Documents were reviewed
We reviewed the submitted documents. Document review and group issuance paths fall here.
Scoring engine
Badges become levels,
levels become limits
HACS decides on two axes. The level is set by what you have proven, and it unlocks credit limits and products. The score says how reliable you are within that level. The score is never used for promotion — a level rises only through proof.
Level ladder
The level ladder
A level rises when every required badge is held at evidence grade L2 or higher (confirmed by an institution against its own records). Level 2 opens only with verified income. Students start at Level 1; migrant workers with verified wages start at Level 2.
| Level | Required badges | Example limit (KRW) | Example limit (LAK) | Lenders shown |
|---|---|---|---|---|
| 0 · NONE | No badges | 0 | 0 | 0 |
| 1 · Identity FOUNDATIONAL | identity_verified + residence_attested (L2 or higher) | 1,000,000 | 8,100,000 | 3 |
| 2 · Income INCOME | + employment_verified (4 major social insurance schemes or employer confirmation, 6 months or more) | 5,000,000 | 40,500,000 | 8 |
| 3 · Assets FINANCIAL | + asset_attested | 20,000,000 | 135,000,000 | 12 |
Amounts are converted by each country’s income level (World Bank 2024: Laos ×0.0544 · Vietnam ×0.1222 · Thailand ×0.1940 · Korea ×1.0). Level rules are data that can be set per country, and they change only through a draft → review → approval process. Final credit limits and interest rates are set by partner lenders.
Evidence grades
Same badge, different weight: the source decides
An institution confirmed it against its own records
Issued or endorsed directly by a university, employer or institution. This is the baseline for level promotion.
Confirmed directly from source data
Facts confirmed directly at the source, such as connectors and lender notices. These receive the highest cap.
| Issuance path | Cap | Issuance path | Cap |
|---|---|---|---|
| ISSUER_DIRECT | L2 | CONNECTOR | L3 |
| ENDORSEMENT | L2 | LENDER_NOTICE | L3 |
| DOCUMENT_REVIEW | L1 | GROUP_ISSUED | L1 |
Final grade = min(path cap, issuer trust grade). The structure itself leaves no room for forgery or exaggeration.
Play it forward
What each proof is worth
Pick some badges. You will see right away the points each axis gains at each evidence grade. HACS shows borrowers, in numbers, what to do next.
Points added per axis
Total axis points of the selected badges (production badge weights)
Each axis score flows into the 0–850 score through the points-table weights
Limit step-up
On-time repayment doubles the limit
| Level | Max with good repayment (KRW) | Max with good repayment (LAK) |
|---|---|---|
| Level 1 | 2,000,000 | 16,200,000 |
| Level 2 | 10,000,000 | 81,000,000 |
| Level 3 | 40,000,000 | 270,000,000 |
Even Level 1 at ×2.0 is designed to sit below the Level 2 base limit, so there is always a clear reason to bring the next proof. Any arrears, default or write-off in the last 12 months resets the multiplier to ×1.0. Multipliers can be adjusted per country.
The score
0–850, 6 axes, 31 factors
Total = Σ(axis score × weight) ÷ Σ(axis maximum × weight) × 850 — rounded only once. An axis with no data, and some factors, are set to a neutral score rather than 0 (the rule differs by factor).
| Axis (expert-v3 per-level points table) | Maximum × weight |
|---|---|
| Identity | 127 × 1.0 |
| Financial activity | L1 168 × 0.75 · L2 218 × 0.5 · L3 313 × 0.5 |
| Payment reliability | 170 × 1.5 |
| Credentials & education | L1 99 × 0.75 · L2·L3 127 × 0.75 |
| Social trust | 85 × 1.0 |
| External data | 128 × 1.5 |
5 tiers
Score bands and tiers
Every factor
Every factor and its points, by level
Every factor and its points are public, with the levels that count it. Level 1 counts 29 factors, level 2 counts 31 and level 3 counts 35.
