Investing all at once beat spreading it over a year 77% of the time
Since 1993, putting money into SPY in one go beat investing it in 12 monthly parts in 77% of start months, by a median 4.4% after a year. Spreading it out halved the worst first years.
- Start months where the lump sum won
- 77%
- Worst 5%, lump sum
- −20%
- Worst 5%, monthly
- −10%
- Won, S&P 500 since 1950
- 65%
- Won, ten years later
- 76%
What we found
From every month since January 1993, we put $10,000 into SPY either at once or in 12 equal monthly parts, with the rest waiting in T-bills. A year after the start, the lump sum was ahead in 77% of 395 starts, by a median 4.4%. Ten years on, it was still ahead 76% of the time.
Spreading it out bought a smoother first year: the worst 5% of starts lost 10% or more that way, against 20% or more all at once, and 5% of starts lost over a tenth of the money against 11%. The lump sum lost most when it went in just before a crash, as in February 2008.
The result holds on the S&P 500 since 1928 (62%) and since 1950 (65%), even though those use prices without dividends, which flatters the monthly plan.
The full results
- A year later
- Ten years later
- All at once
- 12 monthly parts
The 5th percentile of returns a year after the start.
| Data | Starts | Lump sum won | Median lead | Worst 5%: lump | Worst 5%: monthly | Lost 10%+: lump | Lost 10%+: monthly |
|---|---|---|---|---|---|---|---|
| SPY with dividends, 1993– | 395 | 77% | +4.4% | −20% | −10% | 11% | 5% |
| S&P 500 prices, 1950– | 911 | 65% | +2.5% | −17% | −9% | 12% | 5% |
| S&P 500 prices, 1928– | 1175 | 62% | +2.3% | −25% | −14% | 17% | 8% |
Uninvested cash earns the 3-month T-bill. On prices without dividends the lump sum misses more income, so those rows flatter the monthly plan.
| Data | Starts | Lump sum won | Median lead | Worst 5%: lump | Worst 5%: monthly | Lost 10%+: lump | Lost 10%+: monthly |
|---|---|---|---|---|---|---|---|
| SPY with dividends, 1993– | 287 | 76% | +4.3% | −8% | −7% | 4% | 4% |
| S&P 500 prices, 1950– | 803 | 66% | +2.7% | −4% | −2% | 4% | 4% |
| S&P 500 prices, 1928– | 1067 | 63% | +2.5% | −24% | −20% | 8% | 6% |
Lost 10%+: worth less than 90% of the money put in.
What spreading it out buys
Investing in parts holds about half the money in cash over the first year, so it naturally has about half the risk in that year: smaller losses in a crash and smaller gains otherwise. Over time the stocks are owned either way, and ten years on the worst outcomes were about the same.
So the choice is about regret, not returns. Investing all at once wins about three times in four; spreading it out protects against the one start in four that would feel worst.
How it was measured
- SPY with dividends, month-end closes, from January 1993. Uninvested cash earns the 3-month T-bill rate (FRED TB3MS). Compared 12 and 120 months after the first part.
- No costs or taxes.
Caveats
- A lump sum wins because stocks usually rise, so money waiting to be invested usually misses gains. In the minority of starts just before a bear market, spreading it out did better, and that is the case people worry about.
Questions
- Is lump sum better than dollar-cost averaging?
- Usually. Since 1993, investing in SPY all at once beat 12 monthly parts in 77% of start months, by a median 4.4% after a year, and was still ahead 76% of the time after ten years.
- When does dollar-cost averaging win?
- When a bear market starts soon after the money is ready, as in 2000 or 2008. It also halved the worst first-year losses, from about 20% to about 10%.
More research
- Market factBuying at an all-time high did no worse than buying any other month
- Market factWe tested 24 Supertrend settings. None beat holding the S&P 500
- Market factOnly 20% of US stocks beat the S&P 500 over the last ten years
Historical analysis. Past performance does not guarantee future results. Not investment advice. Published Nov 2026. Data: SPY and FRED to Nov 6, 2026.