DCA simulator — the variable the plan does not control
The plan did not change. The year did.
Same asset, same $530 across 53 weekly buys, one year apart. Run your own dates, then see why the usual comparison fails.
Started 9 January 2017
$3,820.16
Up 620.8%. This is the run that gets screenshotted.
Started 8 January 2018
$319.22
Down 39.8%. Same asset, same amount, same discipline.
Live
BTC
Price:
Loading...
Loading...
DCA Simulator Pro
Professional dollar-cost averaging simulator
Show Stablecoins 🔒
Adjust for Inflation (3% YoY) 🔒
☀️
🌙 🔒
Equal DCA
This strategy simulates investing a fixed dollar amount at regular intervals, regardless of the asset's price. It helps reduce the impact of volatility and removes the need to time the market.
Vs. Lump Sum 🔒
Show Bitcoin Halvings 🔒
01 — Why the usual comparison fails
The plan did not change. The year did.
Most DCA calculators hand you one number from one window and let you read it as a property of the strategy. It is not. It is a property of the dates. Below are two runs of the identical plan — $10 a week into Bitcoin, 0.1% fee — over two consecutive years.
Started 9 January 2017
$3,820.16
$3,820.16
$530 went in across 53 weekly buys and came out worth $3,820.16 — up 620.8%. This is the run that gets screenshotted, and it is the reason people believe they know what DCA does.
Started 8 January 2018
$319.22
$319.22
The same $530, the same 53 weekly buys, the same asset and the same fee — down 39.8%. Nothing about the plan was different. This window simply began on the day the one beside it ended.
The reasoning runs in one direction and it is worth following slowly. For a backtest to tell you something about a strategy, the result has to come from the strategy. For the result to come from the strategy, it has to survive a change of start date. And on a volatile asset it very often does not — a single backtest is one draw, not a finding.
That is why the start date is a field on this page rather than a decision made for you, and why the free tool draws the lump-sum line beside your plan instead of leaving you to guess at the alternative. Run both years above yourself. The tool will reproduce them.
Educational content only — not financial advice. Past results describe what already happened; they do not forecast anything.
02 — The arithmetic
Nothing here is hidden. Check every line.
The two runs from section 01, laid out in full so you can audit the method before you trust the output. Both are $10 a week into Bitcoin at a 0.1% fee, 53 buys across 364 days, valued at the closing price on the end date.
| 9 Jan 2017 → 8 Jan 2018 | 8 Jan 2018 → 7 Jan 2019 | |
|---|---|---|
| Weekly buys | 53 | 53 |
| Total invested | $530.00 | $530.00 |
| Average price over the window | $4,365.56 | $7,419.47 |
| Your average cost | $2,104.66 | $6,683.12 |
| Value at the end date | $3,820.16 | $319.22 |
| The same $530 as a single lump | $8,896.61 | $140.49 |
| Result — spread vs. lump | +620.8% vs +1,578.6% | −39.8% vs −73.5% |
The line that is a theorem, and the line that is not
Look at the two middle rows. In both windows your average cost came in below the average price over the same period — $2,104.66 against $4,365.56, and $6,683.12 against $7,419.47. That is not luck and it is not a property of Bitcoin. A fixed sum buys more units when the price is low and fewer when it is high, so what you pay is the units-weighted average, and units-weighted <= quoted average for any set of prices that are not all identical. It held in all eleven January-start years tested, because it has to.
Now look at the last row, which is where the honest version of this page departs from the marketing version. A lower average cost is not a higher return. In the 2017 window a single lump on day one returned 1,578.6% against the plan’s 620.8% — spreading the buys cost you money, badly. In the 2018 window the same lump lost 73.5% against the plan’s 39.8% — spreading the buys halved the damage.
So the thing DCA reliably buys is a lower entry price than the period’s average and a smaller hole in the years that go wrong. It does not buy a bigger number in the years that go right. Any tool that shows you only the first half of that is selling you something. This one draws the lump-sum line beside your plan on the free tier for exactly this reason.
03 — How it works
Every assumption is a field you can see.
Nothing is inferred about you and nothing about your plan is looked up. You describe the plan, the tool replays it against real daily closing prices, and it reports what that plan would have done — including when the answer is unflattering.
Step one
Pick the asset
The free tier opens on Bitcoin and covers 250 crypto assets. Stocks, index funds and commodities belong to the Pro database, and section 04 gives the counts plainly rather than leaving you to find them by clicking.
Step two
Describe the plan
How much per buy, how often, on which weekday, and what your exchange charges. The defaults are $10 weekly on a Monday at a 0.1% fee — deliberately small, so the mechanism stays legible before the numbers get large.
