Seasonality in Bitcoin: Examining Almost A Decade of Price Data

Feb 5, 2020 · 12 min read

With almost ten years of price data for bitcoin now available, this article will investigate if there are any seasonal effects for BTC-USD.

In traditional markets, seasonal trends are often found in financial time series data such as Gross Domestic Product (e.g., recessions and expansions), the price of a commodity because of changes in the weather (e.g., summer versus winter) or calendar events for a company’s stock price (e.g., earnings season).

Image for post
Image for post
BTC-USD price history (log transformed). Source: CoinMetrics.

Bitcoin’s price history — illustrated above — is limited. But since there is now almost ten years of data, we can begin to uncover any seasonal trends.

There are several hypotheses about seasonal trends in BTC-USD — such as the influence of tax season, increased trading/blockchain activity during winter months (shadowing internet usage), and the effect of public holidays such as the Chinese New Year and Christmas (i.e., the “Santa Claus rally”).

The calendar heatmap below displays the rolling 30-day return on investment for bitcoin (BTC) since mid-August 2010. Positive returns were transformed to 1 and we code negative returns as 0 to provide more clarity:

  • A green square on the calendar means that on that day, the 30-day return on investment was positive.
  • A red square on the calendar means that on that day, the 30-day return on investment was negative.
Image for post
Image for post
Image for post
Image for post
Source: CoinMetrics, bitcoin 30-day ROI data for 18/08/2010 to 31/01/2020.

For the 113 months we have full price data for, there have been 28 months where the 30-day return for bitcoin was positive every single day. For example, the 30-day return for BTC was positive every single day in April, May and June during 2019. There’s also a strong momentum effect. Ten of these months that were fully green were followed by another straight month of positive returns.

There’s less tendency for the 30-day return to remain in the red for an entire calendar month, with ten examples so far. Again, there’s a similar momentum effect, with three of these months followed by another month of straight red. The last time this happened was September-October 2014.

Another glance at the calendar heatmap also shows that in July and November, if bitcoin’s 30-day return is positive or negative during the first day of these months, it usually stays that way. There are six instances in both July and November where the 30-day return was negative (or positive) for the entire month.

Looking at public holidays such as Chinese New Year and Christmas, we see that the 30-day return around the start of Chinese New Year has been negative in recent years (2016–2019) while the 30-day return was positive in 2020 and between 2011–2015. There’s not much evidence for a “Santa Claus rally” with bitcoin, since just five of the past ten years display a positive ROI in the run up the Christmas.

The 30-day returns are usually negative around the time of tax deadlines in many countries such as the US and the UK (late March-early April). During this period, investors may engage in tax loss harvesting — selling digital currency to incur losses and reduce their tax burden. Another explanation could be the sale of digital assets by investors to pay their capital gains tax or other tax liabilities. Out of nine years, the 30-day return was negative seven times during this period.


Are there any seasonal trends in the price volatility of bitcoin?

Bitcoin’s price history is cut into two parts to account for the falling volatility over time. The calendar heatmap below shows the raw volatility data for the years 2010-2014 and 2015-2019 separately.

Green/yellow days are where the price action was highly volatile while orange/red show moderate-low volatility days.

BTC-USD was more volatile during the spring/summer months during 2010–2014, with periods of high volatility during June 2011 and April-May 2013. We also saw a period of elevated volatility during November 2010 and December 2013-January 2014.

Image for post
Image for post
Image for post
Image for post
Source: CoinMetrics, volatility of 30-day returns for bitcoin between 18/08/2010 to 31/12/2019.

For the years 2015–2019, the first few months of the year have been volatile as compared to the rest of the year (with 2017 and 2019 being exceptions).

Looking at both calendar heatmaps together, it’s clear that BTC-USD usually experiences heightened volatility in summer (June, July, August) and winter (December, January, February) months.

Let’s look at the seasonal trends in bitcoin differently by plotting the distribution of monthly returns.

The Distribution of Monthly Returns

How do bitcoin returns vary by month?

The distribution of monthly returns are plotted in the chart below and excludes outliers.

Image for post
Image for post
Returns by month — outliers excluded. Source: CoinMetrics, daily bitcoin price data for 18/07/2010 to 31/01/2020.

When excluding outliers, the median returns are negative for March, August, and September. Historically, the largest returns are seen in June and April, with median gains near 20 percent. October, November, and December also show a high median return.

The median returns show that February, April-June and October-December have been much more profitable for bitcoin investors. The chart also shows the volatility of monthly returns, with the ranges of the boxplots being much larger during months with high median returns.

The violin chart below shows the true distribution of the data (outliers included). The density of each plot represents the frequency of observations and the mean return is shown as a diamond. Reading a violin chart is the same as reading a density plot: thicker parts of the chart represent more observations in that range.

Image for post
Image for post
Monthly returns — outliers included. Source: CoinMetrics, daily bitcoin price data for 18/07/2010 to 31/01/2020.

