Topics

The notes in the knowledge base are written for someone who already knows what a yield curve is and why a Treasury auction can ruin a Tuesday. They are terse on purpose. This page is the patient version: it walks through the themes the notes keep returning to, explains the mechanics in plain terms, and points at the notes behind each claim. Links in marker color go to a note; the notes carry the numbers and the sources, this page carries the explanation.

A word on authorship, stated once: the notes are Jörn Dinkla’s own thinking material. This page was drafted by Claude from those notes and read by him before publishing. Neither is investment advice. Both are one person’s attempt to sort what he reads about markets into piles that make sense.

Three terms carry the rest of this page

A yield is the interest rate a bond pays to whoever buys it today, at today’s price. When a bond’s price falls, its yield rises; the two are the same fact seen from different ends. Most of what follows is about the yields on US government bonds, because those set the floor for what everyone else pays to borrow.

The deficit is what a government spends beyond what it collects in one year. The debt is the pile those deficits have made. The distinction sounds pedantic and turns out to be the source of half the arithmetic errors the notes catch.

A real rate is a yield with inflation subtracted. If a bond pays 4% and prices rise 3%, the lender is really earning 1%. For a decade after 2012, real rates in much of the rich world were below zero: lenders were paying for the privilege. That period is over, and its ending is the hinge for most of the topics below.

The map

1. The interest bill starts to steer policy
2. The bond market as the last disciplinarian
3. Treasury plumbing, or how to move a yield without admitting it
4. Why a metal that pays nothing sits at the center
5. When bonds stop protecting stocks
6. The dollar, the yen, and the cost of defending a currency
7. The AI trade and the shape of the US stock market
8. Reading the crowd
9. China and Europe
10. How to read a chart posted by a stranger

The interest bill starts to steer policy

In the summer of 2026 the US federal government was paying about 18.5 cents of every dollar it collected in taxes straight out again as interest — a share last seen in 1991. The comparison hides the interesting part. In 1991 the government paid roughly 8% on its long bonds; in 2026 it pays around 5%. The same share of revenue at a much lower rate means the debt is far larger relative to income than it was then, and every further increase in rates bites harder. The note Fiscal dominance carries the figures and, characteristically, also corrects the viral post that reported them.

Fiscal dominance is the name for what happens next, if it happens. Normally a central bank raises interest rates when inflation is too high and lowers them when the economy weakens, without asking the finance ministry’s permission. When the interest bill is large enough, raising rates blows a hole in the budget. The central bank then faces a choice it was designed never to face: fight inflation and worsen the government’s finances, or keep rates low and accept inflation. The precedents are real. The US Federal Reserve capped long bond yields at 2.5% from the Second World War until 1951; Japan held its yields down by decree for years. “Financial repression” is the polite term for this, and The financial-repression toolkit lists the parts: captive buyers (banks, pension funds, and lately stablecoins) who must hold government paper, a ceiling on yields, and eventually rules that make it harder for savers to take their money elsewhere. The note’s dry observation is that capital controls never arrive announced; they arrive dressed as consumer protection.

Whether a debt is sustainable depends on a single inequality, explained in r minus g — debt dynamics: if the economy grows faster than the average interest rate on the debt, the ratio of debt to income shrinks on its own; if not, it grows, unless the government runs a surplus before interest. Two mistakes recur in the wild. People compare real growth with nominal debt growth, which mixes units. And they multiply the whole debt by a new higher rate, when in fact only the debt being refinanced this year pays the new rate — the error unpacked in Stock vs flow repricing. The correction matters in both directions: a rate rise hurts less and later than the headlines say, but it also means “inflating the debt away” works only briefly, a point picked up in the next section.

