Dollar Cost Averaging — Does It Really Work — Bitcoin Case Study
Let’s say you have some money, have decided you want to invest it, and also already know what you want to invest in. It could be anything but let’s say it’s bitcoin. What’s the smart way to do it. Do you invest the entire amount up front, or do you spread the purchase over a number of days, weeks, or months in order to get the benefit of so called dollar cost averaging.
Dollar cost averaging is often touted as a good solution to the dilemma that if you buy now you might be buying everything “at a high”. What if the market dips right after you buy. In this scenario dollar costs averaging allows you to buy some now and then some later at the new lower price getting you in at a better average price. If the price doesn’t dip then you buy some now and some later at the higher price. Your average price is higher but you are still happy because the price is up and you have made money. Maybe dollar cost averaging is all about pain and regret minimisation rather than return maximisation. If price goes up afterwards you are happy that you at least got in early with a partial puchase before the price rise. If price goes down afterwards you are happy that you can now buy more of something you wanted anyway but now at a lower price. Either way you are happy, or at the very least have a way to rationalise that you did the right thing.
Dollar cost averaging seems to make sense, but what effect does it really have on your expected future returns. I decided to apply dollar cost averaging to bitcoin specifically to see how it compares to buying everything at once.
For each day in the past I compared the returns up until the present (5th April 2019) had you invested all at once on that day, versus dollar cost averaging the investment over the next 12 months in 12 equal instalments. So for example if today was 1st Jan 2017 and you had $12,000 to invest was it better to invest the full $12,000 on 1st Jan 2017 or was it better to invest $1,000 every month for the next 12 months. In this example the future returns for dollar costs averaging come to 174% but had you purchased everything at once the future returns come to 422%.
The chart below shows green when dollar cost averaging resulted in better future returns and red when buying all at once resulted in better future returns.
The next chart shows the same data but this time using a log scale to make it easier to read. Remember green is where dollar cost averaging is better, red is where buy at once is better.
If you count up all the red and green days it turns out 27% of the days are green. So most (around 73%) of the time you are better off investing all at once. On average dollar cost averaging is worse than buy at once but we can go deeper by looking at the specific conditions when dollar cost averaging does work better. First let’s look at the size of the differences in future returns.
The red line shows where there is zero difference. Above the red line is where dollar cost averaging is better and by how much. Below the red line is where buy at once is better and by how much. To make comparisons over time possible the difference in returns has been annualised. For example a value of 20 means dollar cost averaging was 20% per annum better at that specific date. A value of -20 means buy at once was 20% per annum better at that date. The size of the differences is significant. When dollar cost averaging is better the future returns are on average 18% per annum better. When buy at once is better the future returns are on average 32% per annum better. Those are some pretty large differences.
Let’s look now at the specific periods where dollar cost averaging is better. These are the green bits, so let’s zoom in there. There are large green patches in 2011, 2014, and 2018 which as you might guess correspond to periods of “bubbles” and their subsequent bear markets.
The pattern in each case is quite clear. Dollar cost averaging becomes the better strategy when we are already a fair way up a large near vertical rise (which is later called a bubble), and remains better until we are getting close to (but not yet at) a subsequent price bottom. All other times it’s better to buy all at once.
The 2018 chart has no red or green for the most recent 12 months since we don’t yet have the future price data to know how dollar cost averaging over the subsequent 12 months will do. This period is shown as grey on the chart. In order to try guess how things will turn out I changed the rules a bit and said let’s still see how dollar cost averaging would have done over whatever time period we have left. So instead of 12 months let’s use 11 months if there is only 11 months of data left, 10 if there are only 10, and so on down to just 2 months if that is all we have. The “2018 with assumptions” chart shows what happens when you do this.
Give all of the above what would I do right now if I wanted to invest in bitcoin. We are clearly not part way up a vertical rise so that can be ruled out. Are we close to a bottom or have we already bottomed. Maybe. The “2018 with assumptions” chart seems to indicate that we have just transitioned to a red period where buy at once is better. Remember that this chart has extra assumptions and the colors at the end part are still tentative.
From a fundamental point of view my personal take is that it’s too early to call the end of the current bear market. Price may well have bottomed at $3,100 ish in mid Dec, but it’s also possible the final bottom is not yet in. There are prominent bitcoin bulls (yes bulls) making a credible case that we can still fall to $1,000 before the next bull market begins. The purging of excesses from the previous bull market and the ICO unwind might still have further to go.
Since dollar cost averaging has a psychological advantage over buy at once and it’s a close call as to whether we are now in a green or red period, I think on balance dollar cost averaging is still the choice for me as of today (5th April 2019). The recent unexpected rise on 2nd and 3rd April also calls for caution and adds to the case for using dollar cost averaging in case that unwinds.
In closing here is a summary of what I learned:
- Most (but not all) of the time you get better future returns if you buy bitcoin all at once rather than dollar cost average.
- However dollar cost averaging does work better if you are buying during a bear market or if the price is already some way up a large vertical rise.
- Dollar cost averaging is psychologically easier so if in doubt then rather dollar cost average and be happy.
The Python code for this article is available on Github
Please note that nothing in this article should be taken as investment advice. These are my own personal views. If you got this far then you probably already realised that Bitcoin is extremely volatile. Before investing in Bitcoin or anything else please do your own research and make your own decisions.