Posted on 2nd Aug 2020 by the LSS team
Whenever Bitcoin does relatively well for a few weekends in a row, the discussion starts again about whether there is some sort of "weekend effect" present in Bitcoin returns that we should all be aware of. I should know, as I am writing this to you on a Saturday where Bitcoin is going up 5%. We did a little bit of digging of our own based on the latest data available to better understand to what extent this is or isn't true.
To share some examples with you of the research I mentioned, here is Investopedia claiming Bitcoin sees its biggest swings on the weekend. On the contrary, here is a Medium piece on the same topic concluding that there really is no meaningful difference between the different days of the week. But perhaps most authorative is this piece of actual published scientific research which for instance notes that Bitcoin sees outsized returns on Mondays.
To start our own study, I first of all note that looking through Bitcoin returns since September 2014 (thereby deliberately missing out the super volatile early days as cryptocurrencies tend to behave quite differently in the very early days) and dividing those returns up by day of the week, the following is the pattern we see. The best days of the week are by far and away Monday and Friday.
These are days in which volatility tends not to be excessive, while average returns are the highest. In particular on Monday, the 95% percentile (top 5% of returns) is at +7.2% compared to -4.4% on the 5% percentile (bottom 5% of returns), indicating a healthy skew. While Saturday can hold its own, Sunday is not much of a day for seeing outsized BTC returns. So in essence, if Bitcoin has been experiencing positive momentum, Friday-Saturday is indeed not a bad day to dip your toes in the market, but Sunday tends to be a different story. Indeed, Sunday's returns seem low and skewed to the downside.
|Day of week||Average return||Standard deviation||5% percentile||95% percentile|
Note again that our results are consistent with the scientific paper we refered to earlier, as they too saw outsized returns on Mondays.
Now if we look a little deeper and take a closer look at some overall statistics here is the image that emerges. In the median, that is the most typical result seen, the weekend and weekdays are not materially different safe for 1 basis point. But it's in the average where the differences become apparent as the weekday number is significantly higher at +0.18% due to the higher extremes that are seen on the positive side. The 95% percentile of returns is +6.44% which is higher than the +4.97% during weekends. Similarly the figures are greater on the downside, implying that there is both less pump and less dump during weekends.
I really enjoy research like this, but I find myself wanting to approach it from a more practical perspective. What would really have happened if this research had been published in December 2019 and in the January-July period I had followed some of the findings, being in the market only on Friday, Saturday and Monday? The chart below shows this.
Investing in and around the weekend (red line) would have prevented much of the downside and in fact would never have seen you underwater year-to-date (even when Bitcoin was down 50%!). On the contrary, only investing on Tuesday, Wednesday, Thursday and Sunday would have you doing far worse than both buy and hold and the days around the weekend. In fact you would still be down nearly 16% year-to-date! Clear winner though? Buy and hold. But with significantly more risk and higher drawdowns. We may want to avoid the middle of the week for trading for a bit.
And finally, the actual return distribution of Bitcoin during weekends and weekdays looks like this: