Since its creation in January 2009 up until the present day (28 June 2021, as I write this), the price of Bitcoin has increased dramatically. During this time, it has gone through several bull and bear markets featuring spectacular increases in price as well as many significant drops. In this post, I take a long term look at how the price has behaved over the years. This can serve as an idea as to how it may behave in the future although, it hardly needs to be pointed out, prices are random and we can never guarantee future behaviour.
In the first year or two of its creation, there was no reliable bitcoin price data. It was traded very infrequently and, when it was, this was often done via private transactions and not on public exchanges. On 22 May 2010, known as Bitcoin pizza day, Laszlo Hanyecz paid 10,000 bitcoin for two Papa John’s pizzas which, effectively, put the price of one bitcoin at 0.4 cents. Some studies make use of such events in order to give a bitcoin price for the early days, but such numbers are clearly very unreliable and of little use in understanding its later evolution.
The MtGox exchange launched in July 2010 and ran up until its spectacular collapse in early 2014. Other exchanges opened in the years following MtGox, including Bitstamp in 2011 and Bitfinex in 2012. We do, therefore, have exchange price data going back as far as July 2010. This is available in the Tradingview website as the ‘bitcoin all time history index’, which combines data from various exchanges to obtain continuous price data from 12 July 2010 onwards. Figure 1 shows the prices over this time period using a logarithmic scale.
Between 12 July 2010 and the current all time high on 14 April 2021, prices increased from 5 cents to $64,900, a factor of over 1 million. The history can be roughly split into 4 bull markets and 3 intervening bear markets. This is partly subjective, since there is no universally accepted definition of what constitutes a bull or a bear market. Splitting up these periods according to the local highs and lows:
- 12 July 2010 (when the data starts) to 8 June 2011, with the price increasing from 5 cents to $31.91, an increase of about 640x.
- 8 June 2011 to 17 November 2011, where the price decreased from $31.92 to $1.99, a factor of about 16x.
- 8 June 2011 to 29 November 2013, where the price increased from $1.99 to $1,242, a factor of about 600x. This includes a mini bear market where the price collapsed from $266 on 10 April 2013 to $71.51 five days later, a decrease of about 73%, and took over 5 months to recover.
- 29 November 2013 to 18 August 2015, when the price went from $1,242 to $162, a decrease of 87%. This was a single intra-day low, and the low point could instead be considered to be $166 on 14 January 2015.
- 18 August 2015 to 17 December 2017, when the price went from $162 to $19,804, an increase of about 122x.
- 17 December 2017 to 15 December 2018, when the price went from $19,804 to $3,124, a decrease of about 84%.
- 15 December 2018 to 14 Apr 2021, where the price went from $3,124 to $64,898, which is an increase of about 21x. This included a mini bull market where the price went up to $13,870 on 26 June 2019, a mini bear market where it dipped to $6,340 on 18 December 2019, and the coronavirus panic when it dropped to $3,880 on 13 March 2020 from highs just over $10k a month prior.
As things stand on 28 June 2021, prices have dropped about 50% from the recent high of just under $65k. Does this mean that the bull market is over, or can we expect prices to recover within a few weeks or months as they did in the short-term bear market of 2013? Opinion is divided on this matter, and the current post is concerned with long term trends which will not answer such short term questions. If history is any guide, we can expect that it will recover eventually, but could take anything from a couple of weeks to two or three years if a longer term bear market sets in.
Looking at the historical prices, they are characterised by bull markets where prices increase slowly at first and finally show explosive growth before a collapse. Then, the bear market follows where the prices are still volatile but decline over a period, then an accumulation phase followed by the next bull market. Each bull market has always led to prices remaining higher than all previous times, even at the very bottom of the ensuing bear market. The longest period for which you could have held bitcoin and lost in dollar terms is 1,243 days from 29 November 2013 to 24 April 2017.
Dips within bull markets
Zooming out this far and using the logarithmic scale in figure 1 does slightly obscure some important properties. There are many drops of 30% to over 40% even during the bull markets. These can take a few days up to a few months to recover, and sometimes drop again, before continuing the upwards trajectory. Without the benefit of hindsight, this makes it difficult to distinguish between these regular price drops during a bull market, and the top before the oncoming bear market.
Figure 2 shows eight separate occasions during the 2016-2017 bull market where the price dropped by over 30% and, for three of those, over 40%. The time taken to recover back to the previous all time high took anywhere from 8 days up to 206 days. Bearing this is in mind, when the price peaked in December 2017 and dropped significantly, it is not surprising that many people believed that this was just another short-lived dip. In fact, it was not until 30 November 2020 that prices recovered back to the all time high of just under $20k.
Fitting historical bitcoin prices
Looking again at figure 1, there are a few points which should be noted. The prices are generally increasing over time interspersed with several ‘bubbles’ where the prices increase rapidly and decrease back to an increasing baseline over a period of a year or two. Looking at this baseline curve, it appears to be reasonably smooth, increasing, and concave on the logarithmic scale. It is not difficult to draw in a smooth curve by hand which passes through these low values. However, there is also an indicator available in Tradingview to draw this (Bitcoin Logarithmic Growth Curves, by quantadelic), shown in figure 3. The lower curve in particular can be seen to very well match many data points where Bitcoin is in an accumulation phase after a bear market and before the next explosive bull market. The only significant drop below this was during the coronavirus panic in March 2020, which recovered quickly back to the line. The upper curve also very well matches the peaks of the 3 previous bull markets that we have seen so far although, being only 3 data points, this is less useful. Intermediate curves are also plotted, although these are not important for the current discussion. They are just interpolated between the upper and lower curves, including the mid-line and several Fibonacci levels that traders like to look at.
If we were to assume that the prices again reach the top curve in mid to late 2021, this suggests a peak price of $100k to $200k and drop back to about $50k by the end of 2022. If, on the other hand, a longer term bear market ensues and prices drop back to the lower regression line, then we still expect prices to recover back to around $65k by early 2023. Of course, these are just suggestions for possible scenarios based on prices continuing to show similar behaviour as historically, and not firm predictions.
Although I do not attach any specific importance to the exact formula for the curves, the following is being used, with t the time in days since 21 July 2010.
|t(tlog10(t+10) – 1)eIntercept + Slope x t.|
The lower curve is using
|Intercept = -3.0269716,|
|Slope = 0.001329,|
and the upper one is using
|Intercept = 1.06930947,|
|Slope = 0.00076,|
I have seen other parameterizations used which suggest a lower market top, although this one appears to be a better fit to both the top and bottom of the market cycles than the others.
From figure 3, we see that the prices often follow close to the lower line (accumulation phase) with periodic bull market rallies up to the top line followed by bear market declines back to the lower line. As the curves are increasing, the rallies are significantly larger than the subsequent decline. Also, as the curves get closer over time, the volatility and magnitude of the rallies and drops are declining over time (in the log scale). We should put more confidence in the lower line than the upper one, since there are many more data points and because the market tops are very unpredictable by their nature. Also, as more market participants are aware of these historical patterns, it will affect their behaviour and feed back to impact future price movements. For example, if prices were to reach close to the top curve, we can expect selling by people who are aware of what happened on previous occasions, likely causing a dip in the price.