The cheapest day for weekly bitcoin DCA doesn't stay the same
Split nine years of daily candles into three periods and each one crowns a different weekday, with at most a little over one percentage point between best and worst. At $100 a week for a year, that point or so comes to about $64 worth of bitcoin.
Over the full stretch since August 2017, buying bitcoin every Thursday got the most coin for each dollar; over the last 5 years alone, Saturday did; over the last 3 years, Tuesday. In all three periods, the best day bought less than 1.5% more bitcoin than the worst.
How much bitcoin each dollar buys on each fixed weekday
The data is Binance's BTCUSDT daily candles from August 17, 2017 to October 16, 2026, 3,348 in all. Each weekday gets its own run: every week on that day, the same amount buys bitcoin at that day's closing price; the bitcoin bought is added up and divided by the dollars put in, giving BTC bought per dollar; then the seven days are compared, with Monday set to 100. The last 5 years start from October 17, 2021, and the last 3 years from October 17, 2023.
Binance's daily candles split days in UTC. "Thursday" in the table means buying at the close at 24:00 UTC on Thursday.
| Fixed day | Full period (from Aug 2017) | Last 5 years | Last 3 years |
|---|---|---|---|
| Monday | 100.00 | 100.00 | 100.00 (lowest) |
| Tuesday | 99.93 | 99.94 | 101.23 (highest) |
| Wednesday | 99.57 (lowest) | 99.87 (lowest) | 100.69 |
| Thursday | 100.67 (highest) | 100.06 | 100.93 |
| Friday | 100.30 | 100.16 | 100.73 |
| Saturday | 100.27 | 100.35 (highest) | 100.74 |
| Sunday | 100.35 | 100.15 | 100.66 |
| Highest over lowest | 1.11% | 0.48% | 1.23% |
Wednesday comes last over the full period and the last 5 years, then climbs to fifth over the last 3 years, where Monday takes the bottom spot. Thursday ranks first over the full period and slips to fourth over the last 5 years.
The three periods are nested: the last 3 years fall within the last 5, and the last 5 inside the full period. Even with most of the data shared, the rankings keep shuffling, and each period's winner is no longer first in either of the other two. A win like that isn't a pattern you can carry forward.
One study says Monday is cheapest, another says Friday: which to believe
Both claims come from the same company, River. A 2023 article of theirs found that Monday was, in theory, the best day to buy over the previous year; their 2026 research used data from January 2023 to October 2025, tested hour by hour, and concluded that early Friday morning US Eastern time was best and Wednesday evening worst. The 2026 research also cautions that there's no guarantee early Friday will stay the best time (River's research).
The last-3-years column above overlaps River 2026's window by two years; compared at a single moment each day, the UTC close, the winner is Tuesday and the loser Monday. Neither should be taken as settled: both studies and this table only describe their own stretch of time.
The trouble with picking a weekday is that it gets you thinking about timing. This week it's which day; next month it's whether to hold off on this buy for a bit. To me, the answer is a day you won't forget and when the money is sure to be in the account, set to run automatically. One look at this table is enough.
Which day to set your weekly DCA: payday and balance come first
On these gaps, any day will do. What really throws results off is the money not being there on the day of the debit, so that week's buy doesn't happen. Pick the day with these in mind:
- Put it a day or two after payday or another regular deposit, so the balance is always there when the debit runs.
- Avoid days with large outgoings, such as when the credit card bill or rent is due.
- If the platform lets you choose the hour, check whether it's showing your local time or UTC.
Once it's on automatic, you no longer have to remember to place an order each week; schedule and funding are part of setting up automatic DCA on Binance. Once the day is set, leave it alone; when a study changes its answer, you don't need to change yours.
When the gap is big enough to care about
The 0.48% to 1.23% in the table comes from looking back after the fact and putting the best day against the worst. When you choose a day at the start, you don't know which day will win from then on, so what you can expect to gain by picking a day in advance is smaller than that.
Put it in money. At $100 a week for a full year, that's $5,200 in total, and 1.23% is 5,200 × 1.23% ≈ $64 worth of bitcoin; at $1,000 a week, a year is $52,000, and 1.23% is about $640 worth. Only at larger amounts does the gap start to carry weight.
Even at that size, choosing the day only moves you somewhere between best and worst. Missing a month of buys means about 4 fewer purchases, roughly 7.7% of the year's plan (4 ÷ 52); getting the weekday right adds at most 1.23% more bitcoin. If the weekly amount makes you want to stop the moment prices drop, work out first how to stick with DCA.
Binance is not available in every country, including the US. Disclosure: if you enter this site's Binance invite code when you register, Manfu may receive a referral fee, and you pay nothing extra. Investing carries risk; the content is for education only and is not investment advice.
Weekend prices, and splitting a week's buy across two days
Is bitcoin more expensive to buy on weekends?
Not in this data. In all three periods, Saturday and Sunday bought more bitcoin per dollar than Monday, and over the last 5 years Saturday bought the most of any day; over the last 3 years Sunday came second to last, still a little ahead of Monday.
Can splitting each week's buy across two days avoid the worst day?
It pulls the result to the middle of the two days, which also means you never get the best day. With equal amounts on both days, the combined BTC per dollar is roughly the average of the two days' figures. Over the full period, for example, Wednesday is 99.57 and Thursday 100.67; buying half on each gives about (99.57 + 100.67) ÷ 2 = 100.12. Buy an equal share on all seven days and the result lands on the seven-day average.