The Question They Couldn’t Answer

01 / The concession

Ask the people who built the metric what it is, and they will tell you it depends.

Onstage at a Bitcoin conference this June, the founder of the company that created the entire category was asked to define mNAV, the multiple every Bitcoin treasury company is now valued on. He said, “I don’t think there’s a single metric even though there is some that are useful.” The chief executive of the second issuer to follow him echoed it from the same stage, saying “there’s no single metric that will tell a full story and we shouldn’t expect it.”

That is the valuation metric. The per-share metric fares no better, and the sharpest admission of that is the company’s own filing. Strategy’s Q1 8-K, signed under penalty of law, states that buying Bitcoin with preferred stock or out-of-the-money convertibles raises Bitcoin per share while increasing the senior claims ranked ahead of common “in a manner that is not reflected in these metrics.” The same filing says those metrics “might overstate or understate the accretive nature of the Company’s use of capital to buy bitcoin.”

Read those admissions in a row. The valuation metric is not singular. The per-share metric does not reflect the senior claims stacked in front of common. The filing itself says the figures may mislead. None of that is a critic’s claim, and this part of the series does not need to make one. It is the toolkit’s own architects, on the record, describing the toolkit.

02 / The question that earned a mocking

Every one of those concessions is an answer, and every answer was pulled out of someone who builds these instruments because one man kept asking the same question in public and would not stop.

Jack Mallers runs Strike and Twenty One, one of the largest corporate Bitcoin treasuries in the world. In recent weeks he started asking, on his show and from conference panels, a question that sounds like it should be simple. What is mNAV? Not as a trap. As a man who could not get a straight answer. “What is mNAV? I don’t know,” he told his audience after weeks of trying. “All of these metrics were invented out of thin air,” he said. “These were all invented by the company themselves.” Later he added, “these variables are being redefined in real time. Nobody knows the definition of these things.”

For asking, he was called a nepo baby, a performative financial elite, a competitor throwing shade, a child. A clip of him putting the question to Michael Saylor in person was cut and recirculated as proof he did not understand his own industry.

Here is what is strange about that. The discourse had a slot ready for his question, and the slot was labeled either ignorance or motive. A chief executive asking what a metric means must not understand finance, or must be attacking a rival. There was no slot for the third thing, the true thing, which is that a measure can be so unfixed that a sincere and sophisticated person cannot get a stable answer out of it, and says so out loud.

Part 2 of this series described the same trap from the other side of the room. The investigator Coffeezilla looked at a preferred stack, felt the math was wrong, and reached for the only word the discourse had handed him, “Ponzi-adjacent,” because no one had given him a measurement vocabulary. He had the right instinct and the wrong word. The outsider reached for criminality because measurement was missing. The insider reached for the plain question, what is this, and was told he must be a fool or a fraud, for the identical reason. The vocabulary that would let the question be asked and answered does not live inside the toolkit.

The question was never the problem. The absence of an answer was.

03 / What he was right about

Strip the personality away and look at what Mallers was saying, because underneath the naive question is a precise one.

A convertible bond is a hybrid. It is debt that can turn into equity if the stock climbs to the conversion price. While the stock sits far below that price, the bond is not going to convert, and no one valuing the company would treat it as equity. It is debt. Count it as equity anyway, and you inflate the equity value, which makes the mNAV multiple look better than the company earned.

Mallers walked this through on his own company’s bond, repeatedly, in public. Twenty One raised a convertible that converts at $13.88. The stock trades near $5. “That would be classified as debt,” he said, “cuz it’s not converting into equity anytime soon.” To Saylor’s face at the same event he put it the same way, “no one would assume that as equity. It would be classified as debt.”

He was not inventing a standard. He was pointing at the oldest one. “Enterprise value is easy to calculate and universally understood,” he said. “Creating a lot of different metrics makes things complex and very difficult to understand.” Enterprise value has classified an out-of-the-money convert as debt for decades. The measuring problem is not that the answer is unknown. It is that a settled answer was retired and five unsettled ones were issued in its place.

One move ends the “he is just sniping at a competitor” reading for good. He applied the rule to himself. Asked what his own company’s mNAV was, he said, “if someone asked what’s 21’s MNAV, I would not include that convert. I would think that that would be not right.” He held his own balance sheet to the stricter standard, and his own number came out worse for the honesty. He did it without claiming to be certain. “This is me killing ego entirely,” he said. “I could be missing something, uneducated, the dumbest person, and I’m totally open to that.”

A man sniping at a rival does not hold his own filings to the harder rule and admit he might be wrong. A man asking for a standard does.

04 / Why the question couldn’t be heard

The reason a sincere question read as an attack is structural, and the fairest way to show it is to let the toolkit’s architect make his strongest case, because he has one.