What makes the current episode unusual is timing. Past jumps in US debt came after a depression or a world war. This one is running during an expansion with 4% unemployment, which is why Deficits at full employment treats the roughly 6% of output being borrowed each year as a starting point, not a peak. The next recession adds to it. Two side notes complete the picture. The Fed itself lost money three years running — the delayed price tag of its bond buying, explained in The Fed's operating losses — which quietly costs the Treasury the profits the Fed used to hand over. And The Big Debt Cycle template records how one well-known investor compresses centuries of such episodes into a template of three gauges and a sequence. The note takes the template seriously and then names its weakness: with 35 historical cases to choose from, almost any present can be made to look “consistent.”

What is observation here: the interest share, the deficit, the Fed’s losses. What is interpretation: that these add up to a regime rather than a bad year. What is speculation: the timing, on which even the template’s author says “three years, give or take two.”

The bond market as the last disciplinarian

On 16 August 2026 the US Treasury sold thirty-year bonds at a yield of 5.216%, the highest at auction since 2001. Three days later it doubled a program that buys such bonds back. The sequence, recorded in Bond vigilantes and Treasury buybacks are not debt reduction, is the cleanest specimen the knowledge base has of a very old idea: that investors, by refusing to lend cheaply, can force a government’s hand where no election can.

The term “bond vigilantes” dates from 1983 and describes exactly that. Its distinguishing feature is that nobody has to raise rates. The Federal Reserve did not hike in 2026; holders of long bonds simply demanded more compensation, and the yield rose by the market’s own hand. A seasoned investor’s correction, preserved in the note, is worth keeping: at 5.2% against 3.5% inflation and a 6% deficit, the market was not being a vigilante. It was “a pushover that had finally begun to clear its throat.”

To read these episodes you need the yield curve, explained from scratch in Yield curve spread (the basics). The curve plots what the government pays to borrow for two years, ten years, thirty years. Normally the longer loans pay more, since the lender bears more inflation risk for longer. When the curve steepens, the important question is which end moved. Curve steepener (2s/30s) separates the benign version (short yields fall because rate cuts are expected) from the worrying one (long yields rise because lenders want more for the risk). The second is the vigilante mechanism in chart form.

The backdrop to all of this is the theme of The era of free money is over: real yields, the inflation-adjusted kind, are positive again in nearly every developed country. Savers are paid again. Governments face a real financing constraint again. Every asset whose value lies far in the future — growth stocks, real estate, gold — must now compete with an alternative that finally yields something. The note treats this as a regime change, not a data point, and the knowledge base largely agrees with it.

Here the notes divide into two camps, and the division is the most useful thing on this page. One camp holds that the state will inflate the debt away rather than submit to higher rates; that is the Debasement trade, treated in the section on gold. The other camp, stated cleanly in The inflation surprise only works once, points out that inflation only transfers wealth from lenders to the state when it is unexpected. Once lenders expect 5% inflation they demand 5% plus a real return, and the state refinances at the new price within a few years. So inflation can shrink the existing debt but cannot pay for a persistent deficit. The two camps agree on the mechanism and disagree on the politics: will the state suppress yields before lenders reprice them, or will lenders reprice first? Both sides are honest about the fact that the 1940s worked because yields were capped and the deficit vanished after the war, while the 1970s failed because neither held.

A smaller idea belongs here as well. TACO trade records the observed pattern that aggressive policy announcements get walked back once markets inflict enough pain. Its 2026 variant is instructive: the party that flinched at a 5.33% thirty-year yield was not the President but the Treasury.

Treasury plumbing, or how to move a yield without admitting it

A headline in 2026 announced, with two alarm emojis, that the US Treasury had bought back two billion dollars of its own debt. It read like a government paying down what it owes. It was not. The program buys old bonds and pays for them by issuing new ones; the total owed does not move. Treasury buybacks are not debt reduction opens with this correction and then follows the program as it changed character late in the summer, when it was doubled and moved to the long end after the yield spike. A German bank called the enlarged version “soft financial repression.” Two billion dollars against a debt approaching forty trillion is, as the note puts it, a scale joke. The note’s conclusion is that the operation matters as a signal — the Treasury announcing that long yields are now a policy variable — rather than as a purchase.

