How this works

The problem is the benchmark, not the appetite

A rural lender cannot hedge rural distress. A hotel group cannot hedge a collapse in tourist arrivals. A merchant cannot hedge payment-rail failure. The reason is not lack of demand and not regulation — it is that no settlement benchmark exists.

Without an objective published number to settle against, every deal has to be negotiated from first principles. That is a three-month structuring exercise per trade, and the next pair negotiates something slightly different, so nothing is comparable and no price ever forms. The index collapses all of that into one pre-agreed definition, leaving two strangers to agree on price and size only.

Severity, not occurrence

Contracts never settle on whether an event happened. They settle on how far a continuously measured number moved from an agreed reference level. That is what makes the instrument a derivative rather than a wager.

The index quotes an absolute level, not a deviation — a heat index for May reads 41.0 because the mean daily maximum was 41.0°C. It is checkable against the source in ten seconds by a non-expert, it removes the discretionary choice of a baseline, and it survives a data revision cleanly. The baseline does not disappear; it moves into the strike, where it is a commercially negotiated term between two counterparties rather than a parameter we impose.

One multiplier, every contract

₹1,50,000 per index point. A 0.1-point tick is ₹15,000. The minimum clip is twenty lots, there are no fractional lots, and no sub-lot splitting or assignment.

One lot is therefore between roughly ₹45 lakh and ₹3 crore of gross notional. Retail participation is already impossible at the membership gate, but membership rules can be circumvented by nominee structures and introducing brokers, and ticket size cannot. The size barrier is arithmetic, not policy we could later relax.

The honest trade-off: a ₹1.5 lakh multiplier makes the contract lumpy. A hedger whose exposure is ₹40 lakh per point can buy 26 lots or 27 and is about 1.5% over- or under-hedged either way. We accept that. Granularity is a real cost and a smaller one than a retail-accessible product would be.

Institutions with proven exposure

Hedgers document real economic exposure at onboarding. Their position limit follows by arithmetic: documented exposure per index point divided by ₹1,50,000 is the maximum number of lots. The check is not a judgement call, which is exactly what makes it enforceable at the moment a counterparty accepts.

Liquidity providers qualify on balance sheet and risk capability instead, under a market-maker classification. The two-class structure is a legal firewall rather than a marketing segmentation: commercial interest is what removes these contracts from wagering law.

What is fixed and what is negotiated

The platform fixes the machinery: which index, the multiplier, the tick and clip, the settlement mechanics and fixing source, the disruption and data fallback rules, the documentation architecture.

Members negotiate the economics: notional in lots, observation window and tenor, strike or trigger level, price, and payout structure within approved variants. It still feels like a genuine negotiation, because everything a member actually cares about sits on that side — but two completed deals stay comparable, which is what lets a price be published for the next participant.

Pre-trade scrutiny runs before the deal exists

Binding acceptance runs three gates — documentation in place, a bilateral credit line available, the hedger within their exposure-based limit — and then a surveillance pass, before anything is written. A blocking finding means the trade is never formed, rather than formed and then unwound.

The surveillance pass combines a quantum-simulated search for anomalous flow structures with a machine-learning screening layer and a deterministic rule pack. The models can raise a score; only a rule can produce a finding. A block always needs a rule behind it, because stopping a member's deal on a model output nobody can restate as a fact is not something we could defend to that member or to a regulator.

We calculate, we never hold

Money does not pass through the venue at any point. Premium and final settlement move counterparty to counterparty through their banking units. Collateral sits with a third-party custodian; we calculate the requirement and instruct.

This is deliberate. It keeps our capital requirement small, keeps us out of the credit chain, and keeps us clear of custody and payment regulation, which are separate licensed activities with far heavier requirements. There is no payment code anywhere in this system, and that is visible on inspection.

No machine learning in the settlement path

The index computation is deterministic, version-pinned to a published methodology, and computed twice by two independently written engines that must agree before anything is published. A model inside the settlement calculation would make the benchmark unauditable and, we believe, unapprovable.

AI is used everywhere else — anomaly detection on source data, entity resolution, relationship mapping, market-abuse detection. The principle throughout is that AI proposes and deterministic code disposes.

What is honestly still open

This is a pre-authorisation build. Index history is generated rather than ingested from the real bulletins; membership and exposure evidence are self-attested rather than verified by a licensed KYC provider; the custodian and banking-unit integrations are simulated; and the quantum layer runs on a simulator rather than physical hardware. Each of those is labelled where it appears.

What is real is the structure around them: the publication chain, the dual computation, the lot arithmetic, the gates, the surveillance, and the audit trail.