Address Cohorts

For most API endpoints you can see the cohort parameter. A cohort is a group of addresses joined together based on some common property.

We define the following cohorts:

• total = all addresses holding the asset, we divide this category into the following 3 sub-cohorts:

• users = all addresses excluding smart contracts and addresses belonging to exchanges,

• cex = addresses belonging to centralized exchanges,

• sc = smart contract addresses.

Addresses in the 'users' cohort can also be subdivided into different cohorts. We use 3 different divisions:

• balance cohorts divisions,

• balance cohorts (USD) divisions,

• age cohorts division.

Balance Cohorts

Here, addresses are divided into cohorts based on how much balance they have.

This division is very popular for BTC and ETH analysis. In the case of BTC, the most commonly used breakdown is: 0-0.01 BTC, 0.01-0.1 BTC, ...

We decided to replicate this idea for altcoins. However, all altcoins have different supply, due to which we had to make different partitioning into cohorts for each coin.

Definition of cohorts for an asset

We implemented the balance cohorts as follows. Let S be the maximum supply for some asset. Let T = S * 0.01 / 21,000,000 (after calculation, T is usually rounded). Then for that asset, the balance cohorts would be:

• balance_cohort_0 [0, T]: all addresses whose balance of the considered asset is between 0 and T,

• balance_cohort_1 [T, 10*T],

• balance_cohort_2 [10*T, 100*T],

• balance_cohort_3 [100*T, 1,000*T],

• balance_cohort_4 [1,000*T, 10,000*T],

• balance_cohort_5 [10,000*T, 100,000*T],

• balance_cohort_6 [100,000*T, 1,000,000*T],

• balance_cohort_7 [1,000,000*T, 10,000,000*T].

Here 0.01 / 21,000,000 is the percentage for BTC, which is used as minimal threshold in its cohorts sampling.

Example. $GRT has a maximum supply of 10.7B $GRT. Then T = 10.7B * 0.01 / 21,000,000 = 5. Then blanace_cohort_0 for this asset will have a range of 0 $GRT and 5 $GRT, and blanace_cohort_1 will have a range of 5 $GRT and 50 $GRT.

Notes:

• Only addresses that do not belong to centralised exchanges and are not smart contracts are considered for this division. Thus, none of the cohorts listed above include smart contract addresses or addresses of centralised exchanges.

• This division is separate for each asset. For instance, an address can be in balance_cohort_0 for an asset A and in balance_cohort_3 for another asset B.

Balance Cohorts (USD)

Here, addresses are divided into groups according to the dollar value of the balance belonging to each address. We define the following cohorts of addresses:

• $0-100: a group of addresses that have a balance of the asset being considered in dollar terms in the range of $0 to $100,

• $100-1k,

• $1k-10k,

• $10k-100k,

• $100k-1M,

• $1M-10M,

• $10M+.

Notes:

• Only addresses that do not belong to centralised exchanges and are not smart contracts are considered for this division. Thus, none of the cohorts listed above include smart contract addresses or addresses of centralised exchanges.

• This division is separate for each asset. Example: let an address have an asset A worth $150 and an asset B worth $4000. Then in the metrics for asset A it would be representative of the $100-1k cohort and for B it would be representative of the $1k-10k cohort.

• Since this cohort division uses thresholds in dollar terms, the categorisation of addresses depends on both changes in balances and price.

Age Cohorts

Here the category is assigned to an address depending on how long the address holds the coin considered.

To determine the category, we calculate the volume-weighted average holding time of a coin for each address. For example, if an address received 1000 coins 5 weeks ago and 10 coins 3 hours ago, its average holding time at the moment will be equal:

$$\text{Average holding time} = \frac{(1000 \cdot 840\ \text{hours} +10 \cdot 3\ \text{hours} )}{1000 + 10} = 831.7 \ \text{hours} \approx 5 \ \text{weeks}$$

This address therefore has an average holding time of around 5 weeks, as it has acquired most of its coins at this time.

We define the following age cohorts of addresses:

• <1d: all addresses with an average coin holding time of less than 1 day,

• 1d-1w: all addresses with an average coin holding time between 1 day and 1 week,

• 1w-1m,

• 1m-3m,

• 3m-6m,

• 6m-1y,

• 1y-2y,

• 2y-3y,

• 3y-5y,

• 5y-7y,

• 7y-10y.

Notes:

• Only addresses that do not belong to centralised exchanges and are not smart contracts are considered for this division. Thus, none of the cohorts listed above include smart contract addresses or addresses of centralised exchanges.

• This division is separate for each asset. Example: let an address have an asset A with an average holding time of 2 weeks and an asset B with an average holding time of 2.5 years. Then in the metrics for asset A it would be representative of the 1w-1m cohort and for B it would be representative of the 2y-3y cohort.