RabbitSuitThe approach
What this is, why Bitcoin, what we committed to, and what can overrule the rule. For how it is doing, see the fund page →
What this is
A Bitcoin fund run by one rule.
The fund holds R* × NAV in Bitcoin. R is the rules' daily reading, a number from 0 to 1. R* is what the fund actually holds: usually the same number, sometimes following R with a short delay. Nobody sizes positions by hand, and no code path can hold more Bitcoin than R* says.
The rest, 1 − R*, is parked in exactly one of cash (money market), world stocks, gold, the Nasdaq-100 or long US Treasuries, re-picked at each month-end by a rule. Switching it never changes the Bitcoin share.
Four model funds run the same dial, computed once a day from published closing prices. They differ only in instruments and currency; they exist so the rule can be watched in each form.
It runs only our own money. There are no outside investors, no way to invest in it, and nothing on this site is advice or a recommendation to buy or sell anything. What we publish (the position, the record, the journal) is a track record kept in the open, for discipline and for anyone curious about a fully rules-based Bitcoin allocation.
The rules are not frozen. The dial is the best rule we have found so far and may change given strong evidence; the same goes for the parking rule and its list of assets. Every change needs a written reason and a full re-run of the backtest, is published in the journal, and is flagged if made during a drawdown. None of this happens day to day: within the day, the dial is law.
Why Bitcoin
The rule sizes exactly one asset. Five facts about it explain the choice, and each ends in a design decision.
The only real digital asset
Bitcoin is a category of one: the only digital asset that is truly scarce and belongs entirely to whoever holds it. No issuer, no company behind it, no one who can change what it is. It combines things nothing else, digital or physical, does:
- Full ownership
- Held in self-custody, there is nobody between the holder and the asset: no counterparty, no vault, no issuer.
- Censorship-resistant
- Anyone can hold and move it; no administrator can freeze a balance, rewrite the ledger or shut someone out.
- Cheap to hold and move
- No storage cost, divisible to eight decimals, and billions of dollars settle worldwide in minutes for a few dollars in fees, against weeks of vault logistics for gold.
- An open ledger
- Supply, flows and what holders paid are all visible to anyone, in real time. No other asset publishes its own books.
The design decision: these properties, together with the fixed supply, are where the asset's value comes from, and they only fully hold for self-custodied Bitcoin. The fund's job is exposure to that value's price, through whichever instrument fits each fund best. The open ledger also makes the asset measurable in a way no stock, bond or metal is, which is part of why a fully mechanical rule is possible at all.
A scarce asset in an age of money printing
Bitcoin's supply is fixed by code: 21 million, on a schedule no board, committee or government can change. The record it stands against is one-sided: every paper currency ever launched has lost value against hard assets, the major ones by about 99% against gold since leaving it. And the debts of the large governments make more money printing the path of least resistance, not an accident.
The design decision: an asset that cannot be printed is the natural thing to hold against a system whose release valve is printing. That is why the fund exists at all, and why its Bitcoin share can reach 100% rather than a token slice.
A long-term trend you can measure
For sixteen years Bitcoin's price has followed a single power-law trend, through four complete boom-and-bust cycles, with the fit explaining about 96% of the variation in log price. That regularity is the signature of adoption, not of hype. The record is one-sided, though: as a gauge of cheapness the trend has held up well; as a source of price targets it has repeatedly overpromised.
The design decision: the fund never trades on price targets; it sizes to today's conditions. The trend also supplies the test that would end the fund, quoted below: a lasting break below the trend line would mean adoption in reverse, and with it the end of the fund's premise.
One main driver you can watch
Bitcoin follows global liquidity more closely than any other major asset: in a 2013–2024 study it moved with it in 83% of twelve-month periods, ahead of stocks, gold and bonds, with a long-run correlation near 0.94 to global money supply. There are no earnings, dividends, index flows or safe-haven bids muddying the signal, the things that blur the same relationship in every other asset.
The design decision: an asset with one main, published, measurable driver is an asset a rule can size. Most assets never offer that. It is what makes a daily, fully rules-based dial realistic rather than wishful.
Cycles brutal enough to need a rule
Bitcoin is the only asset we know of that has fallen more than 75% on four separate occasions and made a new all-time high every time. Over the decade to mid-2025 it still had the best Sharpe of any major asset class, while carrying the worst drawdowns in the comparison.
The design decision: the return is structural, the drawdowns are cyclical, and the cycle extremes leave visible traces in public data. Holding through pays for the trend with −80% winters; staying out gives up the trend entirely. A dial that scales exposure to conditions is what this asset's shape calls for.
Why it exists
To remove one specific mistake, and to be judged on one specific claim.
The claim
Returns like holding Bitcoin, at a fraction of the drawdown, judged on Sharpe and MAR. Total-return gaps are published too, but they come second to the risk-adjusted comparison.
Every performance view therefore shows two benchmarks as first-class rows: holding Bitcoin (the cost of caution) and holding cash (the cost of waiting).
