Research
Best and worst time to start DCA
Does it matter when you start? We ran $100 a month of dollar-cost averaging from every possible start month and held it for 3 years, for Bitcoin, Ethereum, the S&P 500 and gold. Each cell below is one start month.
Key findings
- The year you start matters more than the month. For every asset here, the gap between the best and worst start years was at least 9 times the gap between the best and worst calendar months. In the closest case, Ethereum, start years differ by about 238 percentage points a year and calendar months by about 26, so "start in January" patterns are mostly noise.
- Share of start months that ended in profit: Bitcoin 97%, Ethereum 88%, S&P 500 (Total Return) 88%, Gold 74%.
- The single worst start in this data was S&P 500 (Total Return) in Mar 2006, at -31.5% a year.
Updated with prices through 21 October 2026.
Bitcoin
97% of start months ended in profit after 3 years. Median +75.2% a year.
| Start year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2014 | +72 | +70 | +91 | +78 | +103 | +159 | +158 | +171 | +248 | +229 | +290 | +369 |
| 2015 | +404 | +322 | +345 | +256 | +291 | +250 | +214 | +229 | +211 | +186 | +168 | +112 |
| 2016 | +96 | +79 | +81 | +81 | +97 | +135 | +149 | +138 | +120 | +93 | +92 | +62 |
| 2017 | +50 | +65 | +48 | +19 | +36 | +40 | +28 | +42 | +40 | +29 | +46 | +72 |
| 2018 | +113 | +126 | +171 | +190 | +184 | +128 | +116 | +131 | +150 | +146 | +170 | +158 |
| 2019 | +130 | +101 | +107 | +103 | +77 | +50 | +15 | +27 | +14 | +10 | +12 | -3 |
| 2020 | -7 | +14 | +12 | +22 | +16 | +9 | +15 | +9 | -2 | +1 | +14 | +18 |
| 2021 | +25 | +23 | +51 | +60 | +45 | +57 | +50 | +52 | +41 | +46 | +56 | +86 |
| 2022 | +82 | +87 | +71 | +70 | +80 | +88 | +85 | +88 | +81 | +84 | +73 | +49 |
| 2023 | +46 | +30 | +15 | +15 | +23 | +14 | -1 | +1 | +13 | +16 |
Typical result by start month
Median annualised return across all the years, by the calendar month you start. Months with too few years show n/a.
- Jan+77
- Feb+75
- Mar+76
- Apr+74
- May+79
- Jun+72
- Jul+68
- Aug+70
- Sep+61
- Oct+65
- Nov+73
- Dec+72
Best starts
- Jan 2015: +404.3% a year
- Dec 2014: +368.9% a year
- Mar 2015: +345.0% a year
Worst starts
- Jan 2020: -6.5% a year
- Dec 2019: -2.8% a year
- Sep 2020: -1.9% a year
Ethereum
88% of start months ended in profit after 3 years. Median +43.1% a year.
| Start year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2016 | +239 | +142 | +138 | +132 | +137 | +187 | +191 | +144 | +107 | +100 | +94 | +63 |
| 2017 | +33 | +35 | +23 | -24 | 0 | +8 | +2 | +40 | +56 | +31 | +40 | +73 |
| 2018 | +92 | +159 | +176 | +206 | +262 | +244 | +211 | +236 | +295 | +269 | +309 | +314 |
| 2019 | +270 | +217 | +216 | +228 | +192 | +132 | +75 | +113 | +105 | +81 | +92 | +67 |
| 2020 | +54 | +71 | +64 | +65 | +51 | +44 | +38 | +27 | +11 | +12 | +9 | +11 |
| 2021 | +15 | +10 | +39 | +39 | +25 | +44 | +36 | +30 | +10 | +11 | +13 | +42 |
| 2022 | +35 | +30 | +6 | -4 | -6 | +16 | +12 | +37 | +52 | +51 | +39 | +14 |
| 2023 | +17 | -3 | -15 | -9 | -5 | -14 | -27 | -18 | -2 | +5 |
Typical result by start month
Median annualised return across all the years, by the calendar month you start. Months with too few years show n/a.
