Dollar-cost averaging vs lump sum: what the historical data shows

Take 120,000 dollars that arrived on October 1, 2007, from an inheritance or the sale of a business, and two ways to put it into an S&P 500 index fund. The first investor buys everything that day. The second buys 10,000 dollars a month for a year and keeps the rest in Treasury bills. Twelve months later, with dividends reinvested, the first portfolio was worth 77,161 dollars and the second 87,898 dollars.

Start the same experiment in March 2009 and the order flips: 187,047 dollars for the lump sum against 143,242 dollars for the monthly schedule, a gap of almost 44,000 dollars.

Across every month from January 1934 to July 2025, the lump sum finished the year ahead in 773 of 1,099 starting points, or 70.3%. On US, UK, Canadian, European, Australian, emerging and global stock indexes, the lump sum beat a 3 to 6 month schedule in 61.6% to 73.7% of one-year periods up to 2022, depending on the market and the length of the schedule. In the remaining periods the investor who spread the money out ended ahead, by as much as 32.7% of the sum for a start in August 1937 and 25% to 31% for starts from May to September 2008.

Lump-sum investing

Putting an amount that is already available in cash into the target portfolio in a single transaction.

Dollar-cost averaging (DCA)

Splitting that same amount into equal installments and investing them on a fixed schedule, for example monthly over 3, 6 or 12 months, while the uninvested balance waits in cash. Vanguard calls this cost averaging, and in the UK it is called pound-cost averaging.

Periodic investing from income

Investing a fixed amount from each paycheck, as in a 401(k). It uses the same mechanics as DCA, but no lump sum exists to invest earlier.

Studies of lump sum vs dollar-cost averaging since 1979

The theoretical result dates from 1979: a fixed, pre-committed purchase schedule is suboptimal, because an investor who can revise the plan each month can always do at least as well as one who cannot.

Longer schedules raised the lump sum's win rate: against a 36-month schedule it came out ahead in about 90% of US 10-year periods. The same result held for an all-bond portfolio, where the lump sum won 65% of US periods, because bonds also earned more than cash over the sample.

One year after a three-month schedule, at the median, the lump sum left an all-equity portfolio 2.2% richer and a 40/60 portfolio 1.2% richer. For a 60/40 portfolio the average ending value was 109,360 dollars against 107,453 dollars, 1.8% more. With Treasury bill interest paid on the waiting cash, the lump sum still won 65% of the time for an all-equity portfolio.

The S&P 500 test, 1934 to 2025

The test in the last row of the table uses Robert Shiller's monthly S&P 500 series, with dividends reinvested, and pays the cash waiting in the DCA account the three-month Treasury bill rate, which starts in January 1934. Each month from then until July 2025 is a starting point. The lump sum puts 120,000 dollars into the index at once. The DCA investor puts in 10,000 dollars that month and in each of the next 11. Both accounts are valued 12 months after the start. The Shiller series uses monthly average prices, so single-day crashes and rallies are smoothed out.

• The lump sum ended ahead in 773 of the 1,099 starting months, or 70.3%.
• In the months it won, it won by an average of 9.0% of the sum, about 10,800 dollars.
• In the months DCA won, it won by an average of 7.0%, about 8,350 dollars.
• With the waiting cash earning nothing, as in a checking account, the lump sum's win rate rose to 75.5%.
• With a 6-month schedule the win rate fell to 66.0% and the median gap to 2.0% of the sum. With a 24-month schedule, measured after two years, it rose to 73.6% and the median gap to 8.8%.

The win rate varied with the decade of the start date, from 55.8% in the 1970s to 86.7% in the 1990s:

Source: MoneyScience calculations from Shiller S&P 500 data and FRED TB3MS. 120,000 dollars, 12 monthly installments, dividends reinvested, no fees or taxes.

In the 2000s the lump sum still won more often than it lost, 56.7% of the time, yet DCA came out ahead on average by 0.5% of the sum. DCA won by more than 10% of the sum in 26 starting months of that decade, all between July 2000 and September 2008, and by 30.7% for a start in August 2008. In the 1970s, three-month Treasury bills paid an average of 6.3% a year, the S&P 500 price rose from 90 to 108 over the whole decade, and the lump sum's average lead was 0.8%.

Four start dates on the S&P 500

Source: MoneyScience calculations from Shiller S&P 500 data and FRED TB3MS, dividends reinvested.

For the October 2007 start, by March 2009, 17 months in, the lump sum account had fallen to 61,157 dollars and the DCA account to 69,968 dollars. Both had lost more than 40%, because the DCA investor had finished buying in September 2008, six months before the low. The DCA account climbed back above 120,000 dollars in January 2011. The lump sum account took until August 2012.

