Dollar-Cost Averaging vs. Lump Sum: What 100 Years of Data Really Says
If you have $50,000 to invest today, do you put it all in now or spread it out over 12 months? The math has a clear answer — and a more interesting nuance that most articles skip.
The two approaches
Lump sum (LSI): deploy the entire amount on day one.
Dollar-cost averaging (DCA): split the amount into equal chunks and buy on a fixed schedule (weekly, monthly) over a defined window — say 6 or 12 months.
Important distinction: true DCA means investing a lump sum gradually. Investing every paycheck as it arrives is periodic investing, not DCA — you have no choice, the money doesn't exist yet. We're talking about the lump-sum-already-in-hand case.
What the data actually shows
Vanguard's 2012 paper "Dollar-Cost Averaging Just Means Taking Risk Later" (and a 2023 update) tested LSI vs 12-month DCA across rolling periods on:
- US stocks 1926–2022 (~100 years of S&P data)
- UK stocks 1976–2022
- Australian stocks 1984–2022
- 60/40 stock/bond portfolios across all three markets
Why LSI wins on average
Markets go up most of the time. The S&P 500 has had a positive year in roughly 73% of years since 1926. Cash earning T-bill rates underperforms equities most years. Every month you delay deploying capital, you're statistically betting against the base rate.
...because the market's expected monthly return is positive.
The 32% where DCA wins
DCA outperforms LSI when the market falls during your DCA window. Each subsequent purchase happens at a lower price, so your average cost basis drops below where the lump sum bought.
Falling market scenario — 12-month DCA wins
The catch: you can't tell in advance whether the next 12 months will be the 32% case or the 68% case. If you could, you wouldn't be DCA-ing — you'd be timing.
Visualizing the distribution
The argument FOR DCA anyway
Vanguard's own paper concedes this point: DCA reduces regret risk and dispersion. The expected return is lower, but the worst-case outcome is also less bad.
If putting $200,000 in on day one and watching it drop 35% would cause you to panic sell, then a strategy that captures, say, 80% of LSI's expected return but you actually stick with is mathematically dominant in real terms.
Crypto: same math, different weights
Crypto's higher volatility makes both tails of the distribution fatter. DCA loses by more in bull runs (because BTC can do +200% in 12 months) but DCA also wins by much more in bear markets — and crypto bears go deeper than equity bears.
For BTC specifically, rolling 12-month DCA windows from 2014–2024 underperformed LSI in roughly 60% of windows but the average loss when DCA won was larger than the average loss when LSI won. Risk-adjusted, the two are closer in crypto than in equities.
The hybrid that's actually defensible
If you genuinely can't decide and don't want to model behavioral risk, a 50/50 split is rational:
- Invest 50% as a lump sum on day one.
- DCA the remaining 50% over 6 months.
This captures most of LSI's expected return while halving the regret if month 1 happens to be the peak. It's a math compromise, not a strategy that dominates either pure approach.
Practical decision framework
| If... | Lean toward |
|---|---|
| Long horizon (10+ yrs), tax-advantaged account, you won't watch it | Lump sum |
| Risk-averse, near retirement, large windfall | DCA over 6–12 months |
| You'd panic-sell after a 30% drawdown | DCA (or smaller allocation overall) |
| Markets near all-time highs and you're nervous | 50/50 hybrid (don't try to time) |
| Receiving paycheck contributions | This question doesn't apply — invest as money arrives |