Several notes describe the same lever from different angles. The core idea is duration, a word that here means how long lenders’ money is locked up and therefore how much interest-rate risk they carry. A government can fund itself with bills (loans of a year or less) or bonds (loans of ten to thirty years). Fund the deficit mostly with bills and the market holds less long-term risk, which frees appetite for other assets; that is what central-bank bond buying did after 2008, and Activist Treasury issuance (stealth QE) describes the Treasury achieving a similar effect without the central bank. The catch is on the other side of the ledger: bill funding means the state refinances constantly at the front of the curve, so any rate rise hits the budget immediately. One critic’s phrasing, recorded in Buying back below par is not refinancing: the Treasury cannot issue long bonds now because that “would crash the economy now,” so it will “crash the economy later when short-term debt matures.” That same note also dismantles the viral claim that buying back 3.44% bonds with 4% bills is like refinancing a cheap mortgage into an expensive one. The Treasury pays the market price for the old bonds, around 78 cents on the dollar, so the swap is roughly neutral in present value and the loss lands on the bondholder. The analogy is right about the rollover risk and wrong about the rate.

The most elegant piece of plumbing is the one in FIMA repo — borrowing against Treasuries instead of selling them. Foreign central banks hold enormous amounts of US Treasuries. When one of them needs dollars in a hurry, the obvious move is to sell some, which would push US yields up. The facility lets them pledge the bonds as collateral and borrow dollars overnight instead — a pawn shop rather than a sale. The bonds stay off the market. The August 2026 yen episode, treated below, is the specimen, and the European Central Bank has since built the same window for its own bonds.

The newest buyer of bills is a regulatory creation. Since a 2025 law, every compliant dollar stablecoin (a crypto token pegged to one dollar) must hold its reserves in cash and short Treasury bills. Each token minted is thus a forced purchase of government paper, and the Treasury Secretary has said out loud that this is how the dollar stays the world’s reserve currency. The stablecoin T-bill bid takes the mechanism seriously and lists what the framing skips: it is demand for bills only, not for the long bonds where the pressure is; part of it is existing money changing costume; and a stablecoin is a money-market fund without a lender of last resort, which makes it a run-prone structure. Put the pieces together — buy back long bonds, issue bills, mandate a buyer for the bills — and you have what one note’s source calls a Treasury-led Operation Twist. The observable fact is narrower than the conspiracy version, but the direction is not in dispute.

Why a metal that pays nothing sits at the center

In one week of August 2026, China’s central bank bought more gold as the price fell, as it had every month since the March top, while retail buyers in a Shenzhen jewelry district paid five percent over the futures price chasing the rebound. Same week, same metal, opposite reflexes. The central-bank gold bid is the longest note in the knowledge base because it keeps collecting such specimens: who is buying, and when, tells more than the price.

The reason gold shows up in almost every other cluster is explained in Real rates and gold. Gold pays nothing. Its price is therefore governed by what you give up to hold it. When inflation-adjusted yields on safe bonds are solidly positive, holding gold is expensive and it tends to lose. When those yields are negative or falling, the “safe” alternatives are quietly losing purchasing power and gold tends to win. This makes gold a running vote on one question: are financial claims on the future delivering real returns, or not? The note is candid that the tidy decade-by-decade story rests on about three observed cycles — “a story, not a statistic.”

The debasement trade is the thesis built on top. As Debasement trade puts it, with structural deficits, high yields and years of above-target inflation, the state can neither cut spending nor afford the kind of rate shock that ended the 1970s inflation. So it will inflate instead, bonds and cash lose, and the trade is stocks and hard assets. The note’s most useful line is about the photograph attached to the original post: an image of the Fed chairman who did raise rates to 20% in 1980. That path crushed gold for two decades. The debasement trade is a bet that a second such episode is fiscally impossible now, and its risk is precisely that bet being wrong. The arithmetic for why it might not be wrong is in the same note: in 1980 the debt was about 30% of output, today it is around 120%.