The claim is judged on Bitcoin's mature era, from April 2018 on, when its volatility settled into a bounded range. Earlier history is shown for illustration only. How the eras were measured →
What we committed to
No averaging into a falling R. R falling is the formula de-risking; overriding it with discretion is the failure mode this fund exists to remove.
Discretion goes one way only: down. Every intervention the fund can make reduces exposure, and each was written down before the situation it applies to.
At the moment it fires it will look like the buying opportunity of a lifetime — that is precisely why the criterion is pre-committed.
How the dial works
What can overrule the dial
Pre-agreed interventions that rank above the formula. Each one can only reduce exposure; none can increase it.
- kill switcharmed
The full stop: the Bitcoin share goes to zero, even below the floor.
- adoption falsifierarmed
The test that would end the fund: pre-agreed evidence that Bitcoin adoption has reversed. While it stands, all buying stops.
- Bitcoin flooractive
Always on: a small Bitcoin share the fund never goes below.
- crash trimarmed
Automatic crash brake: while stock-market volatility spikes into a crash, the Bitcoin share is halved, buying pauses and the parked share moves to cash. It releases by itself as things calm down.
None of them covers a Bitcoin-only crash, by design: a fall with no stress in stock-market volatility for the crash trim to read. In that case the fund accepts a loss of its Bitcoin share times the fall: with 60% in Bitcoin, a −35% crash over two days costs about 21%, and a −50% fall over a month about 30%. Every crypto-only trigger we tested against the record would have cost more than it saved, and none fired before a low, so the dial buys into those crashes rather than fleeing them. An open options position may partially offset such a crash; it does not guarantee coverage or a full offset. The record shows both the at-cost figure and, beside it, the same value with open options at their payout worth.
Every proposal, activation and lift is journaled and audited. Current status →
The options position
A small options position for substantial Bitcoin moves, with partial downside protection while open. It is paid from the parked share and does not change the Bitcoin dial.
The fund sometimes buys a call and a put to participate in a substantial Bitcoin move in either direction. The call participates in an upside move; the put can offset some losses in a fall while open. Each purchase can lose its entire premium. A move must be large enough relative to the strikes and premium paid to produce a profit; quiet trading alone does not mean options are cheap. Entry and size follow rules that are not published.
It is deliberately boring to run: entries are by rule, the size is a small fixed share of the fund, and positions are simply held to expiry. No take-profit, no stop, no judgement calls. Open positions and their cost are on Today; every entry and settlement appears in the journal. Other investors may sell options to earn income in exchange for some upside, while dealers can hedge the risks they sell. The fund pays for a different payoff; this does not require the seller to be making a pricing mistake.
Its rules
- Limited loss
- Only ever bought options: the most a position can lose is what it cost, and that is treated as spent from day one. The position is counted at cost; any value above that is shown but not counted.
- Paid from the parked share
- The premium comes out of the parked share, never out of Bitcoin, and payouts go back there. The Bitcoin share is the same with or without it; the dial stays law.
- Hard-capped
- Each entry is a small share of the fund, total open premiums are capped at cost, and an entry the parked share cannot pay for is skipped. The cap limits simultaneous premium commitments; repeated purchases can lose more over time, and it does not cap fund drawdown.
- Mechanical and self-checking
- Entries fire only when the rule says so. The rule grades itself: completed episodes are scored afterwards, a failed score shrinks the size, and a second failure suspends the position pending a design review.
Glossary
The words the numbers are reported in. Dotted terms across the site link here; the same text shows on hover.
Reporting terms
- R — formula dial · 0 → 1
- What the rules say today: the share of the fund that should be in Bitcoin. What the fund actually holds is R*.
- R* — executed dial · 0 → 1
- The share of the fund actually in Bitcoin. Usually the same number as R. When the rules move R against the wider liquidity trend (a cut while money is getting easier, a rise while it is getting tighter), R* follows R with a short delay rather than all at once. It never holds more Bitcoin than the rules have called for.
- CAGR — yearly growth rate
- The steady yearly rate that would take the series from its first value to its last. Hidden until a record spans 90 days; a few days of value tell you nothing about a year.
- maxDD — worst drawdown
- The biggest fall from a peak to the low that followed. The number the claim is really about: returns like holding Bitcoin, at a fraction of this.
- Sharpe — return per unit of volatility
- Arithmetic mean of daily returns above the cash benchmark, divided by the sample standard deviation of those daily excess returns, multiplied by √365.25. CAGR is reported separately.
- MAR — return per unit of drawdown
- Yearly growth rate divided by the worst drawdown. The second risk-adjusted test, next to Sharpe.
- TWR — time-weighted return
- How much the value grew over the window: what 100 at the start became by the end.
- SI — since inception
- Measured from the fund's start at 100 to the latest close: its whole life, launch costs included, not a window.
- backtest — the simulated record
- Today's rules run over past data (2012-02-01 to 2026-08-02, the day before the funds went live). The headline window starts 2018-04-13, the mature era. No fund existed then and nothing was traded.
- era — an evaluation window
- A stretch of Bitcoin's history with roughly constant volatility, found in the data rather than picked from a calendar. Backtest figures are quoted per era; the headline is the mature era, from 2018-04-13 on.