For the March 2009 start, every monthly installment after the first bought at a higher index level than the lump sum paid, between 848 in April 2009 and 1,124 in January 2010, against 757 in March 2009. The 43,805 dollar gap is 36.5% of the sum.

Why the lump sum usually wins

A DCA investor holds part of the money in cash for the length of the schedule. Over a 12-month schedule the average cash balance is close to half the sum, so for that year the portfolio carries about half the intended exposure to stocks. From 1976 to 2022, US stocks outperformed three-month Treasury bills in 76% of the time, and bonds 68% of the time. Each month spent in cash gives up that premium on the uninvested part.

Three variables set the size of the gap:

1. The length of the schedule. A 6-month schedule leaves money in cash for half as long as a 12-month one, and the median gap in the S&P 500 test fell from 4.3% to 2.0% of the sum.
2. The interest paid on cash. Removing Treasury bill interest raised the lump sum's win rate from 70.3% to 75.5% in the same test. In the 1970s, when bill rates were high, the lump sum's average lead fell below 1%.
3. The share of stocks in the target portfolio. The more of the portfolio that goes into stocks, the more premium the cash forgoes: 2.2% for all equity against 1.2% for a 40/60 mix in global data from 1976 to 2022.

The lower average cost argument

DCA is often sold with a true arithmetic fact: a fixed dollar amount buys more shares when the price is low and fewer when it is high, so the average cost per share ends up below the average price over the period. The fact holds whenever prices move at all, and it has no bearing on expected return, because the lump sum investor bought every share at the first price and the relevant question is which account ends up larger. Treating the lower average cost as a gain is a false comparison, and a strategy using less capital can reproduce the same range of outcomes as DCA.

From October 2007 to September 2008, twelve 10,000 dollar purchases of the S&P 500 bought at an average cost of 1,361 index points against an average monthly price of 1,367, a difference of 0.4%. The 10,737 dollar advantage over the lump sum in that year came from having less money in the market during the fall.

What dollar-cost averaging reduces: drawdowns and regret

Over 1,021 rolling 12-month periods in the US from 1926 to 2011, a 1 million dollar lump sum lost value in 229 periods, or 22.4%, and a 12-month DCA schedule in 180, or 17.6%. The average loss in those periods was 84,001 dollars for the lump sum and 56,947 dollars for DCA. The S&P 500 test shows the same pattern: the lump sum lost money over 12 months in 22.7% of starting months and DCA in 19.0%, and the worst results on 120,000 dollars were 70,911 dollars and 83,541 dollars.

Over rolling 12-month periods, the average annualized Sharpe ratio of an all-equity US portfolio was 0.77 with a lump sum and 0.68 with DCA, and the lump sum led in all nine market and allocation combinations tested.

Using the CRSP database from 1926 to 2003, Michael Brennan, Feifei Li and Walter Torous compared certainty equivalent wealth, the sure amount an investor would accept in place of a risky outcome. For a diversified stock portfolio, DCA over one to six years beat the lump sum on that measure for all except the most risk tolerant investors, while the lump sum kept the higher expected return. A 50% stock and 50% cash portfolio rebalanced monthly beat both, with almost the same expected return as DCA and lower risk.

For the length of the schedule, a DCA investor holds less stock than the target allocation. An investor who wants that lower level of risk permanently can set a lower target and invest it at once. Discomfort with the risk of a lump sum entry can signal a target allocation set too high.

Loss aversion and the case for spreading the money

Loss aversion is the tendency to weigh a loss more heavily than a gain of the same size. A lump sum invested in the S&P 500 in October 2007 was down 49% at the March 2009 low, and the whole loss followed from a single purchase date. The research case for DCA rests on that risk: spreading the entry limits the downside and the regret of a badly timed start, at a cost in expected return.

In a utility model with a loss aversion coefficient of 2.5, the preference switches for some investors. Without loss aversion, a moderately conservative investor preferred the lump sum, with a certainty equivalent return of 1.0443 against 1.0421. With loss aversion, the same investor preferred DCA, by 1.0102 against 1.0100, a difference of 0.02 percentage points. The very conservative profile preferred DCA in both cases, and the adventurous profile preferred the lump sum in both. For loss-averse investors who choose DCA, the same paper recommends a short schedule, such as three months, to limit the opportunity cost.

DCA has been recommended to investors since at least the 1940s, and Meir Statman's 1995 behavioral framework for dollar-cost averaging gives four behavioral explanations for its popularity:

• Prospect theory framing. Investors judge outcomes as gains and losses against a reference point. In the paper's two-period example, 2,000 dollars invested in two halves at 50 dollars and 12.50 dollars a share is worth 1,250 dollars, a loss, yet the average cost of 20 dollars sits below the average price of 31.25 dollars, which reads as a gain.
• Aversion to regret and responsibility. Following a rule fixed in advance leaves the investor less responsible for a bad outcome, so a fall after an installment brings less regret than a fall after a single lump sum decision.
• Cognitive errors. Investors extrapolate recent price trends, so after several falls they expect more and are tempted to stop buying.
• Self-control. A strict schedule keeps the investor buying in a falling market, when that temptation is strongest.