What counts as a hard asset is settled in Hard assets: anything that is not a claim on a fixed amount of currency. Gold, silver, commodities, land, infrastructure. Stocks are the middle case, paper claims on businesses that can raise prices, which is why the phrase is always “equities and hard assets.” The note’s three-bucket version is the one to remember: fixed claims lose to inflation; productive claims resist it over decades but not quarters; scarce unproductive things are the pure hedge, with their own brutal drawdowns. Its five-year scoreboard shows long Treasuries losing a third while gold miners tripled — with the caveat that the window starts at the 2021 peak of cheap money, so the table mostly shows what reversed.

Several notes handle the policy side. Gold revaluation as bond-market release valve presents the argument that a much higher gold price would restore foreign central banks’ reserve capacity outside Treasuries and so relieve the bond market, alongside a rebuttal that historically gold rises before yields do, making it a smoke alarm rather than a valve. The note’s verdict: both agree that much higher gold and a worse bond market travel together; the dispute is over which causes which, and no chart settles it. The President amplifying a $10,000 gold call is filed there as a tell, not a plan. Gold price targets are scenario claims treats every such target as a scenario with the scenario left out: “say which one you are buying.” Two gold markets and Basel III made gold a "Tier 1 asset" — what it actually did cover the plumbing, the first describing how sanctions are splitting physical gold into a Western pool and a Chinese one, the second correcting a perennial meme about banking rules. Miners vs metal reminds the reader that a gold mining company is a stock first and a gold claim second, and Silver's solar demand explains why silver, half money and half industrial input, has a structural buyer gold lacks.

The knowledge base’s stance on all of this is consistent: the mechanism is sound, the specimens are real, and nearly everyone quoting them sells something — a newsletter, a fund, the metal itself. Both facts are recorded, every time.

When bonds stop protecting stocks

For decades a simple portfolio worked: own stocks, own government bonds, and when stocks fell the bonds would rise and cushion the fall. In 2022 both fell together, and in August 2026 a Federal Reserve research letter said the relationship had “flipped for the first time in decades.” Stock-bond correlation flips with the shock type explains why with one distinction. In a world of demand shocks, bad news lowers growth and inflation at once, so bonds rally when stocks fall. In a world of supply shocks, inflation rises while growth falls, and stocks and bonds fall together; the bond allocation becomes a second inflation bet instead of insurance. The note gives the source unusual credit: a Fed research shop with nothing to sell, rare among the knowledge base’s specimens.

The channel that keeps the flip alive is oil. Oil is the inflation transmission traces the chain: crude sits inside transport, agriculture, manufacturing and freight, so a sustained oil rise lifts the whole economy’s cost base, then inflation expectations, then yields, and central banks lose room to ease exactly when governments most need cheap funding. The same note keeps the chain and discards the chart pattern that came attached to it.

Two related notes reframe familiar assets in these terms. Equity duration treats growth stocks as the stock market’s long bonds: their value sits in cash flows far in the future, so rising real rates compress them first and hardest, while value stocks and commodity producers behave more like short bonds. And Shannon's demon — rebalancing turns volatility into return takes apart a viral story about turning two mediocre stocks into a fortune by rebalancing. The real result is modest and well known: rebalancing between volatile, weakly correlated assets adds a fraction of a percent a year, because compounding runs on the geometric mean and volatility is a cost the arithmetic mean hides. The note’s larger point is that this bonus requires assets that actually diverge, which is exactly what the flipped correlation has taken away.

The dollar, the yen, and the cost of defending a currency

At the start of August 2026 the United States and Japan intervened jointly to prop up the yen, spending somewhere between ten and eighty billion dollars depending on who is counting. Four trading sessions later the exchange rate was back where it started. FX intervention has a half-life of days records the episode and the general rule: intervention against a fundamental trend buys days, because reserves are finite and the market’s flow is not, and everyone knows which runs out first.

