Decent Newsroom
Sign In

The Index Was the Buyer

By Roger November 12, 2026
MSCI removes Strategy from its indices at the close on Nov 30. The passive bid that financed the bitcoin treasury model goes with it, and the premium that model depended on is already gone.
The Index Was the Buyer

The Index Was the Buyer

On the first morning of December, a slice of Strategy’s shareholder base will be gone, and nobody will have made a decision about bitcoin. MSCI’s November index review, posted shortly after five o’clock New York time on Wednesday, removes the company from the MSCI USA Index at the close on November 30, and with it from MSCI World and ACWI. Nowhere in the rule that did it does the word bitcoin appear. (Bitcoin Magazine, 11 November 2026)

MSCI adopted that screen on October 15, after a consultation that opened in August, and it replaced an earlier and cruder plan to exclude companies whose digital assets made up half or more of total assets. MSCI dropped that plan in January. The final test asks whether a company’s operating assets equal at least 50% of total assets. Fail that, and five financial ratios come into play; four flags or more and an issuer is ineligible for the whole Global Investable Market Index family. Buffers slow the turnover: a current constituent is removed only if it failed the screen in each of its two most recent annual filing periods, and MSCI reserved the right to read the latest interim filing too.

A filing question, not a monetary one, which is why the same review also removed Metaplanet from the MSCI Japan Small Cap Index, Yellow Cake from the United Kingdom Small Cap Index, Skel Fjárfestingafélag from Iceland’s frontier small caps and Upexi from the USA Micro Cap Index. On MSCI’s October potential-impact list, Strategy alone accounted for more than ninety percent of those five companies’ combined float-adjusted market value. (Bitcoin Magazine, 11 November 2026)

Read the classification honestly and it is not even hostile. A company whose assets are overwhelmingly one financial instrument is not a company in the sense the index means. It is a fund that happens to have a ticker — and funds are priced against their holdings rather than against their prospects. MSCI spent a year looking for a neutral way to say this and found one that applies to uranium trusts and bitcoin treasuries alike.

The number is small, and that is not the point

Index funds must hold what the index holds. Funds tracking MSCI USA therefore have to sell Strategy by the close on November 30, at whatever price the market offers, because the decision to sell was made by a rule and not by a view.

The size of that forced trade is a matter of public record. Strategy told MSCI in its consultation response that funds tracking the global investable indexes hold about 13 million of its shares: 3.1% of basic shares outstanding, worth roughly $1.6 billion at Wednesday’s price of $122.35, and less than a single day’s typical trading volume of around 14 million shares. BloFin Research arrived at the same 13 million independently. JPMorgan, working from MSCI’s initial November 2025 proposal rather than the rules finally adopted, put potential outflows at $2.8 billion, rising to $8.8 billion if other index providers followed. (CoinDesk, 12 November 2026; own arithmetic on the share count and Wednesday’s close)

Three percent of the share count is not a liquidation. That is why the bullish read is available and why it is not stupid: Markus Thielen of 10x Research told CoinDesk the outcome was widely expected, that many funds were already underweight the stock relative to the index weight, and that the shares should find buyers slightly lower. He does not expect it to spill over into bitcoin at all.

Staring at the small number is the wrong move, though. What matters is not how many shares must change hands on November 30. What matters is what the buyer of those shares is buying, and why he would pay more than the coins are worth.

The premium was the machine

Strategy’s accumulation model is not a savings account. It is an arbitrage, legal and disclosed and brilliantly executed, and it runs on a single quantity: the gap between the market value of the equity and the market value of the bitcoin behind it.

Above 1x, every share sold buys more bitcoin than the share itself represents, so bitcoin per share rises for everyone who did not sell. That is accretion, and it is the entire product. Below 1x the same operation runs backwards: issuing shares destroys bitcoin per share for existing holders. Around 1x, equity issuance is a wash with fees attached.

In November 2024 the multiple peaked at 3.4x. This week it is roughly 1.08x on the company’s own measure, as CoinDesk reported. Both numbers need a footnote, because “mNAV” is computed two ways and the two answers differ by a lot.