In January 2020, Bitcoin’s performance was +30%. The distribution for January above shows that the current performance is higher than average, but is not unprecedented — with bitcoin gaining 53% and 75% in January 2013 and January 2011, respectively.

A few observations from the violin chart above:

  • The average monthly returns are all positive — except for September.
  • The lower bound of April’s plot is just below zero, while most of the distribution is in positive territory. Since 2011, you’d have only lost money twice (in 2014 and 2015) if you bought bitcoin at the start of April and sold at the end of April. Even in 2014 and 2015, the losses were -4.39% and -6.29% respectively.
  • Comparing the violin chart with the boxplot for March, one outlier makes March’s average return positive. When excluding outliers, March has the second lowest median return, suggesting we’re more likely to see negative price growth for bitcoin during this month.
  • April, May, October and November have the largest average monthly returns. These months also show the highest variability in returns.
  • The variation in returns is smallest for July and September.

The Distribution of Returns by Weekday

Is there any link between bitcoin’s price changes and the day of the week?

Wednesday is usually the most volatile and busiest weekday in Forex markets, as the data release cycle is in full swing. However, cryptocurrencies are traded 24/7, 365 days a year, so it would be interesting to see if there are any patterns for returns on the weekend.

The true distributions of returns by the day of the week are shown by the violin plot below. The diamonds represent the mean returns.

Image for post
Image for post
Source: CoinMetrics, daily bitcoin price data for 18/07/2010 to 31/01/2020.

The distribution of daily returns are similar for each weekday, with the peak at (or close to) zero percent. The range of price changes are similar for Monday, Wednesday, and Friday, as shown by the length of each violin plot.

The biggest returns have been on Saturdays, but the biggest drops have occurred on Thursdays. These two weekdays also display the highest variation in returns.

Saturday’s distribution appears to have two peaks, unlike all other days, suggesting that while most of the time returns are clustered around 0 percent, returns also tend to cluster (although to a lesser extent) around 4 percent.

Intraday Returns, Volatility and Trends for BTC-USD

Are there any hourly patterns? Is there an effect on the price of bitcoin corresponding with the opening of the global financial centres, such as London, New York or Tokyo?

Average Hourly Returns for Bitcoin

The heatmap below uses hourly price data from BitMEX for its BTC-USD pair, starting from 00:00 on 01/01/2017 up to 00:00 on 31/01/2020. All times are in UTC.

The colour of the squares represent the intensity of hourly price changes:

  • Dark purple squares show the hours where the trend in the price of bitcoin has usually been negative.
  • Orange/yellow squares show the hourly sessions where the bitcoin price action has usually been bullish.
  • Light purple squares show that the price has remained relatively stable.
Image for post
Image for post
Source: CryptoDatum.

The largest, positive price movements are clustered around 14:00–16:00 on Fridays.

Some of the darkest purple squares are for:

  • the 01:00–02:00 trading session on Tuesday,
  • a couple of trading sessions on Wednesday (14:00–15:00, 18:00–19:00),
  • during 16:00–18:00 on Thursdays/Fridays,
  • and Saturdays between 20:00–21:00.

Average Hourly Volatility for Bitcoin

The heatmap above shows the hourly change in prices on average, but does not account for the hourly highs or lows. The same infographic is shown below for the volatility of BTC-USD (which is calculated as [(high-low)/open].

Image for post
Image for post
Source: CryptoDatum.

The price of bitcoin is more volatile around/during the 16:00–17:00 session on Wednesdays/Thursdays, and during Friday’s trading sessions between 13:00–16:00.

The price of bitcoin tends to be leat volatile Saturday and Sundays mornings (02:00–07:00) and evenings (18:00–20:00) as well as Monday mornings (02:00–06:00) before European markets open.

Average Candle Body Ratio by Hour for Bitcoin

To identify when the price of bitcoin has a higher likelihood of trending, we can also look at the Candle Body Ratio by hour and day of the week.

To understand the Candle Body Ratio, let’s look at two extremes: where the ratio is equal to 1 and close to 0.

Image for post
Image for post
The formula for the Candle Body Ratio.

If the Candle Body Ratio equals 1, then the change between the opening price and closing price is the same as the difference between the high and the low. In candlestick analysis, this is known as a Marubozu.

A Candle Body Ratio close to 0 indicates indecision and these readings are associated with Doji patterns. Two examples are shown in the chart below.

Image for post
Image for post
Illustrating the Candle Body Ratio.

The Candle Body Ratio is used to identify trading sessions that are more likely to display trending candles and the days/hours where BTC-USD is more prone to trendless conditions.

The yellow/orange squares show that the Candle Body Ratio is larger on average during that hourly session, while the dark purple/indigo squares show that the Candle Body Ratio is usually lower on average for that trading session.

Image for post
Image for post
Source: CryptoDatum.

