Decent Newsroom
Sign In

The Discount Rate Arrives

By Roger November 15, 2026
Bitcoin fell 14% in a week and nothing in the protocol broke. A Fed preparing to hike into 6.5% producer inflation repriced every long-duration asset; the wreckage is legible in the periphery, not in the core.
The Discount Rate Arrives

The Discount Rate Arrives

An empty concrete hall with one shaft of daylight crossing the floor

Nothing broke.

That is the first thing worth noticing about last week. Bitcoin did not halt, no miner cartel censored a block, no exchange froze withdrawals, no consensus rule failed under load. The chain did exactly what it promised for seven straight days. And the price still fell about 14% in a week, from roughly $75,750 on 7 November to a low near $65,000 on Saturday morning, breaking Friday’s trough of $67,555 and printing the weakest level since mid-August (Bitcoin Magazine, 14 November 2026). CoinDesk reported an intraday push below $63,000 and called it a fourth capitulation wave.

An asset that breaks nothing and loses a seventh of its value in five sessions is not telling you a story about itself. It is telling you about the price of the unit it is priced in. So this piece is not about Bitcoin. It is about the thing that moved, and about why the wreckage showed up everywhere except the one place people were staring at.

The price of money, priced again

Start with Friday’s data, because the sequence matters more than any single print.

The Labor Department said producer prices for final demand rose 0.3% in October, below the 0.4% consensus. The monthly miss was treated as good news for about four minutes. The annual figure, unadjusted, was up 6.5%, the largest twelve-month rise since November 2022 (Bitcoin Magazine, citing the Bureau of Labor Statistics, 14 November 2026). Bond traders read the second number, not the first. The ten-year Treasury yield traded as high as 5.17% and closed at 5.15%, a four-week high. The dollar index sat at a nineteen-month high, its strongest since April 2025.

Then the speaking. Boston Fed President Susan Collins: “soon, some additional tightening of Fed monetary policy may be appropriate.” San Francisco Fed President Mary Daly: policymakers will, “in all likelihood, raise interest rates.” By Friday, CME FedWatch had a quarter-point hike fully priced for the 8-9 December meeting.

Read that again, slowly, because the combination is the whole story. Producer prices are running at 6.5% year over year. The Fed is preparing to raise its policy rate, to a level that is still, on any honest measure, below the rate at which producer prices are compounding. An own calculation makes it explicit: with the ten-year at 5.139% and producer inflation at 6.5%, the real ten-year yield on producer prices is roughly minus 1.36% (own calculation, 15 November 2026, using the ten-year yield of 5.139% and the October PPI print of 6.5%).

Six sovereign ten-year yields, all below the 6.5% US producer price inflation rate

This is the part that gets lost in the daily commentary. What the market is doing is not “tightening.” It is moving nominal rates toward an inflation rate that has already been running hot for years, while still refusing to go above it. Every sovereign yield in the developed world on my screen this morning sits below that 6.5% producer print: the US ten-year at 5.139%, the British gilt at 5.338%, Italy at 4.427%, Spain at 3.943%, Germany at 3.390%, Japan at 2.872% (own measurement, 15 November 2026, 09:30 CET). Six curves, six governments, one shared refusal to pay a positive real return on producer prices.

That is not monetary discipline. That is monetary arithmetic being negotiated. And when the negotiation moves a quarter point, everything whose value depends entirely on the future gets repriced, because a discount rate is a ruler and every long-duration asset is measured with it. Gold, which pays nothing either, is at $3,886 an ounce. Bitcoin, which also pays nothing, is at $64,530, or 16.6 ounces of gold per coin (own measurement, 15 November 2026). Bitcoin is 48.9% below its record intraday high of $126,210.50 (own calculation). If the story were simply “zero-yield assets fall when rates rise,” you would expect both of them to fall, and gold has not been behaving like Bitcoin.

So the rate move is necessary but not sufficient. Something else is doing the work. That something is who owns the marginal coin.

Who has to sell

An ETF share is the most impatient form of ownership ever invented for an asset whose entire thesis is patience. It is marked daily, reportable weekly, benchmarked quarterly, and held by people whose mandate is a calendar year. When the discount rate moves against a position that produces no cash flow, that holder does not get to sit still and wait for the thesis. They get a call from a client.

The flows confirm the mechanics. My own tally of the daily US spot Bitcoin ETF totals gives minus $1,281.7 million for the trading week of 2-6 November and minus $1,065.3 million for 9-13 November, a two-week total of roughly minus $2.35 billion, with the most recent week averaging minus $213 million per trading day (own calculation from the daily fund-level table, 15 November 2026). The Block put the broader picture at about $1.55 billion of outflows across US spot crypto ETFs for the week, noting that the only weekly inflow among the fifteen categories it tracks was $227,000 into Avalanche funds (The Block, 14 November 2026).