- Jan+44
- Feb+53
- Mar+51
- Apr+52
- May+38
- Jun+44
- Jul+37
- Aug+38
- Sep+54
- Oct+41
- Nov+40
- Dec+63
Best starts
- Dec 2018: +313.6% a year
- Nov 2018: +309.5% a year
- Sep 2018: +295.0% a year
Worst starts
- Jul 2023: -26.5% a year
- Apr 2017: -24.4% a year
- Aug 2023: -18.4% a year
S&P 500 (Total Return)
88% of start months ended in profit after 3 years. Median +12.1% a year.
| Start year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2000 | -17 | -18 | -18 | -16 | -11 | -7 | -5 | -5 | -2 | -1 | +2 | +3 |
| 2001 | +6 | +8 | +9 | +8 | +7 | +8 | +8 | +7 | +7 | +9 | +9 | +12 |
| 2002 | +13 | +12 | +13 | +11 | +10 | +13 | +12 | +14 | +12 | +12 | +10 | +13 |
| 2003 | +11 | +13 | +12 | +12 | +11 | +10 | +8 | +8 | +9 | +10 | +11 | +12 |
| 2004 | +13 | +14 | +11 | +12 | +14 | +16 | +14 | +12 | +11 | +14 | +12 | +10 |
| 2005 | +9 | +5 | +2 | +4 | +5 | +4 | -1 | -3 | -2 | -8 | -19 | -28 |
| 2006 | -21 | -26 | -32 | -25 | -20 | -14 | -15 | -11 | -9 | -7 | -6 | -1 |
| 2007 | -1 | -2 | 0 | +4 | +5 | -1 | -3 | +2 | +1 | +5 | +8 | +10 |
| 2008 | +13 | +16 | +16 | +17 | +19 | +16 | +18 | +15 | +10 | +6 | +11 | +12 |
| 2009 | +12 | +15 | +17 | +17 | +16 | +8 | +12 | +12 | +13 | +14 | +12 | +11 |
| 2010 | +11 | +15 | +15 | +16 | +16 | +18 | +16 | +19 | +15 | +17 | +19 | +20 |
| 2011 | +21 | +18 | +20 | +21 | +20 | +21 | +22 | +19 | +22 | +19 | +20 | +20 |
| 2012 | +19 | +16 | +19 | +17 | +18 | +17 | +15 | +15 | +7 | +7 | +12 | +12 |
| 2013 | +9 | +5 | +6 | +8 | +7 | +8 | +8 | +9 | +9 | +8 | +6 | +8 |
| 2014 | +9 | +10 | +13 | +11 | +12 | +13 | +12 | +13 | +13 | +13 | +14 | +16 |
| 2015 | +16 | +19 | +15 | +14 | +14 | +15 | +14 | +16 | +18 | +18 | +12 | +12 |
| 2016 | +5 | +10 | +12 | +13 | +13 | +8 | +13 | +12 | +11 | +11 | +13 | +14 |
| 2017 | +15 | +14 | +7 | -5 | +4 | +9 | +10 | +13 | +17 | +14 | +11 | +18 |
| 2018 | +19 | +19 | +21 | +22 | +24 | +24 | +25 | +26 | +27 | +23 | +27 | +24 |
| 2019 | +27 | +22 | +17 | +20 | +13 | +12 | +6 | +11 | +7 | 0 | +4 | +8 |
| 2020 | +3 | +8 | +4 | +6 | +6 | +6 | +9 | +10 | +9 | +5 | +3 | +8 |
| 2021 | +10 | +12 | +15 | +15 | +12 | +15 | +17 | +16 | +19 | +19 | +19 | +22 |
| 2022 | +20 | +21 | +20 | +15 | +15 | +18 | +20 | +20 | +22 | +23 | +23 | +22 |
| 2023 | +21 | +21 | +19 | +15 | +21 | +23 | +21 | +20 | +20 | +20 |
Typical result by start month
Median annualised return across all the years, by the calendar month you start. Months with too few years show n/a.
