| Loading… |
How the buckets are decided
Every bucket comes from plain arithmetic on what the CAM says. The AI never decides a bucket; it only explains one afterwards.
Secured bucket: three steps
Step 1. Capacity: how big a loan can this person repay?
- Monthly surplus = income × FOIR − EMIs of loans still running today − monthly commitments (SIP, insurance and similar) found in the CAM.
- Tenure = the smaller of 20 years and (70 − age). 70 is the last age we fund a secured loan, so the closer to 70, the shorter the tenure and the smaller the amount. Above 70 there is no tenure and no secured loan. If the age is unknown, 20 years is assumed.
- Loan amount = surplus × the number of months in the tenure. No interest rate is assumed. For example, 20 years is 240 months, 10 years is 120 months and 5 years is 60 months.
- Any loan we already disbursed to the same customer is subtracted.
| Loan amount | Capacity bucket |
|---|---|
| Above 5 Cr | Excellent |
| Above 3 Cr | Good |
| 1 Cr to 3 Cr | Medium |
| Below 1 Cr | Low |
Step 2. Quality: can the person's profile move the bucket by one step?
| Points | When |
|---|---|
| +1 | Lives in an owned house |
| +1 | Doctors: more than 7 years of experience with a graduate degree (MBBS, BDS ...), or 5 or more years with a postgraduate degree (MD, MS ...). Everyone else: 5 or more years. |
| +1 or +2 | Investments and reserves (FD, mutual funds, shares and similar, added together) more than 20 lakh is +1, more than 50 lakh is +2. Only what the CAM shows is counted. |
| −1 | Rented house with family (married), or a hostel / PG. Rented with marital status unknown is also −1. |
| −2 | Rented house, living single (unmarried) |
| −1 or −2 | 3 to 4 credit enquiries in 3 months is −1, 5 or more is −2. More than 10 in 6 months is −2. The worse of the two counts. |
| −1 or −2 | 5 or more small running loans (under 5 lakh) whose EMIs together take 10% to 20% of income is −1, over 20% is −2. Small EMIs are not punished. |
A total of +2 or more moves the bucket one step up. A total of −2 or less moves it one step down. It never moves more than one step, and it never lifts a lead that has no capacity left.
Step 3. Caps: hard limits that can only lower the bucket.
| CIBIL score | Highest bucket (secured and unsecured) |
|---|---|
| Above 750 | No cap |
| 700 to 750 | Good |
| 651 to 699 | Medium |
| 650 or below | Low |
Age is handled in step 1 through the tenure (secured loans end at age 70). If the CAM does not give income or FOIR, the lead shows Insufficient data instead of a bucket.
Unsecured bucket
- Loan amount = surplus ÷ 2 × 100, minus anything we already disbursed to the same customer.
- Above 2.5 Cr is Excellent, above 1.5 Cr Good, above 50 L Medium, above 25 L Low, otherwise Very Low.
- The last age we fund an unsecured loan is 65. Above 65 nothing can be funded (eligibility 0, Very Low, no exceptions). Ages 60 to 65 are capped at Low. Negative remaining capacity is capped at Low.
- The CIBIL cap in the table above applies to unsecured too.
- There is no quality step for unsecured.
Priority slots for unsecured leads
The slot comes from the MIS credit and ops remarks. It is plain keyword matching, and the dashboard shows which words decided it.
| Slot | What the remarks show |
|---|---|
| P1 Act now | Eligibility mentioned, more than 3 enquiries, OGL, RCU, PD call not attended, CIBIL good, documents problem, or customer not picking up |
| P2 Fix in 3 months | Bounces in the last 3 months, eligibility short or "not eligible", or any other reason that can be sorted out |
| P3 Hard to fix | CIBIL low, 3 or more unsecured loans, bounces older than 3 months, or a very negative reason that will not clear in 3 months (written off, settled, legal case, fraud, FOIR or income too weak) |
| P4 No remark | Neither a credit remark nor an ops remark was given |
If a lead matches reasons from more than one slot, it goes to the most serious one. For example, CIBIL good plus a bounce in the last 3 months is P2, and customer not picking up plus low CIBIL is P3. The rules and keywords live in remark_slots.py.
All the numbers above (rate, tenure, retirement age, points, thresholds) live at the top of financial_analysis.py. If you change them there, update this page too.