Hook
Bitcoin is flat. Bitcoin treasury companies are not.
Over the eighteen months ending in late 2025, BTC returned less than 3%. The median drawdown across listed digital asset treasury companies — DATs — was -52%. Nakamoto fell 98%. Fold Holdings -91%. Exodus Movement -83%. Semler Scientific bled into double digits. And Strategy, the company that invented this trade, lost roughly half its market value inside a single year while holding more bitcoin than at any point in its history.
This is not volatility. Volatility is symmetric noise around a mean. This is a mechanism failing in one direction only. When the underlying asset returns +3% and its listed proxies return -52% at the median, the proxies are not tracking the asset. They are tracking something else — and that something stopped paying.
I audited the void and found a backdoor. The backdoor has a ticker: mNAV.
Context
In August 2020, Michael Saylor moved $250 million of Strategy's balance sheet into bitcoin. At the time the move was genuinely novel. A public software company was converting a stagnant cash reserve into a hard-capped monetary asset, and the market rewarded it with a valuation multiple that grew to dwarf the coins themselves.
What followed was not a bitcoin strategy. It was a capital structure arbitrage, dressed in orange. The model needs exactly two variables to function: the price of BTC, and the ratio of the company's market capitalization to the net asset value of the coins it holds. That ratio is mNAV.
When mNAV is above 1, the company trades above the value of its treasury. Every share issued buys more BTC than the share dilutes. Bitcoin-per-share rises without purchasing a single satoshi from operating cash. This is accretive issuance, and when the premium is wide it looks like a perpetual motion machine. When mNAV falls below 1, the same gearbox runs in reverse. Each issuance destroys shareholder value. Each refinancing converts dilution into decay.
By mid-2025, roughly 200 listed companies had copied the structure — from Tokyo to Oslo to Hong Kong. Strategy had six years of operating history. Most of the imitators had less than eighteen months. The window in which the original trade worked depended on two conditions that no longer exist: scarcity of the narrative, and the absence of a cheaper substitute. Both conditions expired. The market has spent the last year repricing that fact.
Core
The data is unambiguous, and it separates cleanly into two columns.
In the winner column: Aker ASA returned +181%, with bitcoin representing 0.4% of total assets. Galaxy Digital returned +114%, as a diversified crypto financial services operation. Tesla returned +59.8%, as an electric vehicle manufacturer with a bitcoin position attached. In the loser column: Strategy, down roughly half over one year. Fold Holdings, -91%. Exodus Movement, -83%. Nakamoto, -98%.
Look at what the winners have in common: not one of them is a treasury company. Aker is an industrial conglomerate. Galaxy is a trading and lending desk. Tesla builds cars. They outperformed despite their coins, not because of them. The market has been misreading this as evidence that the DAT thesis survives. It proves the opposite — the return came from operations, and the bitcoin was dead weight.
Two names still print green when measured from announcement: Metaplanet at +476% and Boyaa Interactive at +456%. I want to be precise here, because precision is the only defense against narrative. Those figures are a timestamp, not an edge. Measure the same stocks from their local highs and the sign flips. Choosing the announcement date as the reference point is a framing decision, not a discovery.
Here is where the mechanism actually broke, laid out as a sequence.
First, premium compression. Investors were paying an ever-shrinking premium over net asset value, and eventually a discount. The flywheel requires a marginal buyer willing to pay above NAV. That buyer left. When I spent two months in 2020 reverse-engineering Curve's stableswap invariant to find where the math misbehaved under volatility, the lesson transferred directly: systems fail at their boundaries, and the boundary of a premium flywheel is the moment the premium inverts.
Second, saturation. A scarcity premium copied 200 times is no longer scarce. The premium was a shared resource, and each new entrant diluted the dilution itself.
Third, and most decisive, ETF substitution. Spot bitcoin ETFs arrived in 2024 and offered purer, cheaper, more regulated exposure inside the same brokerage account. The original DAT pitch — you want bitcoin exposure but you can only access it through equities — was solved by a lower-cost instrument. This is the competitor the DAT sector never discusses on earnings calls, because it cannot beat it. The golden window of 2020 to 2021 was, structurally, the ETF vacuum. Once the vacuum filled, the premium logic was dismantled.
Now the leverage. DATs financed coin purchases with convertible notes — low coupon, high conversion premium. These are debt. When mNAV sits below 1 at maturity, the issuer faces three options: refinance on punitive terms, sell bitcoin, or dilute at a discount. That is not a business risk. That is technical debt in the literal sense, and it matures on a calendar.
The asymmetry tells the whole story. BTC +3% produced a median DAT return of -52%. That is not beta. That is negative convexity — capped upside through premium decay, uncapped downside through leverage. My 2024 ETF basis work trained me to look for exactly this kind of skew: a structure that pays a little when the reference asset rises and loses a lot when narrative fades. The equal-weight portfolio of DAT names is negative over the window. Diversification does not help, because the common factor is itself negative.
Contrarian
The comfortable read is that the market got scared and DATs will recover. That is survivorship bias wearing a suit.
Every performance table in this sector is a selection function. Metaplanet's +476% and a -60% drawdown describe the same equity, separated only by the entry date the analyst chose. Neither number is a strategy. Both are timestamps. And the most dangerous artifact of this bias is that the top three performers — Aker, Galaxy, Tesla — are being cited as proof the DAT thesis still works, when none of them is a treasury company at all. If you defend the trade with a diversified industrial conglomerate, you have already conceded the argument.
The governance layer compounds the problem. Nakamoto paid its executive chairman handsomely while the stock fell 98% — a textbook decoupling of management compensation from shareholder outcome. An announcement-driven pump followed by a rapid fade, replicated across dozens of tickers, is not a market. It is a disclosure pattern that regulators will eventually price in.
And the sector's retreat does not meaningfully damage the underlying network. DATs build no protocol, deploy no developers, settle no on-chain activity. They absorb sentiment and emit equity volatility. Their failure is a capital markets event, contained to retail shareholders — not a crypto infrastructure event.
Takeaway
The trade is not "short DAT." The trade is understanding what resets. Watch three things: the mNAV of the surviving structure, the convertible maturity wall at the largest holder, and whether any single treasury company is forced to sell size into spot. Floor sweeps are just data points in motion, but forced liquidations are not data points — they are the exit door opening on a crowded room.
If the flywheel only turns on a premium that no longer exists, the honest question is not whether DATs recover. It is who holds the bitcoin when the refinancing stops.
