Volatility: Climbing Only to Descend
By looking at the 30-day historical volatilities of BTC, ETH, BCH, LTC, BSV, and ETC, we see that general crypto volatility decreased at the beginning of 2019, ramped up in the middle of the year, and decreased near the year’s end. BTC’s volatility began 2019 at 70.21%, fell to 17.86% by April 1st, climbed to 102.53% by July 20th, and fell to 32.05% by the end of December.
By looking at the 30-day historical volatilities of ETH, BCH, LTC, BSV, and ETC as a percentage of BTC’s 30-day historical volatility, we can see that altcoins experienced several sharp spikes in volatility independent of BTC’s movements throughout the year. At a glance, BSV and ETC appear to be the standouts in terms of large, BTC-independent volatility fluctuations.
Price Correlations: Unexpectedly Low Crypto Correlations
The most recent crypto correlations data show that BTC has a much lower positive correlation with BCH, BSV, and ETC than typical. Whereas these correlations have been as high as 0.8 or 0.9 in the past, they are all currently between 0.4 and 0.6. These correlations began to dip around the 20th of December. Whether these altcoins will continue to be less correlated with BTC into 2020 remains to be seen. All cryptoassets remain largely uncorrelated with gold and the S&P 500.
See the full SFOX crypto correlations matrix below.
For a more complete look at BTC’s correlations with other assets throughout the past year, see the following graph.
Appendix: Data Sources, Definitions, and Methodology
All cryptocurrency prices are denominated in USD unless otherwise noted.
Note that data collection for ETC began on January 16th, 2019, and data collection for BSV began on March 11th, 2019; therefore, this report does not reflect a full year of data for these two cryptoassets.
We use two different in-house volatility indices in creating these reports.
1. 30-day historical volatility (HV) indices are calculated from daily snapshots over the relevant 30-day period using the formula:
30-Day HV Index = ?(Ln(P1/P0), Ln(P2/P1), …, Ln(P30/P29)) * v(365)
2. Daily historical volatility (HV) indices are calculated from 1440 snapshots over the relevant 24-hour period using the formula:
Daily HV Index = ?(Ln(P1/P0), Ln(P2/P1), …, Ln(P1440/P1439))* v(1440)
S&P 500 performance data are collected from Yahoo! Finance using GSPC (S&P 500 Index) data. Gold performance data are collected from Yahoo! Finance using XAU (Philadelphia Gold and Silver Index) data.
30-day asset correlations are calculated using the Pearson method, in accordance with the following formula.
In our calculations, x = 30-day returns for BTC/USD, y = 30-day returns for the other asset in consideration, and r = the correlation coefficient between BTC and the other asset in consideration.
The cryptoasset data sources aggregated for crypto prices, correlations, and volatility indices presented and analyzed in this report are from the following eight exchanges, the order-book data of which we collect and store in real time.
- bitFlyer
- Binance
- Bitstamp
- Bittrex
- Coinbase
- Gemini
- itBit
- Kraken
Our indices’ integration of data from multiple top liquidity providers offers a more holistic view of the crypto market’s minute-to-minute movement. There are two problems with looking to any single liquidity provider for marketwide data.
- Different liquidity providers experience widely varying trade volumes. For example: according to CoinMarketCap, BINANCE saw over $20 billion USD in trading volume in November 2018, whereas Bitstamp saw $2 billion USD in trading volume in that same time – an order-of-magnitude difference. Therefore, treating any single liquidity provider’s data as representative of the overall market is myopic.
- Liquidity providers routinely experience interruptions in data collection. For instance, virtually every exchange undergoes regularly scheduled maintenance at one point or another, at which point their order books are unavailable and they therefore have no market data to collect or report. At best, this can prevent analysts from getting a full picture of market performance; at worst, it can make it virtually impossible to build metrics such as historical volatility indices.
Building volatility indices that collect real-time data from many distinct liquidity providers mitigates both of these problems: collecting and averaging data from different sources prevents any single source from having an out-sized impact on our view of the market, and it also allows us to still have data for analysis even if one or two of those sources experience interruptions. We use five redundant data collection mechanisms for each exchange in order to ensure that our data collection will remain uninterrupted even in the event of multiple failures.