My own arithmetic, on public data: 422.35 million shares at $122.35 is a market capitalisation of $51.7 billion. Against 850,364 coins at $73,433, the spot price on my own board at 09:31 CET this morning, the treasury is worth $62.4 billion, which puts the common equity at 0.83x the value of the coins. The 1.08x figure is an enterprise-level number, and the gap between 0.83 and 1.08 is precisely the preferred stock and the convertible debt that sit ahead of the common. The equity is now a levered claim on the coins, trading below the coins, with roughly sixteen billion dollars of senior claims in front of it. (Own arithmetic; holdings and cost basis per Bitcoin Magazine, 11 November 2026)

An equity that trades below the value of the assets behind it cannot grow by issuing equity, and the machine that ran on that gap has stopped.

Strategy’s 850,364 coins valued at the all-time high, at the average purchase price and at today’s spot price

The filing that shows the machine running in reverse

The clearest evidence is not the index decision. It is the weekly disclosure for the week ended November 8, filed on November 9, which CoinDesk pulled apart: Strategy raised $283.8 million by selling 2.05 million common shares, put $81.9 million of it into bitcoin (1,036 coins), and sent $188.3 million to buy back its own STRC preferred stock, with $13.6 million added to cash.

Read those proportions again. Sixty-six cents of every dollar raised went to retire a preferred claim. Twenty-nine cents went to coins. For most of the last five years the same pipeline delivered nearly everything to the coins. The effective price paid for those 1,036 coins works out to about $79,054 each, above the company’s own average cost and above Wednesday’s spot: a reminder that a weekly average is a timing artefact rather than a strategy, and also of how far the arithmetic has turned.

The STRC programme itself is not panic. Retiring a $100 par claim for $97.74 removes a 12% dividend forever, which is good capital allocation. STRC.live, which tracks the series from the filings, counts 24,960,007 shares retired (23.8% of the issue) at an average of $97.74, $2.44 billion deployed against a $3.00 billion programme, and puts the recurring dividend eliminated at $299.5 million a year. Every share bought below par is a permanent saving on a claim that no longer exists.

But look at what the company is doing with its capital. It is buying its own paper instead of the asset. That is what a fund does when its units trade below net asset value. It is not what a growth company does.

And the obligations are real. STRC.live puts issuer-wide preferred dividends and debt interest at $1.46 billion a year, against a USD reserve of $4.77 billion and $146.2 million of cash: about 39 months of coverage at a weekly burn near $28 million. Nobody is anywhere close to a wall. The direction of the flow has changed all the same: the company drew $21.1 million from that reserve during the week to pay preferred dividends, according to the same filing. Cash is coming out of the reserve rather than going in, and the ATM is being used to shrink the capital structure rather than to buy coins.

What the marginal dollar actually costs

Here the macro matters, and the Federal Reserve published the frame itself this week. Vice Chair for Supervision Michelle Bowman said on November 10 that the policy rate was held at 3.75% to 4%, that core PCE inflation likely rose to 3.3% over twelve months in October, that third-quarter growth probably exceeded 3%, and that progress on inflation has stalled for roughly two years. Tariffs, energy and AI-related demand each accounted for about a quarter of the above-target core inflation over the twelve months ending in September. (Federal Reserve, 10 November 2026)

Put the two numbers side by side, because almost nobody does. The risk-free policy rate is under 4%. Strategy pays 12% on STRC to raise preferred capital. That eight-point spread is the market’s price for financing a leveraged bitcoin treasury: not a rounding error, not a temporary dislocation, but the standing cost of the marginal dollar for this structure.

The convertibles were the other leg, and they were cheap: a weighted-average coupon in the fractions of a percent, because the buyers wanted the equity option rather than the interest. That trade made sense in a world where the equity traded at three times the coins and kept rising. It makes much less sense at 0.83x, where the option has become the whole instrument. The cheap leg was never a subsidy from the bond market. It was a bet on the premium, priced as a loan.