The trend being fought is the carry trade, explained in Yen carry trade: borrow cheaply in yen, invest in higher-yielding dollar assets, pocket the difference. While the gap between Japanese and American interest rates stays wide, the trade is self-reinforcing, and each intervention-driven rally in the yen simply hands the traders a better re-entry price. The note lists the only real exits — Japan raising rates meaningfully, or US rates collapsing — and then records that by the end of August Japan’s two-year yield had reached a 31-year high. The exit, in other words, is being priced.

Why Washington cared is the subject of The yen defense is a Treasury defense. Japan is the largest foreign holder of US Treasuries. A yen in freefall forces Japanese institutions to raise dollars by selling their most liquid foreign asset, which is those Treasuries, into a bond market already under strain. On that reading the US did not intervene to help Japan; it intervened to keep a forced seller out of its own bond market. The note marks this as inference, not announced policy, and notes that no other explanation covers the strange detail that the US first sold euros to buy yen, without telling the European Central Bank.

Paper losses are real when you need cash supplies the pressure behind the scene. Japan’s four largest life insurers carry the equivalent of nearly a hundred billion dollars of unrealized losses on their government bonds. Such losses cost nothing until something forces a sale — a collateral call, a deposit run, a wave of policy surrenders — and then they are losses like any other. The note is equally careful in the other direction: without a trigger, paper losses quietly shrink to zero as bonds mature, so the stock of them is a vulnerability, not a countdown.

Finally, The dollar-shortage thesis presents the mirror image of the debasement trade: the world is structurally short of dollars because so much debt is denominated in them, so the dollar rises in a crisis. The note follows the argument to the step where it breaks — a government cannot profit from a fire sale of its own bonds while it is the largest issuer of new ones — and keeps the half that survives. A small method note, A lone tenor gapping is liquidity, not information, belongs here too: when one maturity of bond moves sharply while its neighbors sit still, that is a single large order meeting a thin market, not information.

The AI trade and the shape of the US stock market

For as long as anyone has kept such lists, the largest issuers of corporate bonds have been banks; balance sheets are their product. In 2026 Amazon and Oracle appeared at fifth and sixth, ahead of Wells Fargo and Citigroup. Big Tech becomes a borrower records the table and draws the conclusion: the AI investment wave is restructuring the bond market’s composition, not only the stock market’s. An investor who bought corporate credit to diversify away from tech now holds the same names on both sides.

The mechanism runs through cash. The Mag-7 free-cash-flow bet shows the combined free cash flow (cash generated after investment) of four giants collapsing from roughly 230 billion dollars in 2024 to about 85 billion in 2026 as spending on data centers ate the surplus, followed by a consensus forecast of an explosion to 640 billion by 2030. The note treats the first half as data and the second as the bull case in its purest form, an estimate from the same analysts whose long-term growth expectations sit at record highs. AI circular financing maps the web of commitments underneath: about 46 billion dollars of equity actually deployed against roughly 879 billion of multi-year purchase promises, or about nineteen dollars of promise for each dollar of cash, with chipmakers investing in labs that commit to buy computing from cloud providers that buy the chipmakers’ chips. The historical rhyme the note offers is the telecom equipment makers of the 1990s lending their customers the money to buy the gear.

Credit noticed first. Credit leads equities explains why bondholders reprice risk before shareholders do: a bond has no share in the upside, so it reacts as soon as debt-funded expansion stops adding up. The signal is not “AI” but the gap between the cash-rich giants, whose default insurance barely moved, and the leveraged periphery that borrowed to build. Zombie firms and Capex booms and the gold hedge supply the longer view — the first on firms that survive only because refinancing is cheap, the second on the recurring pattern in which the builders of a boom lose and the users of the cheap capacity they leave behind win. That second note is also a textbook case of the knowledge base’s habit: it takes a viral statistic about gold during past booms, notices that for four of the five booms the gold price was fixed by law, and keeps only the part that survives.

