# FIRE number via Monte-Carlo NW-threshold on the liquid pot

We compute each Case's target net worth (its "FIRE number") by searching net
worth for the smallest value at which Guyton-Klinger reaches a 99% Monte-Carlo
success rate — **not** a fixed safe-withdrawal-rate multiple (e.g. 25× spend).

A fixed multiple cannot honour the structure the engine already models: the
rising 30→70 equity glide, per-jurisdiction taxes drained from the portfolio,
the kids-cost ramp, an optional one-time home purchase, and a workplace pension
that is locked until ~57. Our block-bootstrap's empirical perpetual SWR is also
~2.5–3%, materially below the textbook 4%, so a 25× multiple would understate
the number anyway.

The solver seeds on **liquid** net worth — it excludes the Fidelity workplace
pension (inaccessible until ~57) and injects that pension as a grown lump at age
57. Early-retirement sequence risk is funded only by spendable assets, so seeding
total net worth would overstate safety exactly where the 99% bar is decided.

## Consequences

- A vectorised NW search (binary search over `initial_portfolio`, monotone
  because the GK year-0 draw is an absolute real amount) populates a
  `fire_target` table, one row per (Case × country × with-home).
- Targets vary by country through **both** COL-scaled spend and the destination
  tax regime (the simulator drains tax from the portfolio since 2026-05).
- The Grafana countdown reads `fire_target` for the selected country and diffs
  it against current liquid net worth from `account_snapshot`.
- Pension growth to age 57 is modelled deterministically (current value
  compounded at an assumed real rate), not per-path — a conservative
  simplification, revisitable if it ever binds.