Strategy (MSTR) and MSCI's Index Exclusion Rule: The Exit Started a Year Ago

In November 2025, J.P. Morgan put a number on what it would cost Strategy to lose its place in MSCI's indexes: about $2.8 billion of selling by funds that track them, and another $8.8 billion if other index providers followed. On October 15, 2026, MSCI made that exit likely. It adopted a test that makes "non-operating companies" ineligible for its global indexes, and when it ran the test on May 2026 data, Strategy was the largest of the three companies it deleted. MSCI publishes the final list on November 11. MSTR was down about 3% before the open the next morning.

The headline version of this is forced selling. I don't think that's the important part, because the forecast is eleven months old and the selling it describes has been priced in installments since. S&P passed Strategy over a year ago, MSCI stopped buying its new shares last October, and by this summer the stock traded below the value of its bitcoin. Thursday's rule is the paperwork for a reclassification the index committees started and the market finished. What the episode shows more clearly is where the durable advantage in this trade sits, and it isn't with the bitcoin treasury.

Strategy's edge was never the bitcoin itself, since anyone can buy that. It was the ability to sell its own shares for more than the bitcoin behind them and turn the difference into more bitcoin per share. That only works while someone pays a premium. Part of what paid it was demand from investors who own stocks through indexes and mandates rather than coins, and index funds that had to buy every new share the index counted.

That second source started closing a year ago, with MSCI's freeze. It wasn't the only cause, since Nasdaq-100 and other trackers kept absorbing new shares and bitcoin's own fall did plenty of damage. But by June the stock traded below the value of its coin on the plain measure and Strategy was selling bitcoin to pay its preferred dividends. A rule written now can't take away a premium the market had already stopped paying.

The ratios read like a description of a bitcoin treasury company, and I think that's deliberate. Strategy says so in its consultation response, where it calls the screen a pretext and argues that "operating" and "non-operating" have no basis in US GAAP or IFRS. It also says it reports its bitcoin business as an operating segment. MSCI kept the screen anyway. It did drop the blunter interim rule from October 2025, which caught any company whose digital assets were half or more of total assets, so the test is now about how a company makes its money rather than what it holds.

I've put Strategy next to the two companies that own the most important labels in passive investing, MSCI and S&P Global, whose index arm runs the S&P 500. I've also added the two companies in the cohort that give investors crypto exposure through an operating business, Coinbase and Robinhood. Robinhood is a useful control. It joined the S&P 500 in the same rebalance that skipped Strategy, and its revenue moves with crypto trading.

The scorecard isn't built to answer this, and valuation pushes these names around more than any index decision. The moat matrix is where the split shows. I've used the four pillars that decide whether a company can set terms for others or has to accept them.

The three groups look nothing alike. The matrix already shows Strategy's wrapper pillars eroded, and it has nothing to rate on system of record or transaction embedding, because it doesn't sit inside anyone's workflow. The regulatory pillar there was the listed wrapper, and the October 15 rule is aimed at exactly that. Coinbase and Robinhood read intact on most of these and Coinbase strong on transaction embedding, which is what a business looks like when it earns fees from activity and doesn't need a committee's blessing to exist.

MSCI and S&P Global read strong on all three of regulatory lock-in, network effects and system of record. The MSCI analysis describes the network effect as the more assets benchmarked to an index, the more a rebalance moves markets, and the more managers have to track it. This week is that loop in action. One methodology change, published on a Thursday evening, decides whether billions of dollars must own a stock, and the company on the other end could only send a letter.

• Index buying of new shares. A company that issues stock to buy assets gets a built-in buyer while the index raises its share count. MSCI stopped doing that for Strategy in October 2025, which made each new share a little harder to place at a premium.
• Mandates that follow the index. Pension and fund guidelines name MSCI benchmarks, so a stock outside them is outside a lot of portfolios by default. That's the same mandate language the MSCI analysis credits for its own lock-in.
• The credibility of being called a company. An operating company gets valued on what it might earn; a fund gets valued on what it holds, usually at a discount. CoinDesk's closed-end fund comparison in June described that switch, and the plain market cap to bitcoin ratio was 0.71 in August and about 0.84 at the October 15 close.

I don't want to wave the flow away, because there are real ways this hurts Strategy more than I've allowed. The first is reverse causality, and it's the strongest version of the other side. If the discount exists because index buyers were frozen out and an exit was telegraphed, then the forced-selling story was right all along and just arrived early. I mostly accept that. It's why I don't think November 30 is a fresh shock. If the exit is in the price, the effective date should pass without the stock falling much harder against its bitcoin than it already has, and the falsifiable claim below tests exactly that.

The second is contagion. J.P. Morgan's $8.8 billion was for other providers following MSCI, and MSTR is in the Nasdaq-100, which sets its own rules. I haven't found any sign that Nasdaq, FTSE Russell or CRSP are considering a similar test. But MSCI has now written down a public test with thresholds, which gives every other committee something to point to, and if Nasdaq's December reconstitution used it the selling would be several times Thursday's.

The third is reflexive. Strategy declared $508 million of preferred dividends in the second quarter alone, on top of interest on $6.7 billion of convertible notes. With the stock below the value of its coin, it can't sell common shares to cover that without diluting holders, which is why it sold bitcoin from the second quarter into August. The $4.88 billion reserve covers a bit over two years at that dividend rate. A lower share price brings the next round of bitcoin sales closer, and sales by the largest corporate holder are bad for the price of the thing it holds. Index selling doesn't need to be large to start that loop if it arrives at a weak moment for bitcoin.

The fourth is Robinhood, which is the cohort's honest counter-example. It rose about 15% the day S&P said it would join the index, so labels clearly move prices. I'd answer that the move came on the announcement, not on the effective date, and that's the point. Strategy's exit has been announced in installments for thirteen months. What I can't rule out is that MSCI's list still contains news, for instance if it also catches names the simulation didn't.

The last is Strategy's own argument, which cuts against my point about the gatekeepers. Strategy says MSCI has moved from measuring markets to making policy calls about which business models count. If big index clients agree, the thing MSCI is selling, a neutral and rules-based benchmark, gets weaker, and a rule that seems designed around one kind of company invites competitors to market themselves as the neutral alternative. I don't think a dispute over a handful of bitcoin treasuries moves the mandates that name MSCI benchmarks. I'd take it seriously if the same approach spread to other sectors.

For Strategy, I'd treat the November 11 list as close to settled and mostly priced. MSCI-linked funds hold 3.1% of its basic shares by Strategy's own count, about 13 million shares, which it says is less than a day's trading volume. At the October 15 close that's about $2.0 billion. What I'd watch is the market cap to bitcoin ratio through the November 30 effective date. If it holds, the index layer and the market had already done MSCI's job. If it breaks sharply, the flow mattered more than I think.

For MSCI and S&P Global, this is what a strong regulatory and system-of-record moat looks like when it's used. Neither company needs crypto to grow for this to matter. An index provider can redefine eligibility, give the affected companies a consultation and a deadline, and have the market adjust around it. That's a better illustration of the pillars than any quarterly revenue line.

Next time a stock's case rests on a category, whether that's index membership, an investment-grade rating or a spot on an approved list, I'll check who writes the rules for that category before I pay for it. This week Strategy was on one side of that question and MSCI was on the other.