In response to yesterday’s Congressional Hearing on Bitcoin in the House’s Small Business Committee, I wanted to do an analysis on the volatility of Bitcoin. But first, it is important to note the well received comments in the hearing such as Jerry Brito’s comment, “Like the Internet itself, Bitcoin has the potential to be a platform for the kind of permissionless innovation that has driven so much of the growth of our economy.” On behalf of the Bitcoin community, coinbase‘s Director of Business Development and Sales, Adam White, also pointed out the potential for business of all sizes to use bitcoin for “the elimination of fraud, reduction of transaction fees, and monetization of new markets”.
Yet despite much of the hype and excitement there still exist some risks, points out Mark Williams, a risk management expert from Boston University. He refers to the “extreme price risk” of Bitcoin, specifically with its volatility. So I decided to do some of my own research and see how this “extreme” volatility really affects merchants using Bitcoin as a payment processor.
I ran an analysis using bitcoincharts historical prices in 5, 15, and 60 minute intervals to calculate volatility. It is within these windows that volatility is relevant to a merchant accepting a payment in Bitcoin (i.e. using the coinbase API). The difference between the 5 and 15 minute volatility is small, both of which are about 0.5% volatility. This means that a $100 Bitcoin payment has a 94% probability of being between $99 and $101 within the first 15 minutes (2 standard deviations from the mean). Considering that there is an equal likelihood that the price rises slightly as opposed to falling slightly means that over time, this should all balance out! So there seems to be insignificant risk for accepting Bitcoin after all.
Another point worth noting is that over time, the volatility of Bitcoin has reduced dramatically. In March 2013, the 1-hour volatility was 2.37% as opposed to the 0.98% in March 2014. This phenomenon is largely due to the fact that Bitcoin is more densely traded. As its acceptance continues to rise, volatility will continue to fall. So for now, we will watch volatility closely but not lose any sleep over it.
Thanks for taking the time to explore this data and sharing your results!
Your last comparison seems skewed though. March 2013 was one of the months with highest volatility; please consider comparing Feb 2013 with Feb 2014, which may not show such a huge decrease in volatility…
Johann, you make a great point. March 2013 was the month that Bitcoin price was extremely volatile. I went back and compared the entire 1Q 2014 to 1Q 2013. I found a similar trend for 1-hour volatility, but not quite as extreme. The volatility in 1Q2013 was 1.64% as opposed to 1.24% over the last 3 months. so there was about a 25% drop in volatility, but volatility over 1% in an hour is pretty high.
The really interesting chart would be a moving average of volatility. Eg., don’t compare March 2014 with March 2013, but plot 1-hour volatility over the course of the past year or two with a moving window size (eg. starting at March 1 2012 to April 1 2012, next point is March 2 2012 to April 2 2012, etc.)
That’s an interesting idea – Another point someone on Reddit made was that it would be interesting to compare volatility during US business hours (i.e. when many companies are transacting with Bitcoin) versus in the middle of the night. I’ll add it to my to-do list, thanks!
Pingback: An analysis of 5, 15, and 60 minute volatility in Bitcoin | Bitcoin News Bits - CoinBits.com
Thank you for the analysis,
Why are different time frames relevant for the Coinbase API? Do you need to wait this times, relative to the tx amount, before Coinbase accepts the transaction?
When you use the coinbase API, there are a few ways for a transaction to process. If your customer pays using their coinbase account, the bitcoins are immediately available to sell because coinbase does not worry about waiting for 6 confirmations. However, if you send the BTC payment from an external wallet, you cannot sell your bitcoins until you receive 6 confirms.
Your comment about being two standard deviations from the mean 94% of the time is only true for the normal distribution. Financial returns typically follow a Student t distribution.