Can You Lose Money With Bitcoin DCA? A Historical Study
We tested every completed monthly Bitcoin DCA start over 1, 3 and 5 years. Explore losses, recovery paths and downloadable historical results.
By The Editorial Team
Yes, you can lose money with Bitcoin DCA, even after years of regular purchases. In this study, 34 of 132 completed 12-month plans and 5 of 108 completed 36-month plans ended below the money contributed. None of the 84 completed 60-month plans did. That last finding describes one historical sample; it does not establish a safe holding period.
We tested every eligible monthly starting date in a frozen Bitcoin price series, with purchases from October 2014 through August 2026. Each plan buys $100 on the first day of the month and values its holdings at the last day of the final month. The rules stay the same for every start, including the uncomfortable ones.
The interactive study below lets you choose a start, change the contribution, and replay the path. You can inspect the exact month-end values or download every result. No account is needed.
What the completed plans actually returned
The table measures terminal loss: whether the portfolio value at the final checkpoint was below cumulative contributions. It does not count every temporary decline along the way.
| Monthly purchases | Completed starts | Ended in loss | Loss frequency | Worst ending return |
|---|---|---|---|---|
| 12 ($1,200 contributed) | 132 | 34 | 25.8% | −53.8% |
| 36 ($3,600 contributed) | 108 | 5 | 4.6% | −10.3% |
| 60 ($6,000 contributed) | 84 | 0 | 0.0% | +39.7% |
The median return on contributions was +40.2% over 12 purchases, +155.6% over 36, and +246.3% over 60. These are cumulative returns, not annualized returns or interest rates. In a monthly plan, the first dollar is invested much longer than the last dollar; dividing the final value by contributions does not give a time-weighted or money-weighted annual return.
The latest eligible starts differ: September 2025 for the 12-month test, September 2023 for 36 months, and September 2021 for 60 months. Every plan ends by August 31, 2026. Newer five-year plans remain unfinished and are excluded rather than treated as successes or failures.
Every start tells a different story
Choose a holding period, then a starting month. Each cell is a completed monthly DCA window, not a forecast.
25.8% · 132 completed windows
Return by starting month
12 monthsOn a small screen, scroll the calendar horizontally to see all months.
Historical loss frequency is not a future probability. Adjacent windows share most of their prices.
Watch the journey
July 2021 – June 2022
Recovery means the first month-end back at or above contributions after the deepest percentage loss, while monthly purchases continue. It can fall below again.
Read the chart as a table
| Date | Contributions | Portfolio value | Return on contributions | BTC held |
|---|---|---|---|---|
| Jul 1, 2021 | $100 | $100.00 | 0% | 0.00297866 |
| Jul 31, 2021 | $100 | $123.99 | +24% | 0.00297866 |
| Aug 31, 2021 | $200 | $258.48 | +29.2% | 0.00548023 |
| Sep 30, 2021 | $300 | $329.63 | +9.9% | 0.00752744 |
| Oct 31, 2021 | $400 | $589.01 | +47.3% | 0.00960571 |
| Nov 30, 2021 | $500 | $641.02 | +28.2% | 0.01124494 |
| Dec 31, 2021 | $600 | $601.63 | +0.3% | 0.01299228 |
| Jan 31, 2022 | $700 | $580.68 | -17% | 0.01508929 |
| Feb 28, 2022 | $800 | $763.24 | -4.6% | 0.01767039 |
| Mar 31, 2022 | $900 | $907.36 | +0.8% | 0.01992494 |
| Apr 30, 2022 | $1,000 | $832.96 | -16.7% | 0.02208562 |
| May 31, 2022 | $1,100 | $784.80 | -28.7% | 0.02468511 |
| Jun 30, 2022 | $1,200 | $554.78 | -53.8% | 0.02804092 |
Start with the default July 2021 entry and 12-month period. Then select 60 months while keeping that same start. The comparison makes the distinction between a painful year and a longer completed outcome concrete. Neither view tells you what a new purchase today will return.
A bad year can overwhelm the benefit of cheaper purchases
The worst 12-month ending in this sample began in July 2021 and ended in June 2022. Twelve $100 purchases contributed $1,200; the holdings were worth $554.78 at the end, a $645.22 loss, or 53.8% of contributions.
Buying at lower prices did accumulate additional Bitcoin. It could not stop the market value of all the accumulated units from falling. This is the central limit of averaging: the rule changes the entry prices of the units you own, not the price at which those units can later be sold.
The arithmetic is simple. At each purchase, Bitcoin acquired equals the contribution divided by that day's price. At the end, portfolio value equals all Bitcoin acquired multiplied by the ending price. If that ending price is below the average cost per Bitcoin, the plan loses money before costs.
A small example shows why the distinction matters. Buy $100 at a price of $100, then another $100 at $50. You now hold three units for $200, with an average cost of $66.67. At an ending price of $40, those units are worth $120. Your average cost fell, but your portfolio is still down 40%. A lower cost basis and a profitable investment are different outcomes.
Investor.gov defines DCA as investing equal amounts at regular intervals regardless of market movements. The definition contains no promise about the asset's future price. FINRA's Bitcoin overview likewise emphasizes wide price swings and the possibility of loss.
Three years did not remove the risk
Five completed 36-month plans ended below contributions. Their starting months were December 2019, January 2020, September 2020, October 2020 and July 2023. The worst began in January 2020 and ended in December 2022: $3,600 contributed, $3,229.48 remaining, a 10.3% loss.
That is a useful counterexample to the claim that a specific calendar commitment guarantees a positive result. Regular purchases spread entry decisions across time. They leave the ending valuation exposed to the market on the chosen exit date.