| Factor | Points | L1 | L2 | L3 |
|---|---|---|---|---|
| Identity — 127 × 1.0 | ||||
| Identity verification completed | 40 | counted | counted | counted |
| Account age | 25 | counted | counted | counted |
| Authority of the ID document | 25 | counted | counted | counted |
| Document standing and rejections | 22 | counted | counted | counted |
| How long the verified identity has stood (full at 2 years) | 10 | counted | counted | counted |
| Reachable by phone and email | 5 | counted | counted | counted |
| Financial activity — L1 168 × 0.75 · L2 218 × 0.5 · L3 313 × 0.5 | ||||
| Transactions in the last 90 days | 40 | counted | counted | counted |
| Months with an incoming payment (6 points each) | 35 | counted | counted | counted |
| Credit line utilisation | 30 | counted | counted | counted |
| Digital-asset holding habit | 25 | counted | counted | counted |
| Wallet age | 20 | counted | counted | counted |
| Bank balance kept out of overdraft | 10 | counted | counted | counted |
| Card held | 8 | counted | counted | counted |
| Transaction amount in the last 90 days | 50 | — | counted | counted |
| Wallet balance | 40 | — | — | counted |
| Balance cushion (months of inflow) | 20 | — | — | counted |
| Savings | 20 | — | — | counted |
| Digital-asset value | 15 | — | — | counted |
| Payment reliability — 170 × 1.5 | ||||
| Months with a payment | 55 | counted | counted | counted |
| Steadiness of payments between quarters | 35 | counted | counted | counted |
| Loan repayment outcome | 35 | counted | counted | counted |
| Payments without a failure | 30 | counted | counted | counted |
| Time since the last failed payment | 15 | counted | counted | counted |
| Credentials · education — L1 99 × 0.75 · L2·L3 127 × 0.75 | ||||
| Issuer authority | 39 | counted | counted | counted |
| Share of credentials still valid | 33 | counted | counted | counted |
| Passes the tamper check | 12 | counted | counted | counted |
| Degree level | 10 | counted | counted | counted |
| Kinds of credentials held | 5 | counted | counted | counted |
| Attested tenure | 28 | — | counted | counted |
| Social trust — 85 × 1.0 | ||||
| How far the employment claim checks out | 25 | counted | counted | counted |
| Connections to verified workers | 20 | counted | counted | counted |
| Digital business-card exchange | 20 | counted | counted | counted |
| Community activity and response | 15 | counted | counted | counted |
| Business card held | 5 | counted | counted | counted |
| External data — 128 × 1.5 | ||||
| External data (points by source below) | 128 | counted | counted | counted |
The 128 external points by source — counted today: in-app activity and referrals 14, phone signals (app use, stability of where the phone spends the night) 7 · connected with a partner: open banking 36, utility bills 28, phone bills 27, business registration and tax 10, e-commerce 4, psychometric test 2. A source not yet connected scores neutral, never zero.
Loan repayment outcome starts from base factors for late (0.35) and default (0.10) and also weighs days past due and later repayment and time. Bank-alert summaries the customer pastes in are collected but not yet scored.
How it moves
How the score goes up and down
↑ Payments
Paying on the due date
Each month paid on the due date adds to the paid-months count, reaching the maximum at 8 months. Repaying in full raises the repaid-loan count (+60~85).
↑ Cash flow
External deposits
6 points for each 1 month of observed external deposits. The bank cash-flow badge counts from Level 2.
↓ Arrears
Arrears, default, write-off
Arrears 0.35x, default 0.10x, write-off 0. Up to 50% additional deduction per 180 days past due.
↓ Anti-gaming
Self-transfer detection
If self-transfers exceed 30% of volume, the two digital-asset factors are set to 0. Each rejected document costs 1/3 of the points.
◆ Digital assets
Multi-chain wallets
Assets on Xphere, Ethereum, BNB Chain and Kaia are assessed with 90-day seasoning and count only within a cap.
◆ Phone signals
Aggregated on the device, numbers only
Only the consented aggregate numbers are sent — never raw records, never coordinates. They are weighted at 7 points, in line with the signal strength measured on public data.
Explainable by design
Every score explains “why”
31 reason codes, up to 4 per response. In an additive scorecard the impact points are mathematically identical to SHAP values — the explanation is not an approximation, it is the calculation itself.
{
"score": 712, "tier": "GOOD",
"reasonCodes": [
{ "code": "BADGE_IDENTITY_VERIFIED", "impact": +30 },
{ "code": "BADGE_RESIDENCE_ATTESTED", "impact": +15 }
]
} // illustrative response shape
Who it serves
One person, one business, one village
There is a model for each country × subject type × segment. The same engine reads different proofs of life, each in its own way.