The valuation notes describe how expensive all this has become and how hard that is to measure. Buffett Indicator and Shiller CAPE both show US stocks near or above their most expensive readings on record; both notes also explain why their denominators have drifted and why the honest use of such gauges is setting expectations for the next decade, not timing the next month. US share of world equity and Top-10 turnover describe the concentration: the US is about 62% of world stock market value, ten companies are about 38% of the US index, and the list of those ten has historically churned almost completely from one decade’s panel to the next. A “global” index fund is, in the note’s phrase, a stack of correlated concentration bets. Now show Japan handles the standard warning about stocks that took 34 years to recover and the standard reply that Japan in 1989 was the most expensive market ever recorded. The note keeps both: the counterexample is weaker than it looks, and it still proves the worst case is unbounded.

One speculative thread sits at the edge of this cluster. AI productivity vs debt-based money presents the thesis that AI-driven productivity is deflationary and therefore incompatible with a monetary system that needs steady inflation to keep debts serviceable. The note’s check is the one worth learning from: if that were the dominant force, bond prices would be rising, and they are not — so the thesis only fits the tape if the state pre-empts deflation by inflating first, at which point it collapses back into the fiscal-dominance story above.

Reading the crowd

On 4 August 2026 more than four million call options on the S&P 500 changed hands, the most ever, on a day the Dow closed at a record. A call option is a bet that prices will rise, and a wave of them near a high tells you that everyone who feared missing the rally has now paid up to join it. Call-volume extremes records the day and then does the arithmetic the headline skipped: option volume has trended upward for years, so part of “the most ever” is growth, but the print still sat 60% above its own trend, which is information.

Nearly every indicator note in the knowledge base lands on the same conclusion, and it is worth stating once because it is so often missed: extremes in sentiment and positioning measure risk, not timing. BofA Bull & Bear Indicator makes the asymmetry explicit by reading its own signal history. Extreme fear has marked sharp, tradeable bottoms because panic is an event. Extreme greed has often fired years before the top because euphoria is a process that can run for quarters. Fear gauges reset in days shows the same thing from the other side: days after the yen drama, with nothing resolved, the market’s fear gauge printed its low of the year. Implied volatility prices hedging flows, not narratives, so a low reading is a statement about the cost of insurance, not a sell signal.

The other indicator notes are variations. Margin debt and net credit balances measures leverage: investors who owe more than they hold in cash are structurally forced to sell into declines. ETF launch boom reads product proliferation as a map of what retail currently craves, and finds the answer is leverage, not diversification. Analyst growth expectations as contrarian gauge treats record analyst optimism as a record hurdle rather than a forecast. Most-shorted baskets outperform explains why the most-hated stocks keep beating the most-loved ones during melt-ups: every short position is a future forced buyer. Crowded bond shorts applies the same logic to long-dated Treasuries, the consensus short of the cycle, while noting how much of that short interest is hedging rather than conviction. Narrative mention counts counts how often a word appears in the financial press and observes that peaks in “debasement” coverage have marked local tops in the debasement trade, not entries; by the time a thesis is in every article, the marginal buyer who reads articles is already in.

Two notes handle rotation rather than extremes. Capital rotates, it doesn't leave observes that when index leadership exhausts, money mostly moves into whatever nobody is discussing, with the 2000 episode as the specimen: real estate stocks bottomed the same month the Nasdaq peaked. Hindenburg Omen is the knowledge base’s treatment of a famous crash indicator, and it is gentler than the joke about it predicting far more crashes than have occurred: clusters of triggers do carry elevated short-term risk, on a tiny sample with a threshold chosen after the fact, and the trigger condition is also the signature of a violent rotation, not only a top. September effect and Follow the fiscal flow round the cluster out, the first sizing a real but thin calendar effect honestly, the second recording the cynical heuristic that each five-to-seven-year market theme is a government program with a stock market attached.

