Radar Timeline

Blockchain data can tell fascinating stories, but few are sifting through the transactions to identify the storylines. Recent blockchain forensic analysis on the fall of QuadrigaCX is a good example of this. A retroactive analysis of Quadriga's on-chain activities highlight just how dire their situation was – something that many now regret not researching earlier. The good news is, tools like Google's BigQuery and eth.events are making it increasingly easy to peer into blockchain data, and begin to tell these stories.

I've chosen to focus much of my ongoing analysis on a specific ERC-20 token, Chainlink. The entire Chainlink transaction data set (excluding internal transactions, and event data) weighs in at a very manageable 55MB.

Measuring Network Activity

There are a number of ways to visualize network activity, including a simple Daily Active Users approach  – counting unique addresses transacting on-chain each day. I've been working on a few ways to visualize network activity in a more insightful way. I often start my research with a question:

Even with a relatively small data set, grouping activity can be a helpful path to meaningful analysis. The Chainlink community created a way to group on-chain addresses using Marine Corps rankings based on how many LINK tokens are currently held. These ranks started as a meme to encourage accumulation, but I've found them to be very helpful during data analysis.

The Chart

This interactive radar timeline chart shows daily net transaction value on the Chainlink network, broken down by group.

Addresses are grouped by Marine rank. The 18 Marine ranks are then divided into percentile groups in increments of 10, resulting in 180 total groups of addresses, based on their current LINK balance. The chart shows the daily net value of LINK transacted by each group in 2019. Addresses are only included if they have a LINK balance > 1 on the last day of the data set.

The axis has a max and min value of 70K and -70K LINK, despite nearly 1% of the data points having values outside that range. Including the entire range on the axis makes the majority of the daily fluctuations difficult to discern and compare, especially in the lower Marine ranks.

Recent LINK network analysis indicates that almost all on-chain activity is related to interactions with exchanges. Understanding this, I've removed exchange addresses from the visualization. Although the exchange addresses are not present, addresses that interact with exchanges are included and the value of those transactions are considered.

Sliding through the days on the chart, you may notice there is more green (likely accumulation) than red (likely selling). Normally, this would be an indication that an increasing amount of LINK is coming off exchanges to users on-chain addresses. If you've been following me on Twitter for a while, you'll know that is not the case - the total balance on exchanges has remained fairly steady for many months, and on-chain address counts have also mostly plateaued.

The way in which the data was compiled has increased the amount of green on the chart. The addresses are grouped by their Marine rank on the last day of the data set, and addresses are only included if they have a LINK balance > 1 on the same day. This means that if an address held 100K LINK at the start of the year, but transferred all of the LINK to an exchange before the end of the sample period, neither their address nor the value of their transactions are represented in the chart.

There are no bombshell takeaways from the visualization, but it does a pretty good job of answering the question of who has been most active on the network.

I haven't seen blockchain transactions visualized this way before, and overall I'm pleased with how the data is presented. As the Chainlink network evolves, and the nature of transactions can be categorized by their utility, I anticipate that radar timelines will be a useful tool to understanding how the network is changing.