Say an inheritance or a bonus lands 100 million KRW in your account. Do you put it into an ETF all at once (lump sum), or spread it over 12 months at 8.33 million per month (DCA, dollar cost averaging)? Intuition says spreading it out to "average down your cost basis" is safer. But Vanguard's classic 2012 study (replicated several times by Of Dollars and Data) found that lump sum beats DCA roughly 2/3 of the time. Why is that — and why does DCA still get recommended anyway?
Deploy the entire cash amount into the market immediately. 100% exposure from day one.
Invest the same total in equal installments over N months. On average, you spend those N months only 50% exposed.
The core difference is time spent exposed to the market. With DCA, roughly half of your money sits in cash for the duration.
Equity markets trend upward over the long run (historically, the S&P 500 has averaged about +10% per year; cash and short-term bonds about +1–3%). That means every month spent holding cash has a lower expected return than the market. DCA parks half of your money, on average, in the low-yield asset (cash) for N months.
With a 10% expected S&P 500 return, 2% on cash, and a 12-month DCA: half the money sits out for six months at an 8pp opportunity cost — roughly 1/8 of the total, or about 4 million KRW of expected shortfall on 100 million. That is the entire source of lump sum's statistical edge.
| Study | Period / market | Lump sum win rate |
|---|---|---|
| Vanguard 2012 | 1926–2011 US/UK/AUS | ~66–67% |
| Of Dollars and Data 2017 | 1920–2017 S&P 500 | ~75% (vs 24-month DCA) |
| Vanguard 2023 (replication) | 1976–2022 multi-country | ~68% |
It is nothing more than a restatement of a simple proposition: markets tend to go up over time.
If the statistics favor lump sum, why doesn't everyone recommend it? Three reasons.
If the market crashes −30% the day after you go all in (as in 2008 or 2020), most investors can't stomach the self-blame and sell at the bottom. Sell, and you miss the recovery (the S&P 500's COVID round trip took five months). DCA offers the comfort of "half is still waiting on the sidelines," which lowers the odds of panic-selling in a crash. Even if the statistical expected return is lower, the outcome including actual behavior can be very different.
The 1/3 of periods where lump sum lost in the Vanguard study are exactly the ones where you went all in right before a market top. Entering just before March 2000 (dot-com), October 2007 (financial crisis), or November 2021, DCA won by a wide margin. Since you can't know the timing in advance, DCA is insurance against that "unlucky one-third."
Auto-investing a fixed share of every paycheck isn't actually DCA — it's simply "invest money as soon as it arrives." True DCA means deliberately splitting up a lump sum you already have. Conflating the two muddles the analysis.
There are also middle grounds between full DCA (12 equal monthly installments) and a full lump sum.
The "mathematically optimal" choice is lump sum. The "psychologically survivable optimum" is DCA or tranches. If you're confident you won't panic-sell in a crash, go lump sum; if not, DCA. The answer lies not in the statistics but in your own behavioral pattern.
Multifolios records buyLots (purchase lots) per holding in chronological order. Useful for DCA users:
The key question in analyzing DCA's effect — "how does my average cost compare to the simple average price over the same period?" — can be computed directly from the buyLots data.
Lump sum is the math answer (given that markets trend up); DCA is the psychology answer (avoiding panic-sells in a crash). Choose based on which one you're more vulnerable to. Periodic contributions and DCA are different concepts — investing money as soon as it arrives is closer to lump sum.