Two hundred twenty-seven thousand dollars. Out of fifteen categories. That is not rotation. That is an exit.

Bar chart of daily net flows into US spot Bitcoin ETFs, 27 October to 13 November 2026

James Check, who has spent a decade watching this asset’s holders behave, framed the human side of it on TFTC this week: “They are now selling at the very price they wish they had taken advantage of three months ago” (quoted by CoinDesk, 14 November 2026). His read of the holder base is that 75 to 80% of the total wealth in the system sits in long-term hands (his marker for a bottom-of-bear, early-bull distribution), while the cohort that arrived in the last five months does the selling. His mean-reversion index has Bitcoin in the bottom quartile of its entire trading history, which by his reading implies a 75% historical probability of being higher from here (TFTC, 14 November 2026).

That 75% deserves the skepticism I will give it below. But the structural point survives the skepticism: the marginal seller in this market is not the same person as the marginal holder. The seller is leveraged, calendared, and accountable to someone else. The holder is not. That asymmetry is what makes a liquidation cascade look like a verdict on the asset when it is only a verdict on the financing.

Where the malinvestment is legible

Here is where the Austrian framework earns its keep, and where most market commentary goes soft. If a rate move is repricing duration, the severity of the damage should scale with how far an asset’s valuation depended on cheap credit and on the promise of a future nobody had to underwrite yet. So look at the periphery, not the core.

Ethereum traded at $1,795 this morning (own measurement, 15 November 2026). Arca’s Jeff Dorman spent a livestream dismantling Vitalik Buterin’s November 8 argument that ETH should be understood as a store of value, calling it “really misguided”. His arithmetic is the interesting part: Ethereum did about $6.2 million of revenue in the last thirty days by his check of DefiLlama. Call it $10 million a month to be generous, annualize to $120 million, apply a 20x multiple, and you get a couple of billion dollars of business value. He then allowed roughly $40 billion for 200 million wallets each holding about 0.1 ETH for gas, arrived at something like $45-50 billion of justifiable value, and pointed out that ETH’s market capitalization was about $246 billion — leaving roughly $200 billion of what he called “just social value or hopium” (Unchained, 14 November 2026).

You do not have to accept every input to see the shape of the argument. A network whose fee revenue is $6 million a month does not have a cash-flow problem you can solve with a narrative. It has a duration problem: its price is a claim on a future adoption curve, and the discount rate just moved.

The rest of the periphery reads the same way. MegaETH cut 20% of its staff while pivoting toward AI, and the tell is in the footnote: decentralized finance value locked on the network fell from more than $245 million to $8.8 million (CoinDesk, 14 November 2026). Kava Lend lost roughly $3.8 million to a hack in which about 49.2 million KAVA was bridged to BNB Chain via LayerZero and sold for approximately $2.1 million in BNB and USDT, with the token down nearly 40% (The Block, 14 November 2026). XRP printed its lowest level since November 2024 as ETF outflows and shrinking stablecoin supply compounded the pressure (CoinDesk, 15 November 2026); it sits at $0.9552 on my screen this morning.

And then there is the quiet one, which I keep coming back to because it is not a price at all. Payward, Kraken’s parent, published third-quarter results led by a new headline number, “total platform margin,” of $487 million, 12% above the $436 million it now shows for the second quarter and 12% below the third quarter of 2025. The footnote says the old metric, Adjusted Revenue, is included “for comparison purposes only” and that “management no longer uses Adjusted Revenue to evaluate our business results.” On the old measure, third-quarter Adjusted Revenue was $553 million, down 15% from a year earlier. Adjusted EBITDA was $60 million, against $181 million a year earlier. Trading volume was $353 billion, down 30% year over year. Funded accounts: 7.1 million, up 37% (Unchained, 14 November 2026).

Seven million funded accounts and 30% less volume. More customers, less business. That is the signature of a boom that converted spectators into accounts and then ran out of spectators. And when the headline stops cooperating, the headline changes. That is not fraud, it is what managements do, but it is a data point about the cycle, and it belongs in the same paragraph as the $227,000 Avalanche inflow.

The operator leaves

There was a second, larger repricing last week, and it was not in crypto at all.

