Graveyard / Funds and structure / Binance perps
Bitcoin treasury companies trading below NAV
Can you profit when a company that holds bitcoin or ether trades below the value of its coins?
The idea
Some listed companies hold large amounts of bitcoin, ether or solana. When the stock trades below the value of the coins (mNAV below 1), buy the stock, short the coin with a perpetual future, and wait for the gap to close.
How we tested it
| Data | Company filings (SEC XBRL), Yahoo prices and Binance perpetual data for 11 US-listed companies that hold a main crypto asset. mNAV = (market cap + debt + preferred − cash) divided by (coins × spot price). Plan: enter at month-end when mNAV is 0.9 or lower, hedge with a short coin perpetual at a 90-day beta, exit at mNAV 1.0 or after 90 trading days. |
|---|---|
| Period | January 2024 to November 2026. |
| Costs | 60 bp round trip on the stock and 15 bp on the perpetual, plus funding. Stress case 90 bp and 22.5 bp. |
| Holdout | Development entries 2024-2025, holdout entries 2026. At least 8 episodes were required, else the result is declared insufficient. |
| Control | A random twin: a random company and month-end with mNAV between 0.9 and 1.1, same holding length, 5,000 draws. It was never run. |
Result
| Episodes with mNAV at or below 0.9 | 4 (8 required) |
|---|---|
| Episodes in the 2024-2025 development window | 0 |
| Episodes confirmed by hand | 0 of 4 |
| Companies that produced an mNAV table | 7 of 11 (4 with unit errors) |
| Returns computed | none |
Why it failed
The test could not be run with free data. Filing data gives only 0 to 7 holdings points per company, often filed late, with inconsistent unit labels, missing share counts and debt or preferred figures that are not the liquidation preference. Weekly holdings announcements would have to be read by hand. Nothing was computed, so this is not a negative result about the idea, only about the cost of testing it.
What would make it worth retesting
A paid point-in-time holdings panel, or a hand-curated weekly holdings table for at least 15 companies with at least 8 independent episodes. The frozen plan would then be run unchanged.
Past results of a backtest say nothing certain about the future.