The heatmap of the Candle Body Ratio shows that the price of bitcoin displays stronger trends during certain hours. For instance, yellow squares represent trading sessions where there is a higher likelihood that BTC-USD will display a strong trend.

The trading sessions with the highest average Candle Body Ratio is 08:00–09:00 on Tuesdays and Saturdays — suggesting the hourly candle on average have displayed stronger trending behaviour as the average candle body ratio is higher.

The dark purple squares show us when the candle body size is low compared to the difference between the session’s high and low. For example, on Sunday and Friday between 07:00 and 18:00, we usually see choppy price action in BTC-USD — which is inferred from the lower average Candle Body Ratio in the heatmap above.

The same metrics are shown below for the 4-hour price data from BitMEX.

Average 4-Hour Return for Bitcoin

Image for post
Image for post
Source: CryptoDatum.

Thursdays and Fridays (16:00–20:00) usually see the largest negative price changes on average according to the past three years of data from BitMEX. The average return for the 00:00–04:00 trading session is on the low end of the scale (-0.20%) for Sunday, Tuesday and Friday.

On average, the biggest percentage increases during the 12:00–16:00 session on Friday.

Average Volatility (4-hour) for Bitcoin

Image for post
Image for post
Source: CryptoDatum.

Volatility is usually highest in the 12:00–16:00 session on Fridays and Wednesdays, as well as all sessions in the second half of Thursday. On average, volatility is lowest in the hours of 00:00–08:00 Sunday morning and 04:00–08:00 Saturday morning.

Average Candle Body Ratio (4-hour) For Bitcoin

Image for post
Image for post
Source: CryptoDatum.

Typically, we see stronger trending candles in the early hours of the day between 00:00–04:00, apart from Thursday, Friday and Saturday. Most of the yellow and orange squares between 00:00–04:00, suggesting stronger trends develop in the Asian trading session.

The highest average Candle Body Ratio is found for the early hours of Wednesday morning (00:00–04:00). The dark purple squares show BTC-USD has a tendency for choppy price action on Tuesday (08:00–12:00), as well as later sessions at 16:00–20:00 and 20:00–00:00 on Wednesday, Friday and Saturday.

The heatmap above suggests if you are a trend following trader, you should be active in the hours that display yellow/orange squares and be more cautious entering positions during time periods represented by dark purple squares. The conclusions drawn here assume that historical price action will be similar to the price action in the future.


The months that have been far more profitable for bitcoin traders have been April, May, October and November, with higher median returns and greater volatility. Returns in September have usually been negative. It’s also interesting to note that April’s returns have been positive seven out of nine years, and in the other years the losses in April were not that large (around -5%/-6%).

Daily returns mostly cluster around zero, although there is a small tendency for returns to cluster around small, positive values on Saturdays. Historically, Saturdays have had the highest upside but Thursdays have had the largest drops. These two days of the week also display the largest variation in returns.

Using the 1-hour and 4-hour data for BitMEX’s perpetual swap, we’ve shown during which hours/4-hour periods are associated with positive/negative growth, high volatility and strong-trending candlesticks for BTC-USD. From the data that is available, some of the findings are:

  • The largest 1-hour/4-hour returns on average are during the 12:00–16:00 session on Fridays. Average returns for the 4-hour timeframe are lowest in the early hours of Sunday, Tuesday and Friday between midnight and 4am, as well as 16:00–20:00 on Thursdays and Fridays.
  • Hourly volatility has been greatest during Wednesday and Thursday afternoon (15:00–17:00) according to the data. The largest average 4-hour volatility is found during Friday’s 12:00–16:00 trading session, as well as midday onwards for Wednesday and Thursday. BTC-USD has been least volatile on Saturday and Sunday mornings (04:00–08:00) and evenings (18:00–19:00).
  • The Candle Body Ratio is highest during the 00:00–04:00 trading sessions on Sunday through to Wednesday, suggesting on average, trends are stronger during these periods. Some of the lowest ratios are seen at Tuesday (08:00-12:00), Friday (00:00–04:00) and Wednesday/Saturday evenings — suggesting price action tends to be choppier around these hours.

Keep in mind that these trends may not always repeat themselves. These insights are just a starting point for looking at seasonality in the price of bitcoin. A deeper dive is coming soon.

All graphics are produced with R Studio. The data and R scripts can be found on GitHub so you can create your own heatmaps.

Disclaimer: This blog post is for informational purposes only and should not be taken as financial advice.

Interdax Blog

Insights into Bitcoin, Ethereum, and cryptocurrency…

Medium is an open platform where 170 million readers come to find insightful and dynamic thinking. Here, expert and undiscovered voices alike dive into the heart of any topic and bring new ideas to the surface. Learn more

Follow the writers, publications, and topics that matter to you, and you’ll see them on your homepage and in your inbox. Explore

If you have a story to tell, knowledge to share, or a perspective to offer — welcome home. It’s easy and free to post your thinking on any topic. Write on Medium

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store