Corrections
The Backroom Brief prints its corrections in a later issue and never edits an issue once it is published. This page collects every correction the brief has printed, newest first, each quoted exactly as it appeared.
-
-
The August sale used as the comparison for the twenty-year reopening was labelled an original issue; the official record shows a $16 billion reopening on 19 August that stopped at 5.204 percent with a cover of 2.53.
The official auction record for that security shows the prior sale was on 19 August and it was a $16 billion reopening, stopping at 5.204 percent with a cover of 2.53. The comparison held up, the stop yield did rise 21.6 basis points month over month, and the label on it was wrong.
-
-
-
The week's Treasury move was framed as front-led, but from 11 to 15 September the two-year and ten-year yields each rose 4.0 basis points, and the five percent ten-year the issue led with was back at 4.977 the next morning.
Yesterday's issue framed the week as front-led, writing that most of the move came from the front of the curve. Measured from 11 to 15 September the two-year rose 4.0 basis points and the ten-year rose 4.0, the same.
-
-
-
The ten-year real yield's nine basis point rise to 2.55 percent was placed on Friday; it happened on Thursday 10 September, the real yield rose five points to 2.60 on Friday, and the 2.36 percent breakeven figure stands.
The lede said the ten-year breakeven "fell four basis points on Friday to 2.36 percent while the ten-year real yield rose nine to 2.55", citing FRED. Treasury's own real yield curve shows the nine-point rise to 2.55 happened on Thursday 10 September.
-
-
-
Brent's Thursday settle was printed at 108.19 dollars, up 6.90 percent and 10.49 percent over two sessions; it settled at 107.63, up 6.34 percent, and the two-session move was 9.92 percent, one error made by applying an intraday percentage to the previous day's settle.
This desk published a 108.19 dollar Thursday Brent settle on 10 September, with a 6.90 percent daily gain and a 10.49 percent two-session move. All three figures were wrong, and they were a single error rather than three.
-
A claim about the United Kingdom thirty-year gilt conflated an auction clearing yield with a secondary market level.
Two corrections against this desk's 10 September issue are printed in the sections above: the Brent Thursday settle, where the mechanism of the error has been identified precisely, and the United Kingdom thirty-year gilt claim, which conflated an auction clearing yield with a secondary market level.
-
-
-
The retraction of the whale distribution call rested on stale, mislabelled figures: the 450,000 counted Binance depositing addresses rather than ether, and the $423 million described whale flows from 6 and 7 May 2026, so the retraction was withdrawn without reinstating the original claim.
Wednesday's issue retracted the line that a 167,855 ether whale distribution was "complete", on the basis of reports that 450,000 ether landed on Binance on Sunday and $423 million more on Monday. Both figures are stale and mislabelled.
-
The claim that the war board itself was under an insider investigation was stated too strongly; the real inquiries named no Iran-conflict contracts, venue or individual, so the asterisk on every quoted probability was withdrawn.
Wednesday's issue asserted that "the war board itself is under an insider investigation" and hung an asterisk on every probability it quoted. That was stated too strongly.
-
-
-
A 167,855 ETH whale distribution was called complete; reporting of 450,000 ETH landing on Binance on Sunday and $423 million more on Monday made complete premature, so the line was retracted.
Monday's issue called the 167,855 ETH whale distribution "complete" and its absorption the expiry of the bear's best evidence; fresh reporting has 450,000 ETH landing on Binance Sunday and $423 million more Monday, so "complete" was premature and the line is retracted, with the caveat that the source is mid-tier and the bands did not move.
-
The difficulty estimate was revised twice in three issues, from +6.71 to +5.24 to +2.045 percent, and the desk conceded it is a moving projection to be labelled an estimate, revised daily.
Second: the difficulty estimate has now been revised twice in three issues, +6.71 to +5.24 to +2.045 percent (mempool.space), and the desk concedes the series is a moving projection, will label it "estimate, revised daily" every time it prints, and will stop treating single revisions as news.
-
-
-
The France card was built around a Tuesday confidence vote that does not exist: François Bayrou lost his confidence vote on 8 September 2025, by 364 votes to 194, and Sébastien Lecornu is prime minister.
This desk owes its readers a correction it will not bury: the France card in Friday's and Sunday's issues was built around "Tuesday's confidence vote," and there is no confidence vote.