Step three
Choose the window
Start and end dates are yours to set, opening on the last twelve months. This is the field that decides your answer, which is the whole argument of section 01. Move it before you believe any single run.
Step four
Read it against the alternative
Tick the lump-sum line and the chart draws the counterfactual beside your plan — the same money committed on day one. A result you cannot compare to anything is not a result.
What Dynamic DCA actually does
The second free engine varies the size of each buy instead of holding it fixed, and the rule is worth stating exactly, because the name suggests something grander than the arithmetic. At each buy the tool takes the mean of every price it has seen so far in the run, measures how far below that mean the current price sits, and scales the purchase by 1 + deviation × band.
On the default Medium band that coefficient is 0.6, with a floor of 0.30 and a ceiling of 1.60. So a price sitting 20% under the running mean buys $11.20 instead of $10, and one sitting 20% over buys $8.80. The Low and High bands tighten or widen that response, the Exponential strategy doubles the coefficient, and Fear & Greed swaps the trigger from price to the sentiment index.
That is a mean-reversion tilt, and it is not the weekly risk reading the newsletter publishes. It sees only the price history inside your own simulation window, which means early in any run it has almost nothing to work with and sits close to a flat $10. Section 04 treats that as the objection it is.
Where the prices come from, and what this page records
Daily closes come from Yahoo Finance for listed assets and CryptoCompare for crypto history. They are real closing prices rather than a model, and the tool will tell you plainly when it has no history for a symbol across the dates you asked for.
One thing this page does that the other calculators on this site do not, and it is better said here than discovered later: each run is recorded. When you press simulate, this site writes a row holding the asset, the strategy, the amount, the resulting value and return, and the IP address the request came from. That is how usage is counted. Nothing you enter is sold or handed to a third party, and the only thing kept in your browser is which crash challenges you have finished — but “nothing is stored” would be untrue here, so this page does not say it.
Educational content only — not financial advice.
04 — Objections
Most complaints about this tool are correct.
A backtester that presents itself as a crystal ball is lying to you. Here are the four strongest arguments against what this page produces, including the two that are limits on the method rather than on the inputs.
“Backtesting Bitcoin’s best decade proves nothing.”
Correct — and the page is built around it
It proves nothing about the future and very little about the strategy. Any asset that rose over a window will make almost any accumulation plan look competent, and the free tier’s asset list is exactly the corner of the market where that distortion is largest.
This is why section 01 leads with two consecutive years that disagree by 660 percentage points on an identical plan. Run your own window, then run the year before it and the year after. If the conclusion moves, the conclusion was never about the strategy.
“Your Dynamic DCA is not a real risk model.”
Correct, and worth being exact about
It is a mean-reversion tilt. It compares the current price to the average of the prices already seen inside your simulation window and adjusts the buy by a bounded multiplier. It reads no valuation, no on-chain data and no macro input, and early in a run it has barely any history to average, so it sits close to a flat contribution.
It is deliberately not the weekly risk reading published in the newsletter. Treating the two as the same thing would overstate this engine and understate that one. What this engine is good for is testing whether buying more into weakness would have helped on your asset and your window — sometimes the honest answer is that it barely moved the result.
“The free tier is crypto only.”
Correct — 250 of 816 assets
Stated plainly rather than discovered halfway through a run. The free simulator carries 250 crypto assets and three engines — Equal DCA, Lump Sum and Dynamic DCA — plus 200 of the 293 historical crash challenges. Every stock, index fund and commodity in the database sits behind Pro, along with the other seven engines.
That is a real limit and it will make this tool the wrong one for some people, which section 05 says outright. If the equity side is what you came for, the honest move is to read the Pro page and decide there rather than to work it out by clicking locked controls.
“You are logging what I run.”
Correct, and disclosed above
Every simulation writes a row: the asset, the strategy, the amount, the resulting value and return, and the originating IP address. No name, no email and no account is attached to it, none of it is sold or passed on, and it exists to count usage rather than to profile anyone.
It is still more than the other calculators on this site keep, and several of those pages say so in the opposite direction because it is true of them. This page will not borrow their sentence. If that is not a trade you want to make, every figure in section 02 can be reproduced in a spreadsheet without running the tool at all.
05 — Who this is for
Built for people testing a plan they intend to keep.
This is a backtester for a recurring buy you are seriously considering, run several times with the window moved. If the question you actually have is a different one, something else will serve you better and it is cheaper to find that out here.
It fits if
- You are about to start a recurring buy and want to see how much of the outcome was the plan and how much was the entry date before you commit to it monthly.
- You already run a DCA and want to test one change — weekly against monthly, a different fee, a larger buy into weakness — against the same window rather than against a feeling.