A hesitant investor can also leave the money in cash. In MSCI World data from 1976 to 2022, a three-month DCA schedule beat staying entirely in cash 69% of the time over one year, and a lump sum beat cash 70% of the time. Regret and trend extrapolation also appear among the behavioral biases that cost investors money.

401(k) contributions: dollar-cost averaging by default

A 401(k) takes a fixed share of each paycheck and invests it on payday, whatever the market is doing. The 2026 limit on employee deferrals is 24,500 dollars, plus an 8,000 dollar catch-up from age 50 and 11,250 dollars at ages 60 to 63. In plans run by Vanguard, participation among eligible employees reached 86% in 2025, the average savings rate was 12.1%, and 5% of participants traded during periods of volatility.

The lump sum studies do not apply here, because the money does not exist in advance. Investing each paycheck as it arrives is the alternative to accumulating the money in cash and trying to time the market later. Warren Buffett described the same approach in his 1993 shareholder letter, where he advised investors who do not analyze individual businesses to own many stocks and space out their purchases, for example through regular investments in an index fund.

Front-loading applies the lump sum result to a 401(k). Paying the full year's contribution as early in the year as possible adds months in the market, which in the lump sum research points to a higher median retirement balance. It requires enough cash to live on while the early paychecks go into the plan.

Five factors in choosing a lump sum or a DCA schedule

The win rates above average more than a thousand start dates. A single investment has one start date, and five factors bear on how it turns out:

1. What the investor would do after a large fall. The lump sum's advantage exists only for an investor who holds through a year like October 2007 to October 2008, when 120,000 dollars became 77,161 dollars. Selling at that point locks in the loss that DCA would have cushioned.
2. The length of the schedule. Every extra month in cash raises the lump sum's win rate, from 66.0% for 6 months to 73.6% for 24 months in the S&P 500 test, and above 90% for 36 months in US data from 1926 to 2011.
3. The rate on the waiting cash. Cash earning close to the three-month Treasury bill rate, 3.72% in August 2026, narrows the gap: in the S&P 500 test, the lump sum won 70.3% of the time against cash earning bill rates and 75.5% against cash earning nothing.
4. The size of the sum relative to existing savings. The choice between the two makes a marginal difference compared with keeping the money in cash for good, and the difference is smaller still when the lump sum is a small fraction of total wealth.
5. Taxes on the windfall itself. Some windfalls create a tax bill, and the amount needed to pay it is money to hold back from investing under either approach.

After the entry, the fund chosen sets the fees paid every year, and our comparison of active and passive investing sets out what those fees have cost.

This article is general information and does not constitute investment advice. Past returns of an index do not predict future returns.

Frequently asked questions

Is dollar-cost averaging better than investing a lump sum?

On historical returns, the lump sum has come out ahead in roughly two out of three periods. In S&P 500 data from 1934 to 2025, a sum invested at once was worth more after 12 months than the same sum fed in over 12 monthly installments in 70.3% of 1,099 rolling windows. Dollar-cost averaging lost less money in the bad windows: the worst 12-month result on 120,000 dollars was 70,911 dollars for the lump sum and 83,541 dollars for the monthly schedule.

How much more does a lump sum typically earn than dollar-cost averaging?

A few percent of the amount invested, and more the longer the schedule runs. For a US 60/40 portfolio held 10 years after a 12-month schedule, from 1926 to 2011, the lump sum ended with 2.3% more wealth on average. On the S&P 500 with dividends reinvested and cash earning Treasury bill rates, the median lead after 12 months was 4.3% of the sum from 1934 to 2025.

Is contributing to a 401(k) from each paycheck dollar-cost averaging?

Yes. A fixed amount goes into the plan every pay period whatever the market does, which is the mechanical definition of dollar-cost averaging. The lump sum comparison does not apply to it, because the money did not exist as a lump sum: it arrives with each paycheck and is invested as it arrives. The 2026 limit on employee deferrals is 24,500 dollars, with an 8,000 dollar catch-up from age 50.

Does Warren Buffett recommend dollar-cost averaging?

In his 1993 letter to Berkshire Hathaway shareholders, Buffett wrote that an investor who does not understand the economics of specific businesses should own a large number of stocks and space out purchases, for example by investing in an index fund periodically. That describes regular investing over time. The letter does not address what to do with a windfall that is already in cash.