject is one step behind: THE BRENT ROLL THIS BLOCK WARNED OF ON THREE CONSECUTIVE COMPILATIONS HAS FIRED, AT THE MAGNITUDE NAMED IN ADVANCE, AND EVERY CURRENCY COLUMN HAS INHERITED IT. The warning read, in as many words, that a publisher rolling on open interest would print a fall of roughly five and a half dollars that is not a price move and that every currency column on this page would inherit it the way they inherited the WTI break. Measured: the October 1 Brent open of $97.89 sits 14 cents beneath Bulletin #188's DECEMBER settle of $98.03 and $5.64 beneath its NOVEMBER settle of $103.53, so the quote has moved from November to December. The November-to-December spread was $5.50, exactly the figure named. The printed fall is $102.58 to $97.89, down $4.69 — of which about $5.50 is the contract changing and about $0.81 is a price move UPWARD. In percentage terms the dollar Brent column prints −4.57 per cent on a session when the barrel rose, and with the currency leg at zero every other column prints the same −4.57 per cent. The inheritance is therefore complete and unmasked, which is the clearest instance of the mechanism this block exists to warn about. 🔴 WHAT THE WARNING GOT RIGHT AND WHAT IT DID NOT, RECORDED BECAUSE A PRE-REGISTRATION THAT IS ONLY SCORED WHEN IT WINS IS NOT A PRE-REGISTRATION. It got the direction right and the magnitude right to the cent. It got the timing wrong four times over: the warning stood unfired across four consecutive compilations, and this block said each time that it was an inference from open interest and not a reading of a contract calendar. And the magnitude was right only because it was re-measured every sweep, $7.45 then $6.43 then $5.50 — the figure named when the warning was first written would have been wrong by $1.95. Open interest says a roll is coming. It does not say when, and it does not say how far until the day before. 🔴 THE EVIDENCE FOR A WTI ROLL IS NOW AVAILABLE, AND IT IS REPORTED AS A STANDING CONDITION RATHER THAN AS A FORECAST — BECAUSE THIS SWEEP ALSO ESTABLISHED THAT THE INDICATOR LAGS. WTI is quoting November while Brent quotes December, so a WTI roll is the one in prospect. RETAINED AND LABELLED rather than re-dated, because its subject is the October 2 session and this compilation steps past it: Bulletin #190 for October 2 supplied the figures an earlier compilation could not obtain — CL NOV26 open interest 258,420 lots, DOWN 18,074 in one session, against CL DEC26 at 236,448, UP 524, the two months 21,972 lots apart with the gap closing by most of that in a single day. No equivalent reading is claimed for October 5: the open-interest levels for CL NOV26 and CL DEC26 in Bulletin #191 were not transcribed on this sweep, and a figure not read is not reported as unchanged. 🔴 THE STANDING CONDITION IS SCORED HERE RATHER THAN LEFT OPEN, BECAUSE A CONDITION ONLY SCORED WHEN IT FIRES IS NOT A CONDITION. Two trading sessions have passed since that reading — October 5 and October 6 — and the WTI roll HAS NOT FIRED. The roll test above identifies the October 6 open as CL NOV26 by a 15-cent margin against CL DEC26's $1.08. So an open interest gap of 21,972 lots closing at 18,598 lots a session, which on the arithmetic alone would have exhausted itself inside two sessions, did not produce a roll inside two sessions. That is the second independent demonstration in this block's record that open interest does not time a roll: the first was the Brent drain arriving the session AFTER the roll, and this is the converse case — an exhausting gap arriving before no roll at all. Both point the same way and neither was predicted. No date is attached to that, and the reason is sharper than the usual caution. The crude rows on the hub page record a correction made on this same sweep: BZ NOV26's open interest went from 19,421 lots to zero on October 2, the session AFTER the roll the publisher made on October 1. The drain FOLLOWED the roll. Three compilations of warnings had rested on open interest as the leading indicator and on the evidence it was the lagging one, so a declining November open interest is confirmation that a roll is available to the publisher and is not a countdown to it. A reader who finds no warning in this block on some future compilation should not read the silence as evidence either way, and a reader who finds this one should not read it as a date.

The one statement on this page that does not depend on a currency

Every column of barrel prices above is the same two facts multiplied together, which means almost nothing in the table survives a change of currency. One thing does. Brent at $113.12 against WTI at $99.36 is a ratio of 1.1385, and that ratio is identical in every row, because the exchange rate multiplies both benchmarks equally. A spread between two crudes is a fact about crude. A price level is a fact about crude and about money.

That is the test to apply to any cross-border oil claim. If the statement survives being restated in yen — a ratio, a spread, a share, a physical volume — it is about oil. If it does not, part of it is about the currency, and the reader is owed which part. The forward-curve ledger makes the same point on a different pair of axes: the Brent–WTI spread is a fact about place and grade, not about time.

For scale on how large the currency term can be, a barrel of Brent costs 17,738 yen and 10,740 rupees on this fixing. Neither figure can be compared to the other, or to the dollar price. All three are the same barrel.

Two consequential oil currencies are missing from this page

The Nigerian naira and the Russian rouble are absent from the ECB reference set and are therefore absent here. Both are consequential oil currencies and both would have to come from a different publisher on a different convention and a different clock, which is exactly the mixing this page exists to avoid. A reader needing either should take it from a single named source and date it themselves.

The rest of the reference