It also matters whether an investor must sell on that date. A backtest can effortlessly hold through a decline; someone paying for a home, education or an unexpected bill may not have that flexibility. This study measures what the portfolio was worth on schedule. It cannot tell us whether a real investor could afford to keep buying, whether they would sell early, or whether waiting longer would suit their needs.
If you want to explore another amount or date range, use the Bitcoin DCA calculator. Its flexible historical tool answers a different question from this fixed-window research. Changing purchase frequency, valuation dates or the underlying data version can produce different results without either calculation being wrong.
A positive ending can hide a long period in loss
Of the 84 completed 60-month plans, 75 had at least one month-end below contributions, even though all 84 finished above them. A green ending cell should not be read as a comfortable journey.
The longest observed run below contributions was 22 consecutive month-end checkpoints. It occurred in the 36-month and 60-month paths starting in March 2021. That is a measurement of sampled month-end balances. It does not establish that the portfolio was underwater on every day between those checkpoints, and it is not a peak-to-trough drawdown statistic.
Return on contributions answers, “Is my portfolio worth more than I have put in?” Drawdown usually asks how far a value has fallen from a previous peak. Those questions can have very different answers, particularly while fresh cash is entering the account. We use the first consistently throughout this study.
The recovery label in the interactive chart also has a narrow definition. We locate the worst percentage return at a month-end, then count checkpoints until the portfolio next reaches its cumulative contributions, while scheduled purchases continue. If no such checkpoint appears before the selected plan ends, we display “Not observed in this window.” We do not invent a recovery date outside it.
For the July 2021 start with 60 purchases, the worst checkpoint was June 2022. The first subsequent non-negative checkpoint arrived 10 months later, in April 2023. Returning to break-even once does not mean the portfolio can never fall below it again. Nor does this result describe someone who stopped contributing at the low: that would be a different cash-flow path.
Why zero observed five-year losses is not a guarantee
The five-year sample is encouraging as a historical description, but it has important limits.
The windows overlap. Two starts one month apart share 59 of their 60 purchase months. Eighty-four completed plans are not 84 independent market histories. They are multiple views of roughly twelve years of one asset. Treating the observed 0% loss frequency as a forecast would exaggerate the amount of independent evidence.
The asset was selected after it survived. This is a study of Bitcoin, not a diversified sample of every cryptocurrency an investor might have selected in 2014. The results cannot be transferred to failed projects or used to argue that averaging into any declining asset eventually works.
Longer horizons have an older starting sample. A 60-month window can only start through September 2021 here. A 12-month window can start through September 2025. Comparing their loss frequencies mixes horizon length with different entry cohorts and market episodes. It is not a controlled experiment isolating the effect of waiting.
The study excludes costs and non-price risks. Trading fees, spreads, taxes and inflation can reduce what an investor keeps. Custody failures and lost access to Bitcoin do not appear in a price-only series. A positive nominal USD return also does not establish a positive inflation-adjusted return or a profit in another home currency.
These limitations do not make the history useless. They define what the evidence can support: under the specified rules and observed prices, certain starts lost money and others gained it. They cannot support a promise that future outcomes will repeat.
Methodology: enough detail to reproduce the numbers
Data and cutoff. We retrieved Yahoo Finance's BTC-USD daily close series on September 26, 2026, and froze observations through August 31, 2026, the most recent complete calendar month. The underlying series begins September 17, 2014. Because September lacks a first-day price, the first eligible purchase month is October 2014. The study contains 143 complete purchase months.
Purchases and valuation. A plan buys $100 at the daily close on the first calendar day of each month, including weekends. There are exactly 12, 36 or 60 purchases. Every checkpoint values the accumulated units at that month's final calendar-day close. There is no extra purchase at the terminal checkpoint, no price interpolation and no assumed fill before a quoted close.
Example calendar. The July 2021 12-month plan buys on July 1, August 1 and each subsequent first day through June 1, 2022. It values the final holdings on June 30, 2022. “12 months” refers to 12 monthly purchase periods, not 365 equal days of exposure for every contribution.
Arithmetic. Fractional Bitcoin is allowed; source prices retain their supplied precision. Units accumulate as the sum of $100 divided by each purchase price. Ending return equals (ending value / total contributed − 1) × 100. Dollar and percentage figures are rounded only for display. The contribution slider scales amounts; it does not change percentage outcomes because the model excludes fixed costs.
Coverage and checks. We require an exact first-day and last-day price for each month and consecutive months within every plan. Results were reproduced with a separate Python calculation and tested with flat prices, known two-purchase examples, missing months, incomplete windows and unrecovered losses.
Download the 324 completed windows as CSV, the full results and monthly portfolio paths as JSON, the monthly purchase and valuation prices, or the frozen daily source observations. The metadata includes the source URL, retrieval time, calculation rules and a SHA-256 fingerprint of the daily observations. The provider's historical data page may later contain revisions; our downloads preserve the snapshot used here.
What to take into your own planning
DCA can make a purchase schedule easier to follow. It cannot turn a risky asset into a guaranteed savings vehicle. A useful plan needs both an entry rule and a response to the possibility that its value is below contributions when the money is needed.
Use the calendar to inspect an unfavorable start as well as a favorable one. Look at the path before the ending number. Then consider whether the contribution amount, liquidity needs and exposure to a single asset fit together. The guide to gradual selling explores what happens on the exit side, and the real cost of DCA adds the trading costs excluded here.
This is historical research for education, not a price forecast or a recommendation to buy or sell Bitcoin. Its most useful result is not a promise of recovery. It is a visible record of how much uncertainty a regular purchase schedule can still contain.