The Buyback Buys the Ratio, Not the Coin
A Bitcoin treasury company carries two numbers that look like the same number and are not. BTC per share is the coin stack divided by the share count. NAV per share is the coin stack minus every senior claim, preferred stock and debt, divided by the share count. The first measures how much coin sits behind the structure. The second measures how much of it reaches a common holder.
On 7 Oct, Grain of Salt SF published a thought experiment. Issue $42bn of STRC preferred at par, use $21bn to buy back MSTR common, hold the other $21bn as cash, and stop buying bitcoin. He asked whether that improves common-shareholder total return more than buying more coin. His arithmetic is exact. The answer turns on a single price: what the buyback pays against what a share is worth.
His arithmetic holds. The tell is what does not move.
Grain of Salt SF published the scenario on 7 Oct, and every line of it reconciles: issue $42bn of STRC preferred at par, repurchase $21bn of MSTR common at an average of $330, hold the remaining $21bn as cash, and buy no further bitcoin, which retires 63.6 million shares against a starting count of 422.1 million.
The share count falls 15.1%, and BTC per share rises 17.8% at every bitcoin price he tested, whether the coin is marked at $100,000, $125,000 or $150,000.
The BTC-per-share bar is also the same height at all three prices. It rises because shares were retired, not because more coin was bought. The metric is indifferent to what bitcoin does, because no new bitcoin enters it.
The two metrics move in opposite directions
NAV per share is the second number, and it is the one that reaches a holder: the bitcoin stack minus every senior claim, preferred stock and debt, divided by the share count, which is a different quantity from the coin stack alone because the superior claims are paid before the common sees anything.
The repurchase pushes it down at all three of his own prices.
$166.35 to $137.29 at $100,000 coin, a fall of 17.5%. $216.57 to $196.43 at $125,000, down 9.3%. $266.80 to $255.58 at $150,000, down 4.2%.
A corporate action that raises BTC per share while cutting NAV per share is not a contradiction, because each dollar spent retires only between $0.50 and $0.81 of net asset value.
The mechanism, and the one price that decides it
The rule is arithmetic and it has no exceptions: a repurchase adds NAV per share only when the average purchase price sits below net NAV per share, so the same transaction can create value for holders at one stock price and destroy it at another, with nothing changed but the mark.
GoS buys at an average of $330.
Because that price is fixed while net NAV per share ranges from $166 to $267 across the three bitcoin endpoints, the transaction buys the stock above the coin it represents at every price he tested.
At $125,000 coin the buyer pays $330 for $216.57 of net asset value. At $150,000, $330 for $266.80. The buyback breaks even at a bitcoin price of $181,461, the level at which NAV per share equals the purchase price. Below that, the repurchase is a coin sale settled in paper.
The same $42bn, put into coin
The comparison the scenario sets up is not buyback against nothing but buyback against the purchase the company is being asked to stop making, and that is the harder test, because the two uses of the same $42bn differ only in what they buy: coin, or the company's own shares.
$42bn into bitcoin at an average of $93,000 adds 451,613 coins. At $105,500 it adds 398,104. At $118,000 it adds 355,932.
BTC per share lands 20% to 30% above the buyback path, and NAV per share is higher at every endpoint: $173.83 against $137.29 at $100,000 coin, $293.78 against $255.58 at $150,000.
The buyback is dominated.
What the scenario must assume to reach $500
The $160-to-$500 path is the re-rating the buyback is meant to justify. It enters the model as an input.
To hold $500, cash-adjusted mNAV must re-rate from 0.93× at the 7 Oct mark to between 1.69× and 2.53×, which is the terminal multiple doing the work rather than any operating improvement, and the $21bn cash reserve that supports it carries a cost of its own.
STRC pays an assumed 12% while the cash earns roughly 4%, a spread of about $1.7bn a year against a reserve that covers 3.17 years of the $6.621bn combined preferred and interest service.
None of those inputs is unreasonable on its own. Together they are the answer rather than the evidence.
Strategy's own balance sheet runs the other way
The company the scenario rewrites is already acting on the distinction. At a gross mNAV of 0.93× on 7 Oct, MSTR common trades below the coin it holds.
In a discount regime an issuer retires senior claims below par rather than issuing common above net asset value, because each retired claim is bought for less than the assets behind it, and that is the direction the repurchase line in the weekly 8-K filings has taken when it is read against the 7 Oct marks.
A buyback can be the right trade. It is the opposite trade at a discount than at a premium, and the sign of the gap decides which one it is.
The counter-argument, stated fairly
There is a real case for the buyback, and it is not the one the share-count metric makes.
BTC per share is the quantity many convertible and preferred instruments are indexed to, and a smaller float can re-rate the multiple if demand for the structure holds, so the mechanism GoS points at is not imaginary even where the arithmetic on NAV per share runs the other way.
GoS framed the exercise as a question rather than a recommendation, and the question is the right one. It resolves against the buyback on his own numbers.
A structure that can sell preferred at par and retire common is a genuine option. It is worth exercising when the common is cheap against coin, not when it is expensive.
What would change the read
The conclusion inverts when the common trades below net NAV per share. At $86,000 coin, net NAV per share is $138.22 against a stock at $160, so the common sits 15.8% above the coin behind it.
That gap, not the headline mNAV, is the diagnostic, because a buyback is accretive when the stock trades below the net asset value it retires and dilutive when it trades above, and on the 7 Oct marks the common trades above net asset value at every bitcoin price in the scenario.
Digital Credit Is a Transmission Belt — where preferred issuance sits in the funding stack
Bitcoin Was Never a Rates Trade — the complex priced through the coin rather than the structure