The Principle and Limits of DCA

DCA (Dollar Cost Averaging) means splitting a fixed amount into regular purchases on a set schedule to spread out your average cost. The goal isn't maximizing return — it's removing the "when should I buy" timing decision entirely. This simulator reproduces DCA and lump-sum performance side by side using real historical data.

Research generally finds that lump-sum investing is mathematically favorable more often in a rising market, but DCA carries a real psychological and risk-management edge during volatile stretches. See a detailed comparison in DCA vs. Lump Sum.

Worked example — following the numbers

In assumed-return mode (no ticker — just an "annual return / volatility" input), contributing 500,000 KRW per month for 12 rounds (6,000,000 KRW total) with the calculator's defaults of 8% annual return and 15% annual volatility produces a DCA final value of 6,139,239 KRW (+2.3%) versus a lump-sum final value (the same 6,000,000 KRW invested all at once in round 1) of 5,864,614 KRW (−2.3%). DCA beat lump-sum here because later contributions bought in below the round-1 price, pulling the average cost down. The path starts at 101.99 in round 1, rises to 106.74 by round 4, falls to 91.72 by round 10, and ends at 99.68 in round 12. The resulting average cost of 97.00 sits below the final price of 99.68, so DCA gained; lump-sum put everything in at 101.99 and the final price finished below that, so it lost. (Assumed-return mode uses a fixed random seed, so identical inputs always produce these same numbers.) — see "How to read the result" below.

What's included, what isn't

Included: with a ticker, recent real prices; without one, a price path generated from your annual return/volatility inputs via a random-normal draw at each contribution date. Not included: trading commissions, taxes (dividend withholding included), and rebalancing.
How dividends are handled — the primary price feed for real-ticker mode returns adjusted closes, which already incorporate dividends and splits, so dividends are reflected in the price path itself. What the tool does not do is accrue dividend cash separately, convert it into reinvested shares, or compute dividend tax. If the primary feed fails and the tool falls back to a secondary source, the adjustment basis may differ. Assumed-return mode builds its path from your annual-return input alone, so dividends do not apply there at all.
The real-ticker mode re-fetches the latest price on every request, so the same inputs can give a different result depending on when you run it.

How to read the result

If DCA came out lower than lump-sum, that means the period was mostly trending up — not that DCA is an inferior strategy. In a steadily rising market, buying most of your shares earlier (lump-sum) wins. Conversely, if DCA beat lump-sum, that period had a decline or large swings, and buying more shares at lower prices along the way lowered your average cost. This tool doesn't prove which strategy is better — it shows the path both approaches took under the inputs you chose.

When your input is out of range

An invalid ticker format (special characters, excessive length) returns an "invalid ticker format" error. A contribution amount of 0 or less won't be calculated. The number of contributions must be between 2 and 360; otherwise it's an error. In real-ticker mode, if fewer than 10 recent price rows can be fetched for that ticker (a newly listed stock, a typo), the tool shows "could not load data" and does not proceed.

Assumptions and limits

Assumed-return mode always uses the same random sequence — the same annual return, volatility, and period always produce the same result on every run (it's not a fresh random simulation each time; it's a single reproducible example). Real-ticker mode is a plain price calculation that excludes taxes and commissions (dividends are already reflected in the adjusted closes, as described under "What's included, what isn't" above; what it does not do is accrue dividend cash separately or compute dividend tax). It also reproduces one specific historical window — choosing a different start date changes the result.