Last Updated on April 19, 2022 by Quantified Trading
Does market timing work on Bitcoin? Let’s play with some numbers and find out how removing certain observations influences the end result compared to buy and hold. We test by owning Bitcoin from January 2015 until the end of September 2021 six and a half years of price data.
Market timing is unlikely to work in a volatile asset like Bitcoin that has doubled many times in a short period of time assuming you are mainly a buy and hold investor. The main reason for that is the enormous rise in the price of Bitcoin. Only time will tell if this repeats in the future.
In this article, we remove the ten best days of the data set, the ten worst days, and the best day of every month. The different results are astonishing. Anyway, as explained in the article, any reference to buy and hold might not be the best benchmark.
First, let’s briefly decide what we mean by market timing:
What is market timing?
Market timing is the exercise of when you try to pick tops and bottoms based on what you believe of the price in the future. You can use a wide range of tools for market timing. This site uses quantified analysis.
Market timing strategies
On our landing page called free trading strategies, we have shown at least 70 practical examples of how you can utilize such strategies for market timing some of them beating any buy and hold handy. We serve better strategies for our paid subscription service, though:
Only you can find out if these strategies have any value. At the end of the day, any comparison to any buy and hold strategy might not be relevant:
Is buy and hold vs. market timing relevant?
It depends. If you’re a long-term buy and hold investor that aims to capture the market’s historical risk premium of about 10% a year, it’s highly likely a futile exercise to try to time the market.
The best option is to simply invest whenever you have capital available and forget about it. Most likely the time spent in the market will make sure you end up with decent annual returns. If you are a trader, we recommend setting aside a certain amount of your capital for buy and hold.
However, for traders, any buy and hold returns might not be a proper benchmark. This is due to leverage, less drawdowns, Sharpe Ratio, low correlation to buy and hold, etc. A trader aims to make money in every trading environment to get a steadily rising equity curve.
Thus, buy and hold might be a completely irrelevant benchmark.
Bitcoin and market timing the results are in:
Let’s return to Bitcoin and market timing. To show how difficult market timing is compared to buy and hold in an asset that rises 126% annually, we backtested three different scenarios:
- Removing the ten best days of the dataset
- Removing the ten worst days of the dataset
- Removing the best day of every month
Let’s see how the results changed:
Buy and hold Bitcoin January 2015 to October 2021 (HODL)
If you bought 10 000 USD worth of Bitcoin on the 1st of January 2015 you would have 2.05 million worth of Bitcoin by the end of September 2021. The chart below is the logarithmic chart of the equity curve (read here for logarithmic vs. linear charts):
Bitcoin’s performance when the ten best days in the dataset are removed
Let’s remove the best ten trading days of the whole sample/dataset. Our 10 000 would only be worth 246 000 USD by the end of September 2021. Quite a drop from 2.05 million! The CAGR is 63% (still high, though).
Why does this happen? It’s the compounding effect. A huge part of the compounding is removed from the dataset, and it’s an enormous drag in performance. Just missing the two best days makes you underperform a lot.
Bitcoin’s performance when the ten worst days in the dataset are removed
When we remove the worst ten days of the whole sample/dataset, the final equity is 11.6 million USD. The CAGR is 196%.
When the worst days are taken out of the sample, the compounding starts at a higher plateau. This has enormous positive effects long-term. Think of it like this: 10% of 100 is much better than 10% of 50 (in absolute terms).
Bitcoin’s performance when the best day of every month is removed
Let’s test something more drastic: we remove the best day of every month: 81 days are taken out from the dataset. This is 3.3% of all observations.
Something interesting happens (linear chart):
The initial 10 000 is now only worth 1 854 USD. You are practically penniless. This is a negative 22% CAGR.
Conclusion: does market timing work on Bitcoin?
If you buy and hold (HODL) Bitcoin from January 2015 until September 2021, you end up with 2.05 million.
However, if we make some editions to the dataset the results are completely turned upside down:
- If the ten best days of the whole sample are removed, you end up with 246 000
- If the ten worst days of the whole sample are removed, you end up with 11.6 million
- If the best day of the month is removed every month, you end up with 1 850.
Thus, in a super volatile asset like Bitcoin that has doubled many times in a short period of time, missing just a few days can change the end result drastically. If you are mainly a buy and hold investor, it most likely doesn’t make much sense to play the market. Market timing in Bitcoin seems like a futile exercise compared to HODL.