How Often Should You Dollar-Cost Average?
Weekly, biweekly and monthly plans, run on the same 10 years of real daily closes with the same total contribution. The table below is the whole answer — and it is smaller than most people expect.
Same money, three cadences — SPY
$1,200 per year into SPY for 10 years (Aug 2016 → Jul 2026), split as $23.08 weekly, $46.15 biweekly, or $100 monthly. Real daily closes, no fees modelled.
Same money, three cadences — BTC
$1,200 per year into BTC for 10 years (Aug 2016 → Aug 2026), split as $23.08 weekly, $46.15 biweekly, or $100 monthly. Real daily closes, no fees modelled.
Across 10 years, the gap between the best and worst cadence was 1.0 percentage points on SPY — and about as small on BTC. Cadence is a rounding error; whether you automate the plan is the whole game.
Data through Jul 2026, refreshed weekly.
Why the cadence barely matters
Dollar cost averaging works by spreading purchases across time so no single day's price decides your outcome. Once you buy 12+ times a year, you have already captured almost all of that spreading benefit — going from monthly to weekly multiplies your number of buys by four but changes the average purchase price by a fraction of a percent, because adjacent weeks' prices are highly correlated.
What does change with frequency is friction: more buys mean more per-trade fees (where they exist), more bank transfers, and more chances to fiddle with the plan. If your broker charges per transaction, monthly keeps costs lowest; if buys are free and automatic, pick whatever aligns with your payday.
Does the day of the month matter?
No — and this is measurable. Salaries cluster around month-ends, so many investors buy on the 1st; others swear by mid-month dips. Across long windows, first-of-month, mid-month and last-trading-day schedules land within noise of each other, for the same reason cadence barely matters: daily prices inside a month are too correlated for the slot to matter.
The practical rule this data supports: schedule the buy for the day after your salary arrives, automate it at your broker or exchange, and stop thinking about it. Every hour spent optimizing the calendar is an hour not spent on the two decisions that do move outcomes — how much you contribute, and whether you keep going through a drawdown.
FAQ
As often as your money arrives, in practice: monthly fits most salaries. On 10 years of real market data, weekly, biweekly and monthly plans with the same yearly total end within roughly one percentage point of each other, so cadence should be chosen for convenience and fees, not returns.
The return difference is negligible on long windows — our 10-year backtests above show the spread. Weekly buys marginally smooth volatility further but multiply transaction count by four; monthly is the default that minimizes fees and effort. If your exchange charges no fees, either is fine.
There is no reliably best day. Backtests of first-of-month, mid-month and month-end schedules on broad indexes land within noise of each other. Pick the day after your paycheck clears so the plan never competes with spending.
Changing cadence mid-crash is timing wearing a disguise. If you want more exposure during drawdowns, the honest tool is contribution size, decided by rule in advance (for example, +50% while the asset sits 30% below its high) — not ad-hoc extra buys your future self may not repeat.
Run the plan on a real exchange
Five exchanges where DCA — and tokenized stocks & ETFs — actually work, each with a real discount. Pick one, copy the code, trade cheaper.
Referral links support the site and never affect the math. Availability varies by country — check your local regulations.