The wrapper, not the asset

Here is the part that separates two stories that get told as one. Over the last 120 trading sessions, bitcoin is down 2.0%. Strategy is down 23.5%. Same window, same asset underneath. (Own calculation: daily closes from Binance spot and from my own board, 22 May to 11 November 2026)

Bitcoin and Strategy share price, both rebased to 100 on 22 May 2026, through 11 November 2026

That is not a bitcoin statement. It is a leverage statement, and it is the arithmetic doing what arithmetic does: a levered claim on a flat asset loses value when the premium that justified the leverage evaporates. Anyone who bought the equity as a “bitcoin proxy” got the beta they asked for on the way up and the decay they did not model on the way down.

The same structure has already failed in its second-largest market. Bitmine, the ether treasury company, reported a $6.3 billion annual net loss as ether fell 44%, and ether traded at $2,045 this morning on my board. (The Block, 11 November 2026; own measurement, 12 November 2026, 09:31 CET) The bitcoin version is better collateral, which is not a small thing and is why the bitcoin vehicle is still standing while the ether one is a cautionary tale, but the wrapper mechanism is identical, and it does not care which coin is inside.

Meanwhile, the physical layer got thinner

While the financial layer argues about index rules, the part of the industry that actually produces blocks had a worse year than the price suggests. Gross issuance revenue to all miners is 450 bitcoin a day, 3.125 per block across 144 blocks. On 6 October 2025 it was $56.1 million a day, with a coin closing near $124,700. At this morning’s $73,433 it is $33.1 million a day, down 41%, while costs stay denominated in the currency that did not fall. (Own calculation on the daily close)

Gross daily issuance revenue to all miners in US dollars since December 2024

Canaan shows what that does to a manufacturer. The Avalon maker’s mining-rig revenue fell 96%, and it has launched a $20 million share repurchase while most of its bitcoin holdings were pledged or placed in a fixed-term product at the end of September, meaning the coins are working rather than sitting. (CoinDesk, 12 November 2026)

This is the point in the cycle where concentration risk quietly rises. When hashprice compresses, small miners sell hardware or hash to whoever will take it, and the pools that survive are the ones with scale. That is why the technical work on decentralised payouts matters more than its obscurity suggests. A discussion paper posted to Delving Bitcoin on November 11 lays out the problem with unusual clarity: a coinbase has a finite number of payout slots, and any rule that hands those slots to the top N contributors by work is a “rich list.” A large miner can game any fixed-slot rule by splitting hashrate across many identities, pushing small miners out of the block reward entirely. The paper walks through Chris Belcher’s 2017 payment-hub design, Braidpool’s rolling coinbase aggregation with a 66%+1 threshold signature over recent block finders, and P2Poolv2’s market for sharechain UTXOs, and shows why each one is exposed to some form of settlement griefing. (Delving Bitcoin, 11 November 2026)

Bitcoin’s censorship resistance is not guaranteed by the size of the hash rate. It is guaranteed by how many independent people can be paid without trusting an operator. That is a design problem, and it is being worked on, mostly by people who will never appear in an index.

The protocol layer, which behaved

The same week, Bitcoin Core published three patched security advisories. The lead one, CVE-2025-62861, rated High, is worth reading for what it says about a specific failure mode: during a chain reorganisation, a defect in the version-3 transaction sibling-eviction logic could leave a node connected, synced, gossiping happily, and unable to produce a block template until restarted. Affected versions were 28.0 through 28.2 and 29.0 through 29.1. The fix shipped in October 2025 in 28.3, 29.2 and 30.0; Core held disclosure for 396 days, until the last affected release line went end-of-life on October 27, 2026. No funds were at risk, and no exploitation has been reported. (TFTC, 11 November 2026; Bitcoin Core security advisories)

Nothing about that story is dramatic, and that is exactly the contrast worth drawing. The layer that secures the asset found a bug, told a small number of people, patched it, waited out the embargo, and published. The layer that wraps the asset in securities found a rule change, and the entire structure had to reprice. One of these systems has a process for bad news. The other has a press release.

