The Distillate — Live

Real money, running the same rule that is sold here. Deposits are shown as deposits, never as return. Not investment advice.

Where it stands

Exposure scales inversely with market volatility and is capped — the position shrinks as the market gets rougher, grows as it calms, and changes only when the gap is worth acting on: quick to reduce risk, slow to take it back. Weeks of silence are the design working. The specific calibration is the product and is not published.

Flight recorder

Did the software deliver what it sold?

Three reference books run beside the account from the same first day, the same deposits and the same daily rule, filled at each session’s close, a dollar a leg, cash earning nothing: the rule with fractional shares (the reference), the rule with whole shares, and the rule with whole shares that cannot spend a sale’s proceeds until the next session. The differences between them are the cost of rounding and of settlement, each a comparison of two model books. The rest of the gap between the account and the reference is what the account held beyond that (a hand trade, the leg mix, the morning-versus-close position), fill prices against the close, commissions, and cash events seen in the cash (dividends, broker fees, interest, named from the broker’s own statement). Whatever is left is printed as unexplained and never folded into another column. The pieces below are linked to the compounded gap, so they add up to it exactly; a market holiday is a valuation boundary with no trade, so a deposit made on one is never counted as return.

Last ten sessions, each day’s own pieces (unlinked). The reference trades only on a signal recorded the previous trading session (a holiday in between is skipped, not read as a missing signal); it is seeded from the account’s own holdings at the first session that has such a signal, so its first position is known, not guessed, and the ledger starts there. A dagger (†) marks a session whose rule was reconstructed from today’s dial rather than recorded that night; those rows depend on that assumption and the reference for them is not frozen. How it is counted, in order: a day’s live return is the positions held at the prior valuation times the day’s price change, plus each execution’s shares times the close minus its fill price, minus commissions, plus what the cash did beyond trades and deposits. The fill column is the second of those, on the day of the trade only; from the next session the position is in the held terms. Nothing is counted twice. A favourable fill figure is an observation against the closing-price yardstick, not a demonstrated timing advantage, and its sign can change. The settlement column reads: settlement restrictions changed this reference book’s return by that much in this sample; it is not the account’s own settlement cost, and no replay of the account’s actual orders is shown here because a faithful one (subsequent cash, positions and sizing included) has not been built.

The configurations

Since-2009 CAGR and the worst full-cycle drawdown (1999–2026), from the same generator as the engine table on the main page. The expansion row is measured on bitcoin’s own history, since 2018, which is all it has. The highlighted row is the one the live account runs.

One construction is being shadowed beside this account, on paper, with the numbers and the wounds: the expansions page.

*day return is flow-stripped: deposits and withdrawals are removed before the percentage is computed, so cash you add can never appear as performance. Methodology and every number this strategy has ever published: the evidence page · what died to make it: the graveyard.