Monte Carlo

Will your bitcoin last?

Every path starts from the last close. 2,000 paths in each of five futures, 30 years of spending. No floor is built in: price can fall below 0.42× the trend and stay there.

Your spending

Change any underlined value

IYou decide

I hold BTC

and spend $ a year,

starting .

IIWe assume

Your costs rise % a year.

On top of history’s crashes, add of .

Spending is in today’s dollars. History’s crashes are already in every future; since 2016 price halved within 90 days five times. An extra crash recovers only at the model’s normal pull toward the trend.

Safe spending, if the trend holds 99 of 100 paths

–a year

Simulating…

FutureRan out · safe spending

History repeats

Turn this into a plan → The planner sizes your stack across twelve futures.

Does the model look like history?

A simulation can only give back what its model assumes. So here are its own price paths, next to what Bitcoin actually did. If a future looks kinder than history, its answer above is too kind.

Month-ends below the floorLongest stay below the floorLowest point, typical path
Historymonth-ends since 2012–––

What the five futures assume

Every future starts from today’s price and moves month by month. Each month’s surprise is a real month from Bitcoin’s history since 2012, replayed in random 12-month stretches, so real crashes and real rallies are in every path. What differs is how hard price is pulled back toward the trend.

History repeats
The trend holds. Swings are as large as they were from 2012 to today, and price is pulled back toward the trend at the speed history shows: about 7% of the gap closes each month.
Calmer
The trend holds and swings are half as large, as if Bitcoin keeps maturing.
Fading swings
The trend holds and swings shrink steadily, to a tenth of today’s size after 30 years.
Stress: trend bends
The trend’s growth halves from today. Price keeps swinging around a line that rises more slowly than the formula says.
Stress: no pull-back
Price still grows with the trend on average, but nothing pulls it back: every loss is permanent, and paths spread further apart every year.

What none of them can tell you: every future assumes Bitcoin keeps working as money. A failure of Bitcoin itself is outside any model built on its price history.

Method and sources
  • The model. The distance of each month-end close from the published trend (10−16.493 × days5.688), on a log scale, follows an AR(1) fitted on month-ends since January 2012: –. The monthly surprises are that fit’s leftovers, replayed in random 12-month blocks. Rebuilt every week from the dataset.
  • Spending. A twelfth of the year’s spending is sold from bitcoin at each month-end, with a 0.25% cost, rising with inflation. A path runs out at the first payment it cannot cover.
  • Same model as the planner and the ebook. These are the statistical futures of the Observatory’s research report F011 (published as Paper 23). With F011’s data and starting point, this page’s engine gives the same safe spending as F011 in all five futures: from half the trend, 50%, 80% and 55% of the floor’s yearly rise in the base, calmer and fading futures (a test checks it every week, before the refitted model is published).
  • Every month of history drawn equally. History is replayed in 12-month blocks that may start at any month and wrap around the end. F011’s first run started blocks only in the first months of the record, so the latest months came up less often and its paths ended about 10% kinder against the trend; that was corrected on 17 November 2026, and its safe levels fell by about 5 points.
  • Inflation. F011 replays inflation from history; this page uses the rate you set.
  • Not in the model. Tax, where you store your bitcoin, income from work, and any other savings. Daily swings inside a month are not simulated; they matter only if you sell on that day.
  • Earlier versions of this page clamped price at a floor 2σ below the trend. That made running out almost impossible and the answers too kind; it is gone.

This simulator is free. Your safe spending: –.

Not financial advice. Simulations of price history, not a forecast.