A note on coverage

We are estimating coverage for some label types where it makes sense to do so: these addresses reflect patterns consistent with the type of address we are labeling (e.g. a contract that is engaging in an event akin to a swap, but for which we do not know the protocol or operator responsible for its deployment).

However, coverage estimates for many label types are not feasible because they reflect only truths that are reported via project documentation, or have small N. Examples of these include L2 addresses, dapps (that don't fit into a standard defi bucket), centralized exchanges, operators, and known garbage or spam addresses that we label as flotsam.