A risk score is only useful if a credit committee can defend it. “The AI said 720” convinces nobody. So when LegiScore returns a property risk score, every point of it decomposes into the evidence underneath. The score is built from five pillars and twenty-nine individual signals, and any number on the report opens into the records that produced it.
Here is what those pillars actually measure, and why each one keeps a lender out of a courtroom.
Pillar 1: Ownership history
The foundational question: does the seller own what they claim to sell, free of competing claims? Signals here include breaks in the chain of title, missing or unregistered links, mismatches between the recited consideration and the encumbrance certificate, benami indicators, and gaps where a transfer should exist but no instrument was found. A single unresolved break can move a property from bankable to do not fund on its own.
This is also where deeds meet the ground. Signals compare the deed against the record of rights and the survey record: does the extent match, and is the survey or sub-division number consistent across sources? A plot that is 200 square yards on paper and 150 on the revenue record is a problem long before it is a default.
Pillar 2: Loans and charges
A title can be clean and still be encumbered. This pillar reads the EC across the full statutory period and checks for subsisting mortgages, equitable charges, court attachments, tax arrears, and releases that were promised but never registered. The most common real-world failure we catch is a mortgage that the deed says was discharged, but for which no release deed was ever recorded.
Pillar 3: Court cases
Property fights leave a paper trail before they reach the borrower's file. LegiScore screens for pending suits, lis pendens entries, partition disputes among co-owners, injunctions, and attachment orders from civil and revenue courts. A clean title with an active partition suit between siblings is not a clean property, and the score says so.
Pillar 4: Approvals and permits
This pillar checks the property against the permissions that can void a transfer outright: building approvals, layout sanction, RERA registration where applicable, whether the land is classified for the use it's being sold for, whether it slipped between agricultural and non-agricultural status without conversion, SC/ST land-transfer prohibitions, and ceiling-law exposure. These are the signals that turn a funded loan into an unenforceable security interest, so they sit at the top of the report, not the footnotes.
Pillar 5: Market conditions
The last pillar looks outward from the file: guideline value against the declared consideration, how deep the local market actually is, and how readily the asset could be sold if the loan ever has to be enforced. A property can be perfectly titled and still be poor security if nobody is buying in that pocket. Security you cannot liquidate is not security, and the rating carries that.
Why twenty-nine signals, not one model
We deliberately did not build a single black-box predictor. Each signal is computed and shown on its own, then rolled into pillar scores and a headline number. That structure exists for one reason: auditability. A risk officer can see that a property scored low specifically because of an unreleased 2016 mortgage and a survey-extent mismatch, and can decide, with full evidence, whether to fund, fund with conditions, or walk away.
A score you can interrogate is a score you can stand behind in front of a regulator. That is the difference between a number and a bankable number.