Over the past year I’ve written (OK, complained) about how the 24/7 trading in cryptocurrencies can upend the weekends for those of us who prefer to end the trading week on Friday afternoon. I’m sure that many of you were equally thrilled to see news alerts on Saturday about a 17% flash crash in bitcoin and other cryptos. I say that sarcastically – I don’t wish any harm upon crypto investors – but because I prefer to clear my head from Friday afternoon until Sunday evening when Australasian markets reopen.No such luck last weekend.

I wish I could say that this was an isolated event, unlikely to be repeated. But it’s not. It’s happened before and it is likely to happen again. Those of you who trade or invest in cryptocurrencies need to be cognizant of the circumstances that allow this to occur.

First, let’s take a look at the action in Bitcoin and Ethereum over the past 7 days:

7 Day Intraday Bitcoin (white) and Ethereum (blue) vs. US Dollar

(Click on image to enlarge)

Source: Bloomberg

We see the drop occur just after midnight (EST) on Saturday morning. Even though it would have been prime time in much of Asia, it is reasonable to expect that liquidity late on a Friday night/early Saturday morning would be impaired relative to normal business hours.^[i]Now consider that there are crypto platforms that allow their customers to lever their positions 20x or more. If a trader is leveraged 20:1 (by putting up only 5% margin), it clearly doesn’t take much of a move to force either a margin call or an automated closeout. Bear in mind that prior to this weekend, Bloomberg data showed that the 20-day historical volatility of bitcoin was about 53. Consider also that that historical volatility is calculated on a “close to close” basis. While “close” is a meaningless term in a 24/7 market, it works as a concept if it is calculated from the same time each day. It is highly reasonable to expect that the daily range is something greater than shown by two data points at an arbitrary time.

Regardless, using the rule of 16, the historical volatility of 53 means that bitcoin was averaging 3.3% daily moves. It doesn’t require much imagination to think that it is likely that a 5% margin does not offer much protection under those circumstances. It is not clear what triggered the weekend’s move, but it would seem to not require too much effort to push bitcoin low enough that margin calls and other forced sales were triggered. I don’t know why the flash crash was triggered, but I can certainly piece together how it occurred.

There are a few takeaways from the weekend activity. First, the conditions that allowed the flash crash to occur certainly do not appear to be all that unique to last weekend. We’ve seen weekend flash crashes in bitcoin before and we’re likely to see them again. Second, we were already seeing weakness in bitcoin prior to the weekend. We noted last Tuesday that bitcoin was acting more like a risk asset in concert with stocks than a hedge against market volatility. On Friday, bitcoin had been outperforming stocks, only to more than catch up over the weekend. Third, while Ethereum also suffered a drop over the weekend, it did not fall nearly as much in percentage terms and is outperforming subsequently. We won’t delve into the pros and cons of one cryptocurrency versus another here, but the relative stability of ether relative to bitcoin could be a real positive.

Last week, we concluded thus:

Perhaps bitcoin is a victim of its own success. It is quite possible that many investors now view bitcoin as a core holding, much as they do stocks or bonds. 

I stand by that statement as an explanation for why bitcoin is now acting more like another risk asset than a poorly correlated hedge. But if that is the case, investors should consider just how much a core holding of cryptocurrency – especially bitcoin – adds volatility to one’s portfolio and whether that volatility is desirable in the current market environment.

[i] I don’t know when the various crypto platforms do maintenance, but it is reasonable to expect that would be done at times when trading is less active. That could have impaired liquidity, but I’m not comfortable asserting that without more data