# KISS 2.3 sleeve-confirmation revision

## Decision

Track the allocation confirmation count independently for stocks, gold, and bitcoin:

- reduce a sleeve after its lower target persists for two closes;
- increase a sleeve after its higher target persists for five closes;
- do not reset one sleeve when another sleeve's target changes;
- implement each confirmed sleeve change on the next trading session;
- hold any temporary difference in cash.

This replaces the KISS 2.1 and 2.2 implementation, which required the exact combined portfolio target to remain unchanged throughout the confirmation window.

## Why it changed

The combined-target counter created an unintended dependency between independent assets. In August 2026, bitcoin reached 100% of its maximum signal target for five consecutive trading closes. Gold joined the candidate allocation after bitcoin's count had started, which changed the combined target and reset the shared counter. Bitcoin therefore remained at 0% even though its own signal had completed the five-close requirement.

Under KISS 2.3, bitcoin confirms on its fifth qualifying close. Gold keeps its separate three-of-five count. A later gold confirmation cannot delay or reverse bitcoin unless bitcoin's own target changes.

## Rotation behavior

A rotation can now occur in stages. The departing sleeve confirms its reduction after two closes, while the arriving sleeve continues toward five closes. Cash absorbs the gap until the increase confirms. This preserves the faster defensive threshold without letting one asset borrow confirmation from another.

## Versioning and history

This changes allocation timing, rebalance events, notifications, and simulated returns, so it requires the `kiss-2.3.0` methodology identifier. Daily signal history and both tax-deferred performance variants must be rebuilt under the same version before publication.