Bitcoin DCA vs lump sum: every start month since 2014

Lump sum usually finished ahead, DCA lost less in the worst windows

We compared two ways of putting the same money into Bitcoin. Dollar-cost averaging (DCA) buys an equal dollar amount on the first day of each month for 12, 36 or 60 months. The lump sum invests the full total on the first of those days. Both are valued on the first day of the month after the final DCA purchase. Every start month from January 2014 whose full horizon ends by the last available close, 23 May 2026, is included.

Over 12 months the lump sum ended ahead in 91 of 137 windows, 66.4%. Over 36 and 60 months the share rises to 73.5% and 91.0%. That pattern is expected for an asset whose price rose strongly over the sample: money invested earlier spends more time exposed to the rise. It is a property of this historical path, not a rule that applies to the next window.

The downside comparison runs the other way. In 12-month windows the worst lump-sum outcome was −72.1% (starting February 2014) against −54.8% for the worst DCA window (starting July 2021). DCA ended below its money invested in 27.0% of 12-month windows and the lump sum in 29.9%. Spreading purchases reduced exposure to a single bad entry date; it did not remove the risk of a loss.

Starting at the top of a cycle

We also started a monthly DCA on the highest daily close of each of four cycles: 4 December 2013, 16 December 2017, 8 November 2021 and 6 October 2025. Later purchases fall on the same day of each following month. Valued on purchase dates, the deepest point was −49.8%, −54.1%, −42.1% and −23.8% of money invested respectively.

For the first three peaks the DCA value was back at or above money invested by the 24th, 18th and 18th purchase. The October 2025 path has eight purchases in the sample and stood at −2.9% on 6 May 2026, so its recovery is not yet observed. Daily lows between purchase dates could be deeper than these figures, and three completed recoveries are a small sample.

How the numbers were calculated

Prices are Coin Metrics’ public PriceUSD series, a UTC day-end reference price, with no missing dates in the period used. Each purchase of $100 buys 100 / price units at that day’s close. Return on money invested is ending value divided by total contributions, minus one. The lump-sum return is the price change from the first purchase date to the valuation date.

Windows overlap: consecutive start months share most of their prices, so the 137 twelve-month windows are not 137 independent experiments. No fees, spreads, taxes, custody costs or interest on cash waiting to be invested were included. Interest on the uninvested DCA balance would narrow the lump-sum lead somewhat; trading costs would affect DCA more because it makes more purchases.

What this does and does not tell an investor

The comparison measures outcomes on one asset that survived and rose a great deal since 2014. Assets that fell permanently are not in the sample, and choosing Bitcoin after its success flatters both methods. The results do not show that either approach suits a particular person, and they are not a forecast of Bitcoin’s future price.

The defensible reading is narrower: with a fixed sum already in hand, investing it at once has historically finished ahead more often in this sample, while DCA changed the range of outcomes and reduced the regret of a single badly timed entry. For people investing from each paycheck, regular buying is simply how the money arrives, and the lump-sum comparison does not apply.

Is DCA better than lump sum for Bitcoin?

Not by ending value in most of this sample: the lump sum finished ahead in 66.4% of 12-month windows and more often over longer horizons. DCA produced a less severe worst 12-month outcome. The answer depends on the period and on how much a bad entry date matters to the investor.

What if I start buying Bitcoin monthly at the top?

DCA started on the 2013, 2017 and 2021 peak closes fell 42–54% below money invested at their worst purchase dates, then recovered to cost by the 18th to 24th purchase. That covers three historical cycles only; a recovery is not guaranteed.

Limitations

One surviving asset, one daily reference-price series and overlapping monthly windows. No fees, taxes, spreads, cash interest or intraday prices are modelled. Historical outcomes are not a forecast or personal investment advice. The raw source can be revised; its fingerprint identifies the snapshot used.

Sources and data access

• Coin Metrics: public BTC daily data
• Coin Metrics: PriceUSD definition and timestamp convention

Cited sources document external facts or data formats; they do not endorse our findings. The public snapshot contains the table and methods. Source-file fingerprints are included for recalculated datasets. Full archived raw files are not distributed here; fingerprints identify the files used, rather than independently proving the results.

Aggregate data snapshot (JSON) · Markdown · Methods and limitations

Live prior 30-day hourly baseline — separate from the historical study

Only the current watch list is shown, up to 12 assets. Missing or stale estimates remain unavailable. Venue and quote currency are kept separate; beta needs matching BTC data. Hourly statistics are not minute-move probabilities.

Observation cutoff (UTC): 23/10/2026, 15:52

Related research