Wallet Classification Attributes - Allium Documentation Hub

wallet_classification is a daily wallet snapshot derived from stablecoin activity. This page documents every column and the exact cutoffs behind each classification. For how rows are selected and how to query efficiently, see the Overview.

Filter by chain on every query — the crosschain table spans 30+ chains and is clustered on snapshot_date. See the Overview for query patterns.

Table Columns

Unique Key: chain, address, snapshot_date

Identity & snapshot

Column Name	Data Type	Description
`chain`	VARCHAR	Blockchain the wallet was active on, lower case (e.g. `ethereum`). Always filter on this.
`address`	VARCHAR	Wallet address. A wallet active on multiple chains appears as one row per chain.
`snapshot_date`	DATE	The wallet’s most recent day of recorded stablecoin activity. Rolling 30-day metrics are computed as of this date.

Size & balance

Column Name	Data Type	Description
`balance_tier`	VARCHAR	Size bucket based on the wallet’s maximum stablecoin balance ever held (see thresholds below).
`current_balance_usd`	FLOAT	Stablecoin balance (USD) as of `snapshot_date`. On chains without native end-of-day balances this is reconstructed from net transfer flow and is best-effort.
`volume_30d_usd`	FLOAT	Total stablecoin transfer volume (sent + received) in the rolling 30 days ending on `snapshot_date`.

balance_tier cutoffs (by lifetime max stablecoin balance):

Tier	Max balance (USD)
`whale`	≥ 1,000,000
`large`	≥ 100,000
`medium`	≥ 10,000
`small`	≥ 1,000
`micro`	≥ 25
`dust`	< 25

Wallet type

Column Name	Data Type	Description
`wallet_type`	VARCHAR	`Consumer`, `Business`, or `Institutional` — assigned by a layered rule set: attribution labels first, then behavioral patterns (cross-chain reach, volume-to-balance ratio, balance volatility, balance drawdown, token diversity), then a balance-tier fallback. Internal-only addresses resolve to `Excluded`.

wallet_type blends signals, not a single threshold. As a rough guide: Institutional = very high volume, multi-chain reach, or large balances with high turnover; Business = high, steady throughput with a meaningful balance floor (≥

1,000 average 30-day balance, ≥

10,000 on Tron to account for high USDT velocity); Consumer = everything else with genuine end-user activity.

Lifecycle

Column Name	Data Type	Description
`behavioral_segment`	VARCHAR	Activity lifecycle stage (see thresholds below).
`combined_segment`	VARCHAR	`balance_tier` + `behavioral_segment`, title-cased (e.g. `Whale Active`, `Medium Dormant`). Convenient single grouping key.

behavioral_segment definitions:

Segment	Definition
`new`	First seen within the last 30 days
`active`	Had activity within the last 30 days
`dormant`	No activity in the last 30 days, but active within the last 90
`churned`	Last activity more than 90 days ago

Activity metrics

Column Name	Data Type	Description
`days_since_last_activity`	NUMBER	Days between `snapshot_date` and the query date.
`days_active_30d`	NUMBER	Distinct days with activity in the rolling 30 days (0–30).
`days_active_lifetime`	NUMBER	Distinct days the wallet has ever been active.
`lifetime_span_days`	NUMBER	Days between the wallet’s first and last activity. `0` if both fall on the same day.
`tokens_used_count`	NUMBER	Distinct stablecoin tokens ever transferred by the wallet (lifetime).
`chains_used_count`	NUMBER	Distinct chains the wallet’s stablecoin activity was observed on, as computed within the source. In the per-chain tables this is `1`; aggregate across rows (`group by address`) for a true multi-chain count.

Engagement

engagement_score (0–100) measures how consistently active a wallet is. It is the sum of three sub-scores.

Column Name	Data Type	Description
`engagement_score`	NUMBER	`recency_score + frequency_score + consistency_score` (0–100).
`engagement_tier`	VARCHAR	Bucketed `engagement_score` (see thresholds below).
`recency_score`	NUMBER	0–40, from how recently the wallet was active: `40` if active in the last 7 days; `30` if within 14 days; `20` if within 30 days; otherwise decays toward 0.
`frequency_score`	NUMBER	0–40, from `days_active_30d`: `40` at ≥25 active days, `35` at ≥20, `30` at ≥15, `25` at ≥10, `15` at ≥5, else `days_active_30d × 2`.
`consistency_score`	NUMBER	0–20, from the share of the wallet’s lifetime it has been active: `min(20, (days_active_lifetime / lifetime_span_days) × 20)`.

engagement_tier cutoffs:

Tier	Engagement score
`highly_engaged`	≥ 80
`engaged`	≥ 60
`moderately_engaged`	≥ 40
`low_engagement`	≥ 20
`inactive`	< 20

Risk

risk_score (0–100) flags balance volatility and anomalous swings. It is the sum of two sub-scores. It is a behavioral/volatility signal, not a sanctions or compliance screen.

Column Name	Data Type	Description
`risk_score`	NUMBER	`volatility_risk_score + anomaly_risk_score` (0–100).
`risk_flag`	VARCHAR	`high` (≥ 70), `medium` (≥ 40), or `low` (< 40).
`volatility_risk_score`	NUMBER	0–50, from the 30-day balance coefficient of variation (stddev ÷ mean): `50` at CV ≥ 2.0, `40` at ≥ 1.0, `30` at ≥ 0.5, `20` at ≥ 0.3, else scaled down.
`anomaly_risk_score`	NUMBER	0–50, from the 30-day balance change: `50` at >500% change, `40` at >200%, `30` at >100%, `20` at >50%; also `40` when the balance is zero but 30-day volume exceeds $1,000 (drained wallet).

Risk scores reflect stablecoin balance behavior only. A high score signals volatility or large swings worth a closer look, not wrongdoing.

​Table Columns

​Identity & snapshot

​Size & balance

​Wallet type

​Lifecycle

​Activity metrics

​Engagement

​Risk

Table Columns

Identity & snapshot

Size & balance

Wallet type

Lifecycle

Activity metrics

Engagement

Risk