Specializes in pattern recognition and trading systems development. He founded Seasonax and publishes the website www.SeasonalCharts.com , which features free-of-charge seasonal charts for interested investors. In his book The Gold Cartel (published by Palgrave Macmillan, see link on the right hand side), Dimitri discusses gold price manipulation and modern-day credit excess.
In the last issue of Seasonal Insights I have discussed how the S&P 500 Index performs on individual days of the week. In this issue I will show an analysis of the average cumulative annual returns of bitcoin on individual days of the week.
Bitcoin, daily. While this is beside the point, we note the crypto-currency (and other “alt coins” as well) has minor performance issues lately. The white line indicates important lateral support, but this looks to us like it could be the beginning of a higher degree correction, mainly because it so far proceeds with greater verve than previous corrections over the past year. Besides, the recent rally in this trading sardine seems rather stretched, and the term stretched by itself actually sounds a bit inadequate as a descriptor. It needs an adverb… insanely might do. [PT] – click to enlarge.
It seems to me that bitcoin is particularly interesting for this type of study: it exhibits spectacular price gains, its is a very new instrument and it is unregulated. Moreover, it trades around the clock, even on weekends.
Given this setup, I found myself wondering: do prices rise uniformly? And does their performance on weekends differ from that on workdays?
The first chart below shows the annualized performance of bitcoin since 2011 in black, as well as the cumulative annualized performance on individual days of the week in blue.
Note: due to the extent of the rally to date, the compounding effect is particularly strong; the average cumulative returns achieved on individual days of the week therefore don’t add up linearly to the total annualized return of bitcoin. I have inserted a break in the black column and the scale of the y-axis in order to improve the clarity of the chart.
Bitcoin, performance by individual days of the week, 2011 to 2017. On Fridays and Saturdays the average gain in bitcoin was negligible.
As the chart shows, the beginning of the week is standing out quite noticeably: if one had invested exclusively on Mondays, an annual return of 39.30 percent would have been attained. The cumulative gain achieved on Tuesdays was even stronger at 59.68 percent.
By contrast, the average return of 0.56% achieved on Friday was essentially irrelevant. In short, even in bitcoin there were days of the week which on average generated almost no gain. Moreover, Saturdays, and by the standards of bitcoin also Sundays and Wednesdays, were rather boring days as well.
The difference in the average annual returns achieved on individual days of the week – measured over no less than 2,333 trading days – is quite significant. This suggests that it is highly unlikely to be a random pattern.
Next we will take a look at the evolution of these trends over time in detail.
The next chart shows the performance of bitcoin since 2011 in black, as well as the cumulative return achieved on every individual day of the week in other colors – indexed to 100.
As the chart shows, Tuesday (yellow line) was well ahead of the pack from the very beginning. By contrast, Monday initially performed poorly, but later caught up rapidly, especially after 2012. Fridays and Saturdays were boring days for bitcoin traders throughout the entire period.
The results of this study underscore that even a very young market that has experienced extraordinary price gains (for which normally a great many trading days should be required), exhibits very conspicuous statistical anomalies.
How can these anomalies be explained? My theory is that many people interested in bitcoin arrive at the decision to make a purchase over weekends. These decisions are then implemented at the beginning of the week. The first purchases are made on Mondays. On Tuesdays those who were waiting for confirmation from rising prices follow suit and buy as well. Thereafter the action tails off until the beginning of the next week.
What do the statistical anomalies in bitcoin tell us generally about investing in financial markets?
Contrary to what some theories assert, efficient markets do not exist in the real world. Human habits, regulations or manipulation can result in recurring disparities in price patterns. Statistical oddities such as those in bitcoin discussed above and many other such patterns can be quickly found with the help of the Seasonax app on Bloomberg or Thomson-Reuters systems; they exist in stock markets, currencies and commodities as well. Don’t let the apostles of the efficient market hypothesis keep you from taking advantage of them.
Charts by: Cryptowatch, Seasonax
Image captions by PT where indicated
Dimitri Speck specializes in pattern recognition and trading systems development. He is the founder of Seasonax, the company which created the Seasonax app for the Bloomberg and Thomson-Reuters systems. He also publishes the website www.SeasonalCharts.com, which features selected seasonal charts for interested investors free of charge. In his book The Gold Cartel (published by Palgrave Macmillan), Dimitri provides a unique perspective on the history of gold price manipulation, government intervention in markets and the vast credit excesses of recent decades. His ground-breaking work on intraday patterns in gold prices was inter alia used by financial supervisors to gather evidence on the manipulation of the now defunct gold and silver fix method in London. His Stay-C commodities trading strategy won several awards in Europe; it was the best-performing quantitative commodities fund ever listed on a German exchange. For detailed information on the Seasonax app click here (n.b., subscriptions through Acting Man qualify for a special discount).