Dollar-cost averaging is buying a fixed amount on a fixed schedule regardless of price. It is recommended constantly, usually with the claim that it reduces risk or improves the average price paid. The first claim is true in a narrow sense. The second is false in general, and the arithmetic that shows why is short enough to do here.
The uncomfortable part first: if you already have the money, investing it all immediately beats drip-feeding it, most of the time, by a lot. That is not a market opinion. It follows from the assumption that markets rise on average, and it is arithmetic once you accept that assumption.
The short version
Money invested earlier is exposed to growth for longer. If the expected return is positive, every month you hold cash instead of investing it is a month of expected return you chose not to have. Averaging in guarantees that most of your money is invested later than it could have been.
What averaging in actually buys you is a narrower spread of outcomes — you cannot put everything in the day before a crash, because you never put everything in on any day. That is real, and for some people it is worth paying for. It is just not free, and the price is measurable.
Worked example, with every number shown
500 a month for ten years, an expected 7% a year, in a fund charging 0.5% a year, with inflation running at 2.5%.
| Input | Value |
|---|---|
| Contribution | 500, monthly |
| Term | 10 years |
| Expected annual return | 7% |
| Annual fee | 0.5% |
| Annual inflation | 2.5% |
periods = 12 × 10 = 120
paid in = 500 × 120 = 60,000
final value = 83,351.24
growth = 83,351.24 − 60,000
= 23,351.24So 60,000 of contributions becomes 83,351.24. Now the same 60,000 invested on day one instead.
The lump sum ends 29,277.01 ahead — 35% more than the drip-fed pot, on identical assumptions, identical fees and identical total contributions. The only difference is when the money was exposed.
Why that comparison is usually unfair
Look at what the lump-sum column assumes: that you had 60,000 in hand at the start of year one. If you did, the comparison is fair and the answer is invest it. If you did not — if the 500 a month is your salary arriving 500 at a time — then there was never a lump sum to invest, and the comparison is between averaging in and doing nothing.
That is the case for most people, and it is why the DCA-versus-lump-sum argument generates so much noise: the two sides are usually answering different questions.
- You have a windfall and are deciding how to deploy it. The arithmetic above applies. Averaging in costs expected return in exchange for a narrower range of outcomes.
- You are investing income as it arrives. There is no choice being made. You are averaging in because that is the shape of the money, and the lump-sum figure is a fiction.
The genuinely useful question in the second case is not DCA or lump sum. It is how much of this is being eaten by fees and inflation — which is the part almost nobody computes.
What the fee actually costs
Run the same schedule with no fee and no inflation.
2,174.63 over ten years, from half of one percent a year. That is 3.6% of the total contributed, and 9.3% of all the growth the portfolio produced. The fee is charged on the balance, so it grows as the balance grows — it takes its cut of the compounding, not just of the contributions.
The comparison people usually make is 0.5% against 0.25%, and it looks like nothing. Over ten years on this schedule the gap between those two is most of a thousand. Over thirty years it is not close.
What inflation actually costs
The same run reports both a nominal and a real figure:
nominal = 83,351.24
in today's money = 65,113.85
power lost = 18,237.39The pot grows to 83,351.24 and buys what 65,113.85 buys today. The inflation adjustment is seven times larger than the fee, and it is the number that answers the question people actually have, which is what the money will be worth rather than what it will say on the statement.
Note that this is not a loss the fund caused. It is the yardstick moving. But a plan that targets a nominal figure — “I want 80,000 in ten years” — is targeting the wrong number, and the size of the error here is 22%.
The mistakes that actually cost money
Treating averaging in as risk reduction rather than risk deferral. Once the money is fully deployed, you hold exactly the position you would have held anyway. The protection lasts only as long as the deployment period, and after that it is gone.
Comparing DCA to a lump sum you never had. If the money arrives monthly, the lump-sum figure is not an alternative you rejected. It is a different life.
Quoting nominal returns over long horizons. Over ten years at 2.5%, a fifth of the number is gone. Over thirty it is more than half.
Assuming a smooth return. Everything above uses a constant 7%. Real markets do not deliver a constant anything, and the order of returns matters enormously while you are contributing — a bad first decade with a good second is a very different outcome from the reverse, even at the same average.
What this does not model
The arithmetic assumes a fixed expected return, a fixed fee taken annually, contributions arriving on schedule and never missed, no taxes, no transaction costs per contribution, and no rebalancing. It says nothing about whether 7% is the right assumption for any particular market or decade — it takes the number you give it and shows the consequence.
Sequence-of-returns risk is the big omission. A constant-return model systematically understates how much the order of returns matters to someone contributing over time, and that is precisely the risk averaging in is sometimes claimed to address. This model cannot settle that argument, and any tool that says it can is showing you a straight line and calling it a market.