What a recurring buy would be worth — and whether that start date was lucky
Buys happen at the daily close from Binance, with no trading fees, no spread and no slippage. Real exchange fees would lower every figure here by roughly the fee rate. Past prices are also survivorship-filtered by definition: coins that collapsed and were delisted are not in this list at all, so any "average coin" intuition drawn from these pages is biased upward.
The bigger caveat is the sample. Crypto price history is short, and overlapping windows make it look far longer than it is — which is why the number of non-overlapping windows is shown next to every spread. A backtest tells you what one particular past would have paid. It is not a forecast, and this page does not turn it into one.
Dollar-cost averaging means buying a fixed amount on a fixed schedule regardless of price. Put in an amount, a frequency and a period, and this page replays it against real Binance daily closes: how much you would have put in, how many coins you would hold, what it would be worth now, and what your average cost would have been. Everything runs in your browser from public market data.
The part most calculators leave out is that the start date does most of the work. "$100 a week into Bitcoin since 2020" and the same plan started eighteen months later are not two versions of one strategy — they are two different outcomes of the same strategy, and the gap between them is usually larger than any difference between strategies. So this page runs your plan from every possible start date in the coin's history and shows the whole spread: the worst, the median, the best, and how often it finished in profit at all.
It also checks the common claim that DCA beats investing a lump sum. That is a testable statement, not a given: spreading purchases lowers your average cost only if the price falls after you begin, and in a market that mostly rose, buying later means buying higher. The page reports how often each approach actually won across that coin's history, alongside the number of genuinely independent windows behind the figure — because crypto history is short, and overlapping windows make a small sample look like a large one.
⚠️ Not investment advice. A backtest describes one particular past with fees, spread and slippage excluded, and coins that failed and were delisted are absent from the data entirely. Past results carry no promise about the future. All decisions and risks are your own.
It replays a recurring fixed-amount buy against real Binance daily closes. Buys land on the close at each interval, so you get the total invested, the quantity accumulated, the average cost and what the position would be worth at the end of the window. Fees, spread and slippage are excluded.
Because in crypto the start date usually matters more than the strategy. The same plan begun a year apart can end up with completely different results, so a single hand-picked window mostly reflects the choice of window. Running the plan from every possible start date shows the worst, the median and the best outcome, which tells you whether the headline number was lucky.
Not automatically, and this page measures it rather than asserting it. Spreading purchases lowers your average cost only when the price falls after you begin. In a market that mostly rose, buying later means buying higher, and lump sum wins more often. What DCA reliably reduces is how much your entry timing decides the outcome, not the average outcome itself.
Overlapping windows re-measure the same history. Three years of daily data yields hundreds of one-year windows but only three that do not overlap, so the apparent sample size is far larger than the real one. The page shows the non-overlapping count next to every spread, and flags results below six as unreliable.
Usually the window, the price source or the buy timing. This page uses Binance daily closes and buys at the close of each interval; sites using a different exchange, a different close convention or a fee assumption will land somewhere nearby but not identical. Large gaps almost always come from a different start date rather than a different method.
No. A backtest describes one particular past, and crypto has only a few genuinely independent multi-year windows in total. Coins that collapsed and were delisted are also missing from the data, which biases any historical average upward. Treat these figures as a description of what happened, not a forecast.