Michael Saylor’s argument is coherent and it is held by a serious person who built a category from nothing and outperformed Bitcoin itself for five years doing it. His keynote this June laid out an entire worldview, Bitcoin capitalism, in which his instruments are products with precise jobs. “MSTR is amplified Bitcoin,” he told the room. “High volatility, high amplification.” STRC is “short duration, high yield, fixed income.” On dilution, his case is direct and not obviously wrong. Selling equity above net asset value to buy Bitcoin is not dilution, it is expansion of the capital structure, and the assets per share hold. On the perpetual preferred, he argues it is not even a liability in any state the company will realistically reach, only “a liability in liquidation, but actually an asset in operation. That’s why it’s a hybrid.” Each of these is a real position, seriously argued.

Now ask him for the number. Asked, at that same event, what a shareholder should use to value the company, his answer was not a number. It was a method. You have to form your own opinion of Bitcoin’s forward price and forward volatility. You build a full model of the company’s assets and liabilities and the cost and duration of its capital. The model spits out a fair value. “The MNAV calculation isn’t the only calculation,” he said. “I don’t think there’s a single metric.”

That exchange is the whole problem in miniature. The answer to “what is mNAV” was “build your own.” A measure that returns a different number for every set of private assumptions is not a measure. It is a position wearing the costume of a measure. And a question with no fixed answer cannot be sincerely asked in public, because the audience cannot tell the difference between a person who does not understand the answer and a person who has noticed there is not one.

This is what Part 2 called the captured insider, the analyst who calls a single coherent metric too complex while fluently juggling three contradictory ones. The capture is not stupidity. It is fluency. Spend enough years translating between five mNAVs and the translation starts to feel like expertise. The person who asks why there are five starts to look like the one who is behind.

The toolkit did not fail to answer the question. It is built so the question has no answer.

05 / The cost is real, and someone pays it

There is a reason the answer keeps coming back as “it depends,” and the reason is that a real cost is being moved around in a place the metric cannot see it.

Mallers framed it as a problem he could not resolve, and the framing was exactly right. A company like Strategy now answers to four constituencies at once: Bitcoin holders, common shareholders, preferred holders, and debt holders. When the Bitcoin is underwater and the perpetual preferred demands its dividend anyway, the company has to raise cash, and every route takes from one group to pay another. Sell Bitcoin, and the Bitcoiners absorb it. Sell common stock below value, and the common shareholders absorb it. Defer the preferred dividend, and the preferred holders absorb it. “Who’s burdening the cost?” he kept asking.

He answered his own question without the framework to name it, and in answering it he reached, unprompted, for the exact words this site has used since January. New shares, he said, inherit a permanent share of the dividend obligation, and “we call that a drag. That drag has a cost.” The preferred, he said, is “a perpetual tax on every bitcoin in the pile.”

Drag. A tax on every Bitcoin. He had the concept and grabbed for the word because the toolkit gave him no instrument to hold it in. The toolkit has its own word for that same number, and the word is Amplification, and it points the opposite way. Amplification sells the senior claims as upside. Drag names them as cost. Same claims, same balance sheet, opposite sign. That is the contradiction Part 1 documented, surfacing now from the mouth of a chief executive who reached for the cost framing on instinct.

He is not alone in the conclusion, only in the volume. NYDIG, an institutional research desk, wrote that “financing decisions are being driven by the wrong metric.” The analyst Jeff Park wrote a year earlier that BPS and mNAV, taken together, are “intellectually inconsistent.” Strategy’s own disclosure now states that issuing equity is accretive to Bitcoin per share only above roughly 1.22 times mNAV, not the 1.0 the market assumes, and the company attributes the higher bar to its growing preferred stack. The gap between those two numbers is the drag, priced. The cost is real. It has a number. The toolkit’s job is to keep that number from acquiring a name.

Call it amplification and it is a feature. Call it drag and it is a bill. It is the same number, and someone pays it.

06 / The framework finishes the question

There is an answer to “what is mNAV.” It has existed since day one. It is one number, it does not move when your assumptions move, and it is the answer Mallers was reaching for and the toolkit’s builders declined to give.

CEBE. Common Equity Bitcoin Exposure.

Total Bitcoin, minus every senior claim translated into Bitcoin, divided by the basic shares common holders actually own. Its valuation companion, CEBE mNAV, measures market cap against that same net reserve. Because the claims are subtracted out of the asset rather than added into a numerator, the per-share number and the valuation number cannot contradict each other. There is no second definition. There is no forward-assumption dial to turn. The accretive threshold is 1.0, and nothing is hidden above it.

Run it on the question that started all of this. What is mNAV? On CEBE’s definition it is market capitalization measured against the Bitcoin that belongs to common equity after the company settles what ranks ahead of it. The out-of-the-money convert Mallers flagged is debt, subtracted, exactly as he said it should be. The perpetual preferred is a claim, subtracted, exactly as it behaves. The number does not depend on your view of Bitcoin’s forward volatility. It is true on the filing, today.

