How long have Bitcoin DCA buyers actually stayed underwater?

Across 624 weekly start dates since September 2014, 80 are below cost today — about one in eight. The deepest is down 9.4 percent. The longest any cohort spent continuously below what it had paid in was 486 days, and every one of the 624 has been above water at some point.

Bitcoin is 36.7% below its October 2025 high and has been for 324 days. That makes this a reasonable moment to ask a question most calculators will not answer: if you had been buying steadily, would you be down right now — and if you were, how long would you have stayed there?

The usual answer is an anecdote. This one is the whole distribution.

What was measured

Every Wednesday-to-Wednesday start date from Bitcoin's first daily close on 17 September 2014 to today — 624 of them. Each one buys $50 a week and holds. No selling, no timing, no changes. Then each cohort is marked against the closing price on 26 August 2026.

Nothing here is a projection. It is 4,362 daily closes and a schedule, and you can reproduce any row of it in the simulator. Where the numbers come from is on the sources page, and how missing days are handled is written up in the knowledge base.

One in eight is below cost today

Every underwater cohort started between February 2024 and September 2025 — a band of 18 months. Buy for longer than that and, so far, no start date has finished below what went in.

The chart makes the shape of it plain. Almost every start date sits far above the break-even line — a 2014 start is worth around 50× what it paid in — and the underwater cohorts are a shallow dip at the recent end, not a cliff.

The worst case is worth sitting with, because it is unspectacular. A buyer who started on 13 November 2024 has paid in $4,700 and holds $4,259. That is the single most unfortunate week to have begun in eleven years of history, during a 36.7% drawdown.

The longer question: how long do you stay down?

"Are you down today" is the easy question. The one that actually decides whether someone keeps buying is how long it lasts. For each cohort I measured the longest unbroken run of days where the position was worth less than the money paid into it.

Sorted, the distribution is steep at both ends and flat through the middle: most cohorts cluster well under a year, and only the tail approaches the worst case.

The worst stretch on record belongs to people who started within days of the December 2017 top. They spent 486 days continuously below cost. They are now up several thousand percent, which is easy to say in hindsight and was not easy to sit through.

The median is the more useful number: about three months. Half of all start dates never spent longer than that continuously underwater.

That is where it finished. How far it fell along the way is a different measurement and a much larger one, and it is the subject of how far down Bitcoin DCA has actually gone.

What this does not tell you

Six things, plainly.

Past distributions are not forecasts. Every cohort here benefits from Bitcoin having risen over the measured window. A period where it does not would produce different numbers, and nothing in this dataset rules that out.

Survivorship is baked in. This measures buyers who kept buying. It cannot measure the ones who stopped in month four, because they are not in the data — and the whole difficulty of DCA is that stopping feels most reasonable exactly when the numbers here look worst.

Fees, spreads and taxes are ignored. A real $50 weekly buy carries costs this simulation does not model.

The cohorts overlap, heavily. Two start dates a week apart share every purchase but one, so 624 cohorts are nothing like 624 independent tries — they are a single price history read 624 ways. The spread above is real, but it is the spread of overlapping windows on one history, and it will understate how differently this could have gone.

The longest run is a floor, not a ceiling. 80 of these cohorts are still below cost today, which means their stretches have not ended: the longest has been running 321 days since 2 July 2025, and nothing here can say where it stops. 486 days is the worst stretch observed so far, not the worst this data will eventually show.

Underwater against what? A cohort down 9.4% is only half a fact. The other half is what the same $50 a week would have done somewhere else across the same weeks, and this piece does not measure it — being down against your own cost basis is a different question from being behind the alternative you passed up. The same schedule into 18 other assets measures that side of it.

Run it on your own dates

The figures above are one schedule — $50 weekly, held to today. If your start date, amount or cadence differ, the answer differs. Open the simulator and use your own.