Cash is a position, not a residue

For weeks my equity book held more cash than its own policy said it should, and every day the system produced a perfectly reasonable explanation. Nothing cleared the research gates today. Risk is near its cap. Liquidity is thin. The regime is hostile.

Each excuse was individually true. Collectively they were a portfolio decision nobody had written down: stay under-invested indefinitely. Idle capital doesn’t feel like a position because nothing happens — no order, no confirmation, no line in the log. But a book that is 20% cash has made a 20% allocation to earning nothing, renewed silently every session.

The fix was to give cash the same treatment as any holding: a written policy with numbers, and consequences when the numbers are breached.

The band

Cash gets a working band — a floor, because dry powder and settlement friction are real, and a hard ceiling a couple of points above it. Inside the band, discretion. Above the ceiling, discretion ends.

Below the ceiling, one rule still applies every session: cash above target must be explained by a named binding constraint — which gate failed, which cap is full, what specifically is missing. “No trade” stays a legitimate outcome, but only as a sentence with a subject in it. “Nothing looked great” does not qualify. The naming matters because vague constraints renew themselves for free; named ones get falsified by next week’s work.

The ladder

Above the ceiling, an explanation stops being sufficient. The system walks a written ladder, top down, and takes the first rung that clears:

1. A cleared idea. If research has already produced a candidate that passed every gate, deploy into it. This rung existing first keeps the ladder honest — the whole point of the pipeline is that this is usually where it ends.
2. Add to a proven winner. If nothing new cleared, up-size an existing high-conviction, profitable holding — with a freshness condition (the conviction score must be recent, not remembered) and a per-position cap it cannot breach.
3. Fallback exposure. Buy broad index exposure sized to bring cash back under the ceiling. This is a waiting room, not a strategy: it is excluded from the book’s risk accounting, capped in total size, and is by rule the first thing sold to fund any real idea.

The ordering encodes a preference — real conviction beats mechanical deployment — but removes the option of doing nothing while over the cap.

The relabeling trap

One clause did more work than I expected: parking cash in a liquid fund still counts as cash against the ceiling. It earns a little, it settles a day slower, and it changes the book’s exposure not at all. Without that clause the ladder has a cheat code — sweep idle cash into a park, watch the “cash” number fall, deploy nothing. Renaming a balance is not an allocation decision, in a portfolio or on a balance sheet.

The principle underneath

The mandate this came from is one sentence long: nobody allocates capital in order to hold cash. The engineering translation: every rupee needs a written reason, including the ones doing nothing — and above a stated line, a reason is no longer enough; only an action is.