China and Europe

Two smaller clusters look east and across the Atlantic. On China, China's money-supply overhang starts from a chart showing China’s money supply at more than twice the US figure for an economy a third smaller, discards the standard explanations along with the poster’s teased secret one, and keeps what survives: because capital controls block the exits, Chinese savings sit in bank deposits, and a captive pool of that size is potential energy. The note then watches the energy leak out through the one door that is open domestically, gold funds, and observes that Chinese government bond yields sit three percentage points below American ones despite a comparable debt load — the price of the closed door, and financial repression running at full capacity. China moves up the value curve records the flip of Germany’s trade in machinery with China from a fifteen-year surplus to a deficit, and the tension that the German stock index hit records in the same weeks. Index and economy have decoupled.

On Europe, German fiscal Zeitenwende follows Germany’s turn from fiscal anchor to structural borrower: over a trillion euros of planned new debt by 2030, interest costs nearly tripling, and a bond market described as “whispering, not shouting.” The note tracks the whisper getting louder through the summer of 2026. The US–Europe prosperity gap survives the median is mostly a method lesson: the usual European comfort is that American averages are distorted by billionaires, so the note switches to the median and finds the gap survives.

How to read a chart posted by a stranger

Most of the knowledge base’s raw material arrived as posts on X, each with a chart attached and a conclusion on top. The most-linked note in the whole collection is not about markets at all but about reading: Check the stat against its own chart. Its instruction is simple. Before believing the text, read the table or chart the post itself attached. The headline is routinely rounded up, “eight of nine” becoming “every single time,” or shifted, a claim about two weeks holding only at three. The image often says more than the words and sometimes contradicts them.

The other method notes are a small toolkit for the same task. Records in a growing series observes that “highest ever” is the cheapest headline in finance, because any series that grows over time sets records in most normal years; the honest question is how far above its own trend a print sits. Reference-class shopping catches arguments that smuggle their conclusion into the choice of historical comparison, with one week in which the same bank’s chart book found “room to run” by two different yardsticks while a value manager found the opposite by a third. Ratio charts explains the legitimate use of pricing one asset in units of another to see a relative regime, and the abuse of freehanding the future half of a pattern onto the chart. Dual-axis divergence charts shows how two series on independently scaled axes will display whatever “divergence” the axis choice manufactures. "Priced in gold" claims takes apart a claim that US stocks, measured in gold, are worth less today than in 1971, by noticing the endpoint chosen and the dividends omitted, and then keeps the real content: the unit you think in shapes what looks like “up.”

Two notes cover the sources rather than the charts. Underwriter research conflict observes that research on a big new stock listing is written by the banks that earned the listing’s fees, so uniform buy ratings measure the fee pool, not the company. Float and lockup overhang explains why the early price of a stock with only three percent of its shares trading is not evidence about the other 97%.

Running through all of these is a habit the notes never drop: naming who is talking and what they sell. Newsletter writers, fund managers, gold dealers, residency brokers, a Fed research department with no book to talk. The point is not cynicism. A claim can be true and still be an advertisement, and the notes’ consistent move is to discount the frame while keeping whatever mechanism survives without it. It is, when you notice it, the same discipline the analytical observer owes the reader — and the reason a page like this one ends by saying that nothing here is advice, and everything here can be checked against the note it links to.

What the notes do not yet cover

Writing about the notes shows where they thin out. There is almost nothing on credit spreads beyond the AI cluster, nothing on private credit, one note on options, and the commodity coverage stops at gold, silver and oil. The notes themselves point at gaps they already know about: the term premium, the extra yield lenders demand for locking money up longer, is referenced repeatedly and not yet written; so are the kelly criterion and market breadth. Those dashed links are the queue.