Nvidia’s chief executive Jensen Huang resigned on Thursday. AI stocks sold off, and Bitcoin Magazine’s account is that the shake-up is a plausible contributor to Bitcoin losing $72,000 on Thursday evening (Bitcoin Magazine, 14 November 2026). I want to be careful here, because one resignation is a corporate event and I am not going to dress it up as the end of a technological era. But the financial structure around it is worth naming, and Nathan Lambert at Interconnects named it plainly: the resignation is “a major setback for the US development of an open model ecosystem,” because Nvidia’s support for open models was “one of Jensen’s more ambitious, founder-style bets,” and he “was more than a cheerleader. He would stop a lot of shit from going wrong once early winds were stirring. He was willing to use his power on open models.” Nvidia’s near-term goal, in Lambert’s reading, is to “stabilize the ship, deliver to customers in the next 24 months, and prevent implosion” (Interconnects, 14 November 2026).

Notice what kind of capital a founder-style bet is. It is capital that does not have to clear a hurdle rate this quarter, because one person with authority decided the option value was worth it. That is precisely the capital that a rising discount rate eliminates first, not through malice, but because the person who could authorise it is gone, and the spreadsheet that replaces him has a hurdle rate in it.

The counter-signal is real and I will not pretend otherwise. llama.cpp and ggml both hit version 1.0, SGLang announced 1.0 at its summit, and a “No AI Slop Manifesto” emerged from the AI Engineer conference and reached the front page of Hacker News (Latent Space, 14 November 2026). Models keep shipping: Gemini 4 Hydrogen went generally available in GitHub Copilot; Gemini 4 Carbon took first place on MathArena, displacing GPT-6.1 Sol; Claude Fable 5.2 entered the Agent Arena at number one with a net improvement of 30.4%, and its API price fell from $10/$50 to $8/$40 per million input/output tokens. Independent checks are less flattering: Roboflow’s vision benchmark has Hydrogen at 29.3 seconds per image in high-effort mode against 51.4 for Argon and 132.3 for Carbon, while Carbon detects better, and one 3D-modelling test found Hydrogen more expensive than Argon because it loops on over-verification.

That is what a maturing technology looks like from the inside: the benchmarks converge, the prices fall, the marketing gets louder, and the marginal dollar starts asking what it is buying. None of that requires a crash in Bitcoin. It does, however, explain why the same week produced a rate-hike repricing, an ETF exit, and the departure of the single operator whose personal conviction was underwriting a chunk of the open-model ecosystem.

What keeping the books actually looks like

Now the part that will not trend and should.

On 14 November, a disclosure landed on Delving Bitcoin: LNF-2026-0004, a duplicate funding outpoint denial-of-service in Core Lightning, fixed in v26.06.8. Every release from v0.12.0 through v26.06.7 accepts a funding_created message that reuses another channel’s funding outpoint, so two channels end up sharing a v1 channel_id. If both are live, reconnecting trips an assertion in connectd and crashes the node, at no cost to the attacker, and repeatably on every restart. If the first channel is already closed, closing the second trips an assertion in lightningd and the node crash-loops on restart until it is upgraded, at the cost of on-chain fees only. The bug was found by smite, using an IR scenario while fuzzing CLN’s funding flow, and flagged because CLN answered the malformed message with funding_signed. The closed-channel variant was found separately by Crypt-iQ (Delving Bitcoin, 14 November 2026).

Sit with what that is. A remotely triggerable, free, infinitely repeatable crash in the node software that a large share of the Lightning network runs on. Not a theft, not a consensus break, an availability bug, the least glamorous class there is. It was found by fuzzing, disclosed in public with the version range affected, attributed to the people who found it, and fixed in a point release. No token was launched to celebrate it. Nobody called it a narrative.

That is the maintenance layer, and it is the reason Check’s remark on TFTC is worth quoting rather than dismissing as cheerleading: he called Bitcoin Core “some of the best written code ever” and noted that the codebase came through a year of AI-enabled exploit waves untouched, while programming languages, servers, and credential stores across the internet were being stress-tested. I cannot verify the claim that Core was untouched by any specific exploit — nobody can prove a negative of that shape — but the direction is consistent with the disclosure above: the failures in this stack get found, published, and patched by named people, on a schedule.

Compare the incentives honestly. Kava lost $3.8 million in a bridge and the token fell 40%. Core Lightning had a free remote crash in every release for four years and the fix cost a version bump. One of those systems has a maintenance culture and the other has a marketing department. That difference does not show up in a weekly candle, and it will matter more than the candle.

The steel-man case

The bears are not stupid, and the strongest version of their case is not “Bitcoin is dead.” It is this: a Fed that hikes into a nineteen-month-high dollar, with the ten-year at 5.15% and producer prices at 6.5%, is draining the liquidity that every speculative asset has lived on for a decade; the ETF holder has no obligation to be brave; and the “discount zone” argument is a statement about the past, not a promise about the future.