- You want the lump-sum counterfactual drawn beside the plan, because you accept that a number with nothing to compare it to is not evidence.
- You are comfortable re-running with different dates and reading the disagreement between runs as the actual finding, rather than picking whichever run flatters the plan.
It does not fit if
- You want to backtest stocks, index funds or commodities on the free tier. You cannot — those are Pro. Section 04 gives the counts rather than making you find out by clicking.
- You want to know what happens next. Nothing here forecasts. It replays recorded prices, and the two runs in section 01 exist to show how little a single replay settles.
- You want to be told what to buy. The tool has no opinion about any asset and never will; it prices the plan you describe and stops there.
- You need a decision today from one run. The method requires several, and a single backtest is the one output this page argues you should distrust.
06 — Questions
The questions people ask before they trust the output.
Answered against what the tool actually does, not against what would be convenient to claim. Educational content only — not financial advice.
Does dollar-cost averaging actually work?
One thing about it is guaranteed and the rest is not. Spreading a fixed sum across many buys means you purchase more units when the price is low and fewer when it is high, so your average cost always lands below the average price over the same period unless every price was identical. That is arithmetic, not a market view, and it held in all eleven January-start years tested on this page.
What is not guaranteed is a better return. Over the twelve months from 9 January 2017, $10 a week into Bitcoin returned 620.8% while the same $530 committed on day one returned 1,578.6% — spreading the buys cost money. Over the twelve months from 8 January 2018 the plan lost 39.8% against the lump sum’s 73.5%. DCA reliably lowers your entry price and softens the bad years; it does not beat a lump sum in the good ones.
Is it better to dollar-cost average or invest a lump sum?
It depends entirely on what the price did after you started, which is the one thing you cannot know in advance. A lump sum puts every dollar to work on day one, so it wins whenever the asset rises from that point and loses hardest when it falls. Spreading the buys does the opposite: it gives up some of the upside in exchange for a smaller hole if the first months go against you.
The two runs above are the same money on the same asset one year apart, and they land on opposite sides of that trade. The useful question is not which method wins on average but which loss you would still be holding through. The simulator draws both lines on the same chart so the comparison is in front of you rather than in your head.
How do I backtest a DCA strategy?
Pick the asset, state the plan — amount per buy, how often, on which weekday, and the fee your exchange charges — then set a start and end date and let the tool replay it against real daily closing prices. The free tier opens on $10 a week into Bitcoin at a 0.1% fee across the last twelve months.
The step most people skip is the one that matters. Run it again with the window moved. A single backtest tells you what one start date produced, not what the strategy does; if the answer changes materially when you shift the dates by a year, the original number was a fact about that year. Pro sweeps every historical start date for you, but moving the field by hand costs nothing and makes the point just as well.
What is dynamic DCA?
It is a recurring buy that changes size instead of staying fixed. In this simulator each purchase is scaled by 1 + deviation × band, where the deviation is how far the current price sits below the average of the prices already seen in the run. On the default Medium band the coefficient is 0.6 with a floor of 0.30 and a ceiling of 1.60, so a price 20% under the running average buys $11.20 instead of $10 and one 20% over buys $8.80.
Be clear about what that is and is not. It is a bounded mean-reversion tilt calculated from prices inside your own window — no valuation, no on-chain data, no macro input — and early in a run it has little history to average, so it behaves almost like a flat contribution. It is not the weekly market risk reading published in the newsletter, and the two should not be treated as the same thing.
Is the DCA simulator free, and what does Pro add?
The simulator is free to use and the free tier is real rather than a demo: 250 crypto assets, three engines — Equal DCA, Lump Sum and Dynamic DCA — and 200 of the 293 historical crash challenges. None of it asks for payment or an account.
Pro widens the asset database from 250 to 816 by unlocking every stock, index fund and commodity, adds seven more engines — Buy The Dip, Future Projections, a ten-asset Portfolio, Take Profit, Monte Carlo, Value Averaging and Start-Date Analysis — and adds the benchmark comparison, inflation and currency controls, saved plans with alerts, and CSV, screenshot and embed export. It is $9 a month, $90 a year or $249 once. If you came to backtest equities, the free tier will not do it, and the Pro page is the honest place to decide.
Does the simulator store what I run?
Yes, and it is better said here than found out later. Each simulation writes a row recording the asset, the strategy, the amount, the resulting value and return, and the IP address the request came from. That is how usage is counted. No name, no email and no account is attached to it, and none of it is sold or passed to a third party.
The only thing kept in your browser is which crash challenges you have completed. Several other calculators on this site state that nothing you enter is written to a database, which is true of them and is not true here, so this page does not borrow the sentence. The arithmetic behind every figure on this page is reproducible in a spreadsheet if you would rather not run it at all.