Mistakes and overstatements found after publishing, and how they were fixed. Figures show before → after. Newest first.
Each site also checks its own wording against the data every morning, and any mismatch is fixed and logged here.
The key points of “Why is the yen weak?” still said more yen was sold than bought after net yen demand over the last 12 months had turned positive. The sentence is now written from the sign of the data.
Net, more yen is sold than bought (¥+1.3 tn)The last 12 months tip slightly toward buying yen (¥+1.3 tn), down from ¥10.8 tn of net selling in 2024
Reversed the 30 Sep change that stopped counting cash matched by customer advances (contract liabilities). Like Pia's ticket money owed to promoters, such float keeps funding the business as long as it lasts, and judging each balance-sheet line company by company would break the one-yardstick rule. The past tests were rerun: only Japan period 1 now beats random picks (US period 1 is within chance). The lists now show cash as a share of market value, so readers can see how much of the cheapness is hoarded cash.
Chiyoda: EV/EBIT 2.4x (#3)0.7x (#1); cash is 83% of market value
US sales and profit now come from whichever XBRL tag a company uses for its latest period. Companies that had switched tags were showing figures from an old year.
Comcast: sales $121.6bn (FY2023 figure)$124.9bn (latest twelve months)
Fixed missing US debt. Comcast's long-term debt was filed under a different XBRL tag and was dropped, leaving $8.9bn (correct: $90.4bn). Some companies also had years-old balances mixed in.
Comcast: EV/EBIT 4.3x (#2 in the US)EV/EBIT 9.0x (#18)
Enterprise value no longer subtracts cash that customers paid in advance (contract liabilities), and now adds minority interests. All of Chiyoda's ¥145bn of cash was customer advances (¥191bn).
Chiyoda: EV/EBIT 0.6x (#1 in Japan)2.4x (#3)
Set the four past tests against what random picks give. Only Japan period 1 and US period 1 (narrowly) beat that; the other two were within chance.
Beat the market in all four (no chance range)Beat it in all four; two beyond chance
Corrected 'oversold works for stocks and gold miners'. With month-block bootstrap ranges, it held only for Japanese stocks overall and US stocks; miners and value stocks were within chance. 'Overheated' was not conclusive either.
Oversold: +1.7% to +3.8% for stocks and miners+1.7% to +3.2% for Japan overall and the US (miners and value unclear)