That is the difference between a measure and a position. A measure returns the same answer no matter who is asking. The toolkit could not give Mallers a stable answer because it was never built to return one. The framework returns one because that is the only thing it was built to do.

The strongest evidence that this is the right construction is that the company’s own people reach for it the moment they are asked to be precise. Pressed in public on the gap between Bitcoin per share and mNAV, Chaitanya Jain, who leads Bitcoin product and investor strategy at Strategy, walked through what he called Net Assets per Share, Bitcoin assets less all senior claims to common equity, the net debt and the preferred, divided by basic shares, with a breakeven at one times mNAV. That is CEBE, line for line, under a different name. He went further and granted that the clean metric is buildable, that you can construct a version of mNAV that maps one to one to per-share accretion and breaks even at 1.0, and the only reasons he offered for not leading with it were simplicity and not wanting to introduce a new measure. He declined to call any of this a concession, and in good faith it need not be. Asked to be exact, the people inside the building describe the same number as the people outside it. They have only chosen not to print it on the front page.

He asked what the number was. This is the number.

07 / The cleanest proof is his own company

The clearest place to watch the framework answer the question is the company run by the man who asked it, because Twenty One’s entire capital structure is the single instrument he kept describing. One out-of-the-money convertible. No preferred. No perpetual obligation. Nothing else standing in front of common.

And even that clean company is misread by the toolkit, in the most literal way the measuring problem can take. Pull up Twenty One’s mNAV today and you get one of two answers, depending only on which shares a data vendor decides to count. Twenty One has two share classes. Class A carries the economics, the dividend and the liquidation right, 346,548,153 shares. Class B carries votes and nothing else, no dividend and no claim on liquidation, set out in the company’s own Certificate of Formation. It is held by the controller, and it owns no Bitcoin.

Count Class A alone, and Twenty One trades at roughly 0.84 times the Bitcoin its common shareholders own. A discount. Count the non-economic Class B as if it were stock, as several vendors do, and the same formula on the same afternoon prints a premium. One screen says the stock is cheap to its Bitcoin. The next says it is rich. Both call the number mNAV, and the entire gap between them is a vote being counted as a share.

That is the measuring problem compressed into one company, and it dissolves the instant you apply the rule. CEBE measures on the Class A base, the same base Twenty One’s own reported Bitcoin per share already uses, and nets the one convert. The convert is debt, because it converts at $13.88 while the stock sits near $5, the exact out-of-the-money security Mallers described on every stage he stood on. The answer comes out to 0.84 times. A discount. On the filing. With no assumption required.

Notice what the answer does not need. It does not need a grudge against the company. It does not need a conspiracy about the vendors. The premium some screens display is not a lie anyone told. It is an undefined metric meeting a non-economic share, which is the thesis of this entire series appearing one more time in its cleanest possible form. The man asked what his company’s mNAV was and said he would not count the convert. Compute it the way he asked, on the economic shares with the convert as debt, and the answer is a discount, the opposite of the premium the vendor screens advertise.

The question was the right one. This is the answer to it.

08 / The standard

Part 1 named the contradiction, three metrics that disagree across every line of the same balance sheet. Part 2 asked why it is tolerated, and found that Bitcoiners who would never trust an unverified ledger had learned to trust unverified, contradictory numbers for the companies that hold their Bitcoin.

This part is the answer to a question Part 2 left implicit. What happens when someone refuses to tolerate it? They are told they are stupid, or that they are lying, or that they should ask their questions privately at dinner. And then, slowly, in public, the people who built the measure concede that the questioner was right. The metric is not singular. The two numbers do not map. The filing says they might mislead. Build your own model.

None of this means the companies are dishonest. They are not. Every metric they publish is valid in isolation, and the architect of the category is a serious person who built something real and durable. The failure is not moral. It is structural, and it was inherited whole from the fiat system Bitcoin was built to leave, the system where four answers to the same question is called sophistication, where you learn which number to read for which purpose, and where you stop noticing that you are doing it.

Bitcoin fixed the measuring problem for the base layer. One ledger. One truth. One set of rules. Verifiable by anyone, with no authority deciding what the number means. The companies that hold Bitcoin can meet the same standard. The question a chief executive got mocked for asking has an answer, and it has had one all along.

Finish the math, and the measure holds.

cebetracker.io

Part 1 of The Measuring Problem

The Measuring Problem

Why every Bitcoin Treasury Company reports three incompatible metrics for the same balance sheet, and what it costs the people who hold them. Strategy’s 7.7pp gap. Metaplanet’s opposite-direction quarter. The toolkit that always has a flattering number among its outputs.

Part 2 of The Measuring Problem

Why We Tolerate Contradiction

Why Bitcoiners tolerate three incompatible metrics for the same balance sheet. The captured insider, the TradFi inheritance, and why both bulls and bears win when the math is complete.