Jonathan Heusser writes at The Genesis Block that Bitcoin markets now offer enough data that we can make some interesting predictions of the currency’s price movement.

This analysis hinges on the idea that events within a set of data actually cause similar events to reoccur later. The term he uses is “self-exciting” to describe financial markets, and he says that’s something we can now find in Bitcoin markets.

The thing is trades usually occur in clusters rather than a set of evenly spaced trading events. That is, something motivates trading activity, and that activity clumps around certain points in time.

Heusser took this idea and applied it to a set of trading data from September 4 and September 5, a period in which Bitcoin prices fell more than 10% in just less than 20 hours. Such wild periods reveal excitability.

Excitability looks like this when plotting it on a graph: Clusters of trades build up, each peak becoming more intense than the last, until something tips a spike in intensity. Shortly thereafter, trades return to a mean level of intensity.

Statistical analysis of historical data reveals two things: We can anticipate intensity, and we can estimate what percentage of future trades will be the direct result of current trades.

That second insight, called branch ratio, allows analysts to spot potential signs that a market is bottoming out. Simply put, when earlier trades stop driving current or future trades, the price tends to sink.

Calculating a branching ratio and applying it in real time, however, is tricky and requires a great deal of insight. It’s one thing to draw conclusions from a historical data set, after all, and quite another to predict behavior at the speed of market activity.

Nevertheless, that’s exactly what kind of analyses are being done in traditional financial markets, and Bitcoin appears to have matured enough that such calculations are applicable to that market, as well.