Quantifying Blockchain-Based Asset Mobility

Flow analysis is vital to any distributed networking system

Distributed ledgers power the digital asset universe. Reformulating these ledgers as directed graphs (with wallets as nodes, transactions as edges, and transaction amounts as edge weights) enables Boltzmann to efficiently discover and analyze latent structural activity across diverse microcosms of blockchain networks in real-time. One framework we developed to leverage the graph structure is called asset mobility.

Mobility measures an asset’s relative tendency to enter wallets with higher “action potential”, where a wallet’s action potential is defined as its capacity to influence the network (and by proxy, the markets). In a graph formulation, a natural way to quantify the influence of a given node is to consider its out-flow (how much asset has been sent by this node) and its out-degree (how many wallets this node has sent asset to). To quantify wallet influence, we developed a concept called action potential, given by A = F * O where F is a proprietary function of the out-flow and O is a proprietary function of the out-degree. Note that both F and O are dependent on the analysis window; given longer windows, we would expect both F and O to be larger since there are more opportunities to transact. At Boltzmann, we compute mobility in five windows ranging from one week to two months with each window yielding its own nuanced view of the framework.

Define a transaction to be a (u, v, t) tuple in which wallet u sends wallet v an amount of asset t. For a collection of transactions, let the total amount of asset transferred (i.e. the sum of the individual t values) be T. The mobility of this collection is given by taking a weighted average over the action potentials of the receiving nodes v:

Note that since the action potential is computed for the receiving node as opposed to the sending node, mobility signals impending network (and by proxy, market) activity — it is a leading indicator. Because the F and O terms are multiplied, wallets with higher action potentials must push a relatively large amount of asset to a large number of wallets, an attribute we believe is characteristic of exchange and active trader hot wallets. We can observe how mobility evolves over time on the Ethereum network in Figure 1:

Figure 1: Mobility and market price of ETH vs time. Each series is smoothed using a 14-day rolling mean.

Two things become apparent in the above plot. First, mobility is increasing nearly linearly with time. This implies that the average action potential of transacting wallets on the network is increasing at a fixed rate — as the network grows, more wallets tend to be increasingly connected (higher out-degrees) and active (higher out-flow). As a blockchain network matures, we expect its mobility slope to slowly level off until the average action potential stabilizes. Second, the hypothesis that mobility captures instantaneous volatility appears to have merit: mobility fluctuations either lead or move in tandem with price fluctuations. This is perhaps more evident when we detrend mobility by considering the absolute value of the first differences as shown in Figure 2 below:

Figure 2: Absolute first differences in mobility with the same smoothing as above

In general, we find that absolute first differences in mobility are elevated during spans of high price volatility; this enables mobility to effectively serve as an on-chain indicator for volatility. Comparing the value of mobility and other graph-derived metrics across various blockchain-powered assets can yield insights into the differential evolution of the associated network structures and consequently their connectivity and utility.

Boltzmann provides investors with a suite of metrics that proxy supply, demand, and flows within token-enabled blockchain networks to enhance risk management strategies in today’s markets.