The steel man, and where it fails

The bullish case deserves its strongest form, because it is not empty. The forced selling is 3% of the share count and less than a day’s volume. Strategy does not have to sell a single coin, and has not — the filing reports no bitcoin sales. Its convertibles are long-dated at coupons near zero, with no margin calls on the coins, and there is a $4.77 billion reserve in front of the preferred dividends. Meanwhile the institutional rails are still being built: Citigroup disclosed a new $119 million FBTC position and grew its IBIT share count by 50% in the third quarter. Banks as a category doubled their reported spot bitcoin ETF holdings from $107 million to $258 million, and Deutsche Bank went live with crypto custody for corporate clients under a MiCA notification that passports across thirty countries. (TFTC, 11 November 2026; Unchained, 11 November 2026; ESMA interim register via Unchained)

All of that is true, and none of it addresses the premium. A treasury company at 0.83x equity-to-coins does not need a better narrative about institutional adoption; it needs a reason for someone to pay more for the shares than for the coins. Losing index eligibility does not just remove buyers for one day in November. It removes the category of buyer that never asked the price question at all. It also closes the door on being added later.

A sharper version of the bear case exists too, and it is not about Strategy’s balance sheet. The entire “institutional adoption” story leans on the treasury-company complex as its proof, because that complex was the visible, listed, auditable form of institutional demand. Reclassify it as a fund, price the fund at net asset value, and the proof evaporates with the premium. Derin Olenik, an independent analyst quoted by CoinDesk, called the exclusion “good riddance” and predicted it would be a major blow to that narrative. He and Thielen are both probably right, which is the uncomfortable part: the price impact can be small and the narrative damage large.

What I do not know

I cannot verify whether FTSE Russell or S&P Dow Jones will adopt a similar screen. JPMorgan’s $8.8 billion figure is conditional on exactly that, and it is a year old and built on a proposal MSCI did not adopt. I do not have a clean, current breakdown of the senior claims, preferred plus convertibles by instrument, which is why I derived the enterprise view from the two published multiples rather than adding the pieces myself.

And I cannot separate this decision from the rest of the tape. Bitcoin has spent fourteen months between $57,800 and $126,211, and the drawdown began long before an index committee met. Brent crude is $93.47 and West Texas Intermediate $87.21 on my board this morning, against the Fed’s own note that Brent ran above $110 in the second half of October as the Strait of Hormuz, roughly a fifth of world oil supply before the conflict, was disrupted. Core inflation at 3.3% with policy at 3.75-4% is a real rate of about half a point: not restrictive, not accommodating, and certainly not the free money that built the treasury-company complex between 2020 and 2024. (Own measurement, 12 November 2026, 09:31 CET; Federal Reserve, 10 November 2026)

Anyone who tells you the November weakness is purely about MSCI is selling something. Anyone who tells you MSCI is irrelevant is selling the same thing.

What actually changed

Twenty-one million coins. The same issuance schedule, the same difficulty adjustment, the same signature scheme, the same node software with one fewer bug in it. December 1 changes nothing about any of that.

What it changes is who the marginal buyer is. For four years the marginal buyer of bitcoin through public equity was an index fund that had been instructed to buy, financed by a convertible market that priced the option on a premium, inside a structure that turned rising share prices into coin purchases. Remove one leg, the instructed buyer, and the other two lose their reason to exist at that price. This is not a conspiracy and it is not a crash. It is price discovery, which is the process of finding out what the coins are worth to people who are not obliged to buy them.

The Austrian reading is unromantic: the boom did not need fraud, it needed a rule. Index inclusion made a fund look like a growth company; near-zero coupons made the acquisition of a scarce asset look like financial engineering; and both were artefacts of a monetary regime that punished saving and rewarded leverage. When the artefact is withdrawn, the structure reprices toward the only thing that was ever real: the asset, and the time preference of the people holding it.

Strategy will very likely still be here in a year, with most of its coins and a smaller, cheaper capital structure. That is the good outcome, and it is the one the company is actively buying: retiring 12% claims at 97 cents on the dollar is not the behaviour of a firm in distress. But it is the behaviour of a firm that has stopped being a growth story, which is what the index rule said out loud.

The coins do not care. That was always the point of them.

Sources

Write a comment