- Jan+12
- Feb+13
- Mar+13
- Apr+12
- May+12
- Jun+12
- Jul+12
- Aug+12
- Sep+11
- Oct+11
- Nov+11
- Dec+12
Best starts
- Jan 2019: +27.1% a year
- Sep 2018: +26.8% a year
- Nov 2018: +26.6% a year
Worst starts
- Mar 2006: -31.5% a year
- Dec 2005: -28.1% a year
- Feb 2006: -25.9% a year
Gold
74% of start months ended in profit after 3 years. Median +5.3% a year.
| Start year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2011 | -15 | -13 | -9 | -11 | -11 | -12 | -8 | -9 | -9 | -12 | -14 | -11 |
| 2012 | -12 | -6 | -9 | -9 | -10 | -9 | -9 | -13 | -10 | -10 | -8 | -12 |
| 2013 | -11 | -7 | 0 | -1 | +4 | 0 | +6 | +7 | +5 | +5 | +4 | -3 |
| 2014 | -4 | 0 | +2 | +2 | +2 | +3 | +2 | +3 | +6 | +4 | +3 | +3 |
| 2015 | +4 | +7 | +5 | +5 | +4 | +3 | +1 | -1 | -2 | -3 | -1 | -1 |
| 2016 | +2 | +3 | +1 | +1 | 0 | +2 | +6 | +8 | +12 | +10 | +11 | +9 |
| 2017 | +11 | +13 | +12 | +12 | +16 | +18 | +18 | +25 | +25 | +22 | +20 | +17 |
| 2018 | +19 | +17 | +11 | +11 | +12 | +16 | +11 | +12 | +11 | +8 | +9 | +7 |
| 2019 | +8 | +7 | +11 | +9 | +8 | +5 | +3 | +1 | -2 | -3 | -4 | +1 |
| 2020 | +2 | +5 | +2 | +6 | +6 | +5 | +3 | +4 | +4 | +1 | +5 | +8 |
| 2021 | +8 | +7 | +8 | +12 | +14 | +14 | +14 | +16 | +18 | +22 | +23 | +20 |
| 2022 | +19 | +23 | +23 | +29 | +30 | +31 | +31 | +30 | +32 | +39 | +39 | +42 |
| 2023 | +42 | +47 | +54 | +45 | +41 | +36 | +27 | +26 | +29 | +25 |
Typical result by start month
Median annualised return across all the years, by the calendar month you start. Months with too few years show n/a.
- Jan+5
- Feb+7
- Mar+5
- Apr+6
- May+6
- Jun+5
- Jul+6
- Aug+7
- Sep+6
- Oct+5
- Nov+5
- Dec+5
Best starts
- Mar 2023: +54.1% a year
- Feb 2023: +47.1% a year
- Apr 2023: +44.8% a year
Worst starts
- Jan 2011: -15.2% a year
- Nov 2011: -13.5% a year
- Feb 2011: -12.9% a year
How to read this
Each row is a start year and each column a start month. A cell is a plan that buys $100 every month for 3 years from that month, valued at the end in US dollars using daily closing prices, with no fees. The number is the plan’s annualised return, which takes into account that the money goes in gradually. Hover a cell for the exact start month and the total return. The S&P 500 figures are total return (dividends reinvested); gold uses COMEX front-month futures. Bitcoin starts from 2014 and Ethereum from 2016, skipping their earliest, very illiquid years.
Neighbouring cells overlap and share most of their history, so they are not independent results. The most recent start months are missing because the holding period has not finished yet. Bitcoin and Ethereum have short histories dominated by a few large cycles, so their patterns describe those cycles more than they predict anything.
Frequently asked questions
- What is the best month to start dollar-cost averaging?
- There is no reliable best month. In this data the calendar month you start in moves the result far less than the year you start in, and the month-to-month differences are small enough to be mostly noise. The practical answer is to start when you are ready and keep going.
- Does the year I start matter?
- Yes, much more than the month. Plans that started just before a long decline, such as the 2018 or 2021 crypto peaks, spent years underwater, while plans started near a low did far better. The heatmap above shows this as bands of colour running across whole years.
- How is each cell calculated?
- Each cell is a plan that buys the same amount every month for the holding period, starting in that month, valued at the end in US dollars with no fees. The number is the annualised return of that plan, which accounts for the money going in gradually.