-
The next bitcoin difficulty step was printed at +6.71 percent around 18 September; the live projection had cooled to +5.24 percent around 19 September, a normal revision that should have been labelled an estimate.
Second, smaller: Sunday's issue printed the next difficulty step at +6.71 percent around September 18; the live projection has cooled to +5.24 percent around September 19 (mempool.space), a normal mid-epoch revision the desk should have labeled as an estimate more loudly than it did.
-
-
-
Hormuz traffic of four to six ships a day against a normal ten-plus paired a tanker-only count with another vendor's baseline; on one consistent all-commercial series the comparison is six transits against a pre-crisis 85.
A note on units, correcting Friday's issue: the "four to six ships against a normal ten-plus" line this desk printed Saturday paired a tanker count with a tanker baseline from a different vendor; on straits.live's consistent all-commercial series the comparison is six against 85.
-
The stablecoin float was printed at $312.2 billion; a later pull of the same series reads $311.6 billion, a vendor revision rather than a flow reversal.
A small correction to Friday's issue: the total printed there, $312.2 billion, reads $0.6 billion above today's pull for the same series; DefiLlama revises, the desk discloses, the direction of travel is unchanged.
-
-
-
Swiss consumer prices were reported as rising 3.5 percent in August, the highest since 1993; the August figure was 0.8 percent, from 0.4 in July, and the 3.5 percent figure does not exist in the Swiss series.
First, Wednesday's issue said Swiss consumer prices rose 3.5 percent in August, the highest since 1993, and built a story and part of the Economist's Corner on it. The August figure was released on Thursday and is 0.8 percent, from 0.4 in July, per the Federal Statistical Office as reported by FXStreet, FX.co and Bloomberg via swissinfo.
-
The MANTRA and TAC drains were dated to Monday and Tuesday, which were rekt.news publication dates; the exploits ran on 20 and 22 August, and six chains were already affected by 25 August.
Second, the issue dated the MANTRA drain to Monday and the TAC drain to Tuesday, which were rekt.news' publication dates; the exploits ran August 20 and 22, per rekt's own KiiChain post, Cosmos Labs acknowledged them on August 24 and urged halts on August 25, and six chains were already affected by then.
-
A payrolls consensus of 42,000 came from one source; Investing.com's calendar shows 58,000 and TradingEconomics 79,000, and the desk could not find 42,000 in any survey.
Third, the issue quoted a payrolls consensus of 42,000 from one source; Investing.com's calendar shows 58,000 and TradingEconomics 79,000, and the desk could not find 42,000 in any survey today.
-
-
-
The ETFs were said to have taken in 142 million dollars on the session the crowd reloaded; issuer data shows a 236.46 million dollar outflow that Tuesday and a 216.70 million inflow on Monday, so the figure was wrong in sign.
First, Tuesday's issue said the ETFs "took in 142 million" on the session the crowd reloaded; the issuers' data for Tuesday shows a 236.46 million dollar outflow, and Monday's inflow was 216.70 million, so the 142 figure matched neither day and the sentence was wrong in sign.
-
The Binance top-trader ratio was counted at two-to-one for an eighth day, but the streak began on 23 August, which made Tuesday the tenth day.
Second, it counted the Binance top-trader ratio at two-to-one for "an eighth day"; the exchange's own daily series shows the streak began on August 23, which made Tuesday the tenth day and Wednesday the eleventh.
-
Korea's export bid was called the one bid in the world that did not care about Tuesday, a line written off one session and one trade print while 9.84 trillion won of foreign net selling since 20 August was already on the record.
Third, it called Korea's export bid "the one bid in the world that did not care about Tuesday" off one session and one trade print, when 9.84 trillion won of foreign net selling since August 20 was on the record and would have qualified the line; the KOSPI fell 3.99 percent the next day.
-
The JOLTS release was listed on Wednesday's calendar, but the Bureau of Labor Statistics published it on Tuesday at 7.271 million.
Fourth, it listed the JOLTS release on Wednesday's calendar; the Bureau of Labor Statistics published it on Tuesday at 7.271 million, and the desk missed it as a same-day print.
-
-
-
A Norges Bank holding of roughly 11,549 bitcoin, worth about 725 million dollars, was attributed to the fund's half-year report, which contains no reference to bitcoin, cryptocurrency or digital assets.
That issue stated that Norges Bank holds roughly 11,549 bitcoin worth about 725 million dollars and attributed the figure to the fund's half-year report. It does not.
-