Network participants process transactions and receive fees from users (supply-side fees).

Cosmos Hub is an interoperability & security platform for blockchains and applications. Cosmos was founded by Ethan Buchman & Jae Kwon and launched in 2019.

ATOM holders own and govern the protocol. They currently take a cut of the total transaction fees paid by users (revenue). Cosmos’s investors include Paradigm, IOSG Ventures, 1confirmation, etc.

governance

github

twitter

discord

etherscan

website

Blockchains (L1)

Cosmos Hub

# Cosmos Hub Active addresses (weekly) Methodology

## 1. Overview

Weekly active addresses for Cosmos measures the number of unique addresses that initiated transactions on the Cosmos Hub blockchain within any 7-day rolling window. This metric provides insight into user engagement and network activity by tracking distinct transaction senders over weekly periods. The calculation treats transaction sender addresses as "active addresses" and uses approximate counting techniques for efficient computation across rolling weekly windows.

## 2. Data sources

The calculation relies on the following raw data sources:

- **Cosmos Hub blockchain events**: Message events from the Cosmos Hub blockchain containing transaction metadata and event attributes
- **Transaction sender data**: Sender addresses extracted from message event attributes in JSON format  
- **Block timestamp data**: Block-level timestamps from Cosmos Hub transactions for time-based aggregation
- **Event type filtering**: Message events filtered specifically for `message` event types to identify transaction activity

## 3. Step-by-step calculation

### Component A: Transaction sender extraction
- **Purpose**: Identify unique transaction senders from Cosmos Hub blockchain events
- **Logic**: The system processes Cosmos Hub message events and extracts sender addresses from the JSON-formatted `event_attributes` field. Only events with `event_type` equal to `message` are included, ensuring focus on actual transaction activity. Each transaction sender is classified with metadata including chain identifier (`cosmos`), market sector (`blockchains_l1`), business segment (`blockchain`), and user type (`Transaction sender`)
- **Key fields**: Block timestamp (`block_timestamp`), transaction hash (`transaction_hash`), and sender address extracted from JSON (`sender`)

### Component B: Daily user aggregation  
- **Purpose**: Calculate daily unique active addresses as the foundation for weekly calculations
- **Logic**: Transaction data is grouped by day using truncated block timestamps. The system counts distinct sender addresses per day, creating daily snapshots of unique users. This daily aggregation includes all relevant metadata dimensions like chain, user type, market sector, business segment, and product identifiers
- **Key fields**: Daily timestamp (`timestamp`), timestamp grain (`day`), and distinct user addresses (`user_dau`)

### Component C: Weekly rolling window creation
- **Purpose**: Generate 7-day rolling windows for weekly active address calculation  
- **Logic**: The system creates a comprehensive date spine covering the full time range of available data. For each day in this spine, it generates a 7-day lookback window by creating an array of dates from 6 days prior through the current day. This ensures every day has a complete 7-day window for weekly calculation, even when some days within the window have no transaction activity
- **Key fields**: Window timestamp (`timestamp`) and constituent daily timestamps (`joinday`)

### Component D: HyperLogLog approximation
- **Purpose**: Efficiently count approximate distinct users across 7-day windows
- **Logic**: For memory efficiency with large datasets, the system uses HyperLogLog (HLL) probabilistic data structures to approximate distinct user counts. Daily user data is first converted to HLL sketches using 15-bit precision. These sketches are then merged across each 7-day window to produce approximate weekly unique user counts. The HLL approach provides high accuracy while using significantly less memory than exact distinct counting
- **Key fields**: HLL sketches (`hll`) and extracted approximate counts (`user_wau`)

### Component E: Dimensional breakdowns
- **Purpose**: Calculate weekly active users across multiple analytical dimensions
- **Logic**: The system generates weekly active user counts across four dimensional breakdowns: project-level aggregation (`project_dbt_tag` - representing overall Cosmos metrics), chain-level breakdown (by individual chain), user type breakdown (by transaction sender type), and product-level breakdown (though product_id is null for Cosmos). Each breakdown uses HLL sketch merging to combine daily user data into weekly approximations
- **Key fields**: Dimension type (`dimensions`), chain identifier (`chain_id`), user type (`user_type`), product identifier (`product_id`), and weekly user count (`user_wau`)

### Final aggregation
- **Combining components**: The system takes the weekly active user calculations and filters specifically for `chain` and `project_dbt_tag` dimensions, then renames the `user_wau` field to `active_addresses_weekly` to reflect that addresses represent active users in blockchain contexts
- **Currency conversion**: Not applicable for this address-count metric
- **Time aggregation**: Weekly periods are calculated as 7-day rolling windows, where each day represents the end of a weekly period looking back 6 days
- **Final calculation**: The final metric represents the approximate count of distinct addresses that sent transactions within any 7-day period, with missing time periods filled with zero values to ensure complete time series data

*This methodology was automatically generated by Token Terminal's methodology analyzer.*

Unique active addresses over a rolling 7-day window.

Cosmos HubATOM

Active addresses (weekly)

Active addresses (weekly)

Competitive landscape