TL;DR: FIREproof runs your plan once for every possible historical starting year. Each run is a cycle: a window of real equity, bond, cash, and inflation returns played back in the order they actually happened. Your success rate is the fraction of those starting years that finished the horizon with money still in the portfolio.

What a cycle is

A Cycle is your plan replayed against one specific slice of history. If your horizon is 30 years and the dataset starts in 1871, a cycle that starts in 1929 plays the 1929 equity returns in year 1, the 1930 returns in year 2, and so on through 1958. Another cycle starting in 1965 replays 1965-1994. Every cycle gets its own copy of your accounts, adjustments, and spending plan, then runs the year-by-year loop independently.

Historic vs Constant

• Historic (default) - one cycle per valid starting year. Returns and inflation come from the actual sequence, paired together.
• Constant - a single cycle that uses fixed annual return and inflation assumptions you set. Useful for sanity checks and back-of-the-envelope projections, not for stress-testing.

Reordering the Select Cycle slider

On Proof, the "Select Cycle" slider above any cycle-aware view has a small Year / Percentile toggle next to its title. Year (the default) orders cycles chronologically by their historical starting year - the same order they're indexed in. Percentile reorders them by ending portfolio value, highest on the left and lowest on the right, so slot 1 is the best historical outcome and the final slot is the worst. The slider's floating label switches to a P0-P100 label (P100 = best, P0 = worst); the "Data Start" sublabel still shows the actual historical year of the cycle you've landed on. The first time you flip to Percentile in a session, the slider jumps to roughly the median (P50) so you start near the typical outcome. After that, toggling between modes preserves your current selection - the thumb just moves to its new position in the new ordering.