I will go further, because the honest part of this piece requires it. Check’s 75% is an unconditional historical frequency: given that the index sat in the bottom quartile, Bitcoin was higher a year later in three out of four observed cases. It is not conditional on a hiking Fed, a 19-month dollar high, and two consecutive weeks of billion-dollar ETF exits. His own stated level to hold was $72,000; he said a break meant retesting the mid-$60s. It broke, and here we are at $64,530, about 5% below the 200-week moving average he puts near $68,000. His host opened the conversation by declaring the bottom is in or close enough not to matter, which is a position, not a measurement, and both men are talking a book they clearly hold.

There is also a discrepancy worth flagging rather than smoothing over: CoinDesk reported a print below $63,000, Bitcoin Magazine’s low was around $65,000. Different venues, different timestamps, different index construction. Anyone quoting a single low without saying whose is quoting a mood.

And the fair bear case on the periphery: Dorman may be wrong about Ethereum’s terminal value, and MegaETH’s 20% cut may be a strategic retreat rather than a collapse. But his structural question, what does this thing earn and what multiple does a rising rate justify on those earnings, is the question that actually gets asked in a repricing, and it is being asked everywhere at once.

The audit of your reasoning

One more item from last week, and then the consequence.

The UK’s Finance Bill 2026-27 would rewrite HMRC’s Schedule 36 information powers, replacing the existing standard of “computer records” with the phrase “any relevant thing.” Tax advisers read the new wording as wide enough to capture prompt histories and full conversation logs from AI tools such as ChatGPT, Copilot and Claude, wherever those tools played a role in preparing tax figures. Third-party notices could be served directly on the AI platforms, so an individual’s refusal to cooperate does not close the door. The same Budget package grants HMRC direct bank-account debt-recovery power from July 2029. HMRC’s spokesperson insists it “is not about going through everyone’s use of large language models” and that information can only be requested where a lawful basis already exists. Nimesh Shah of Blick Rothenberg notes HMRC wants to know “how tax figures have been produced using software or automated processes”; Seb Maley of Qdos puts it more bluntly: “It wants to know not just what tax figure a system produces, but how it arrived at that number” (TFTC, 14 November 2026).

That is a different kind of expropriation than the ones usually discussed. It is not a wealth tax or a banning order. It is a claim on the reasoning — the audit trail of how a number came to exist, held by a third party who is now compellable. Every AI assistant you use for anything financial becomes a potential witness against you, and you cannot self-host your way out of it if the tool lives in someone else’s datacentre.

Bitcoin sits outside that perimeter for one reason only, and it is not ideology: there is no counterparty holding your ledger to be served with a notice. The record is on your side of the table. That property is worth exactly nothing on a week when the price falls 14%, and it is worth a great deal on the week when someone asks you to prove how you arrived at a number.

The consequence

So where does that leave the arithmetic?

The one number I would watch is not Bitcoin’s price. It is the ten-year. As long as nominal sovereign yields stay below producer-price inflation, the repricing is incomplete: the market is negotiating the discount rate, not settling it, and every asset whose value is a claim on the future will keep getting marked against a ruler that is still moving. A Fed that hikes a quarter point into 6.5% producer inflation has not become restrictive. It has become slightly less generous.

Within that, Bitcoin’s behaviour is exactly what its supply schedule predicts. Its issuance does not answer price; it is fixed, and it will be fixed in December whatever the FOMC does. Demand is the only variable, which means a demand shock has no supply response to cushion it: no producer to cut output, no cartel to defend a price, no central bank to buy the dip. That is why a 14% week feels like a verdict. It is not a verdict; it is the sound of an asset with an inelastic supply being repriced by whoever happens to hold it that quarter.

The holders who remain are the ones who were never on a calendar. Check’s 75-80% of wealth in long-term hands is the closest thing to a structural answer this market offers, and it will be tested further, because the December meeting is three weeks away and the dollar is at a nineteen-month high.

What I will not do is dress up $64,530 as a floor, a bottom, or an opportunity. What I will say is narrower and more defensible: nothing in the protocol failed this week, a four-year-old remote crash in Lightning’s node software was disclosed and patched by named researchers in public, and the reason the price moved is that the price of money moved — which is a sentence about the dollar’s managers, not about Bitcoin’s engineers.

The bill for the last decade of cheap credit is being presented, one quarter point at a time. It is being paid, so far, mostly by the assets that were bought on the promise that the bill would never arrive. Bitcoin is on that list because it is the most liquid thing in the room, not because it is the most guilty. The distinction does not pay this month. It has paid every other time it mattered.

Sources

Write a comment