No server is involved. The ERA5-Land rainfall history is real. The pricing model runs on this page, in your browser. The guided demo replays, step by step, a real run on a private test blockchain recorded on 16 October 2026; its tokens have no value.
Local test identity
Public test accounts and valueless cgLOCAL tokens. Roles are demonstrative; they do not establish independent human governance.
Loading application configuration…
SEASONAL RISK, MADE VISIBLE
A steadier season. A clearer price.
Understand drought exposure before the season starts. Turn rainfall history into a transparent, testable insurance estimate.
EVIDENCE BEFORE COMMITMENT
Conceptual field plan · not a map
01 / THE EVIDENCE
30+complete seasons required for analysis
02 / THE SIMULATION
50kmaximum trials for each scenario
03 / THE PROTECTION
100%maximum exposure reserved as collateral
EXPLORE THE WORKSPACE
From evidence to understanding.
Start with a scenario. Inspect every assumption.
ON-CHAIN / SEPARATE FROM THE LAB
Live coverage & capital
Pool capital—Capital-provider equity · USDCReserved exposure—Active policy collateral · USDCApproved, unpaid claims—Reserved until payment · USDCActive policies—Read from deployed contracts
Coverage catalog
Loading contract state
Research estimates are kept separate from live products.
A policy, from agreement to payment.
FOUR DEFINED STEPS
01Agree the terms
Fix the region, index, premium and maximum payout before purchase.
02Reserve the capital
Fully fund the maximum exposure when the premium is received.
03Verify the observation
The approved oracle reports the agreed rainfall index.
04Settle the obligation
Pay the fixed benefit if triggered. Otherwise release the capital.
STRATEGY A / MODEL-BASED PRICING
Good decisions start with the weather.
Explore the uncertainty behind a season’s price.
01 30+ complete seasons→02 Monte Carlo simulation→03 Indicative premiumOFF-CHAIN
ERA5-LAND RESEARCH DEMO
Historical rainfall. A fitted Gamma model.
Run the model on the team’s ERA5-Land rainfall series to see the rainfall distribution and indicative price.
The default reproduces the team’s R script: the Gamma fit of MASS::fitdistr, R’s own random draws from seed 42, and a trigger at the fitted 10th percentile.
ERA5-Land history · model-simulated trialsInputs changed
INDICATIVE PREMIUM / HECTARE
—USDC
Research estimate. Not an offer of cover.
Expected loss—
Capital cost—
Operating fee—
Failure allowance—
Premium per hectare—
USDC / ha · display rounded to 2 decimals
Drought probability—Under the selected modelComplete input seasons—99.5% tail loss (VaR)—USDC · all covered hectares
The shape of a season
TRIALS
Hover over or focus a mark to inspect its value.
CAPITAL THAT STAYS RESERVED
—
Full maximum exposure. Tail-risk estimates do not reduce the collateral needed to activate cover.
99.5% VaR minus expected loss—
95% expected shortfall—
99% tail loss (VaR)—
One shared index: all hectares trigger together. Loss is either zero or the full payout.READ THE UNCERTAINTY
—
95% Monte Carlo interval for the trigger probability, conditional on this fitted model.
Input-season trigger rate—
Historical Wilson upper bound—
More trials reduce simulation noise. They do not create more climate history.
BASIS RISK · INDEX AGAINST A SIMULATED FARM
—
Share of trials in which the index payout and the simulated farm agree. Farm rainfall is the index rainfall plus normal noise with a standard deviation of 0.3 × the seasonal standard deviation (— mm), as in the team’s R script.
Index
Farm in drought
No farm drought
Index pays
—
—
Index does not pay
—
—
Farm drought, no payout—
Payout, no farm drought—
The noise model is the R script’s assumption, not measured farm rainfall.Method, provenance & limitations+
YOUR COVERAGE
Policies & payments
Connect a wallet to read its balances and policy records.
▤
No wallet connected
Your private keys remain in your wallet. This application never asks for a seed phrase.
FULLY COLLATERALIZED
Capital pool
Closed-season pool
Actual USDC balance—Token balance in the vaultAvailable for underwriting—Not necessarily withdrawableUnallocated premiums—Not counted as upfront capitalAccrued operating fees—Separate from investor equity
Provide capital
Deposits require an approved capital-provider wallet and an open funding window. Capital is exposed to insured losses.
Approving tokens does not deposit them. Each step requires a wallet transaction.
Redeem your shares
Redemption is available before the funding cutoff, after a finalized failed fundraise, or after sales have ended and every active policy and unpaid claim has been resolved.
Returns are not guaranteed. A successful claim reduces capital-provider equity.
Contract state has not been loaded.
ETHIOPIA · ERA5-LAND · OFF-CHAIN RESEARCH
Rainfall, with a record.
Persistent evidence
Download official ERA5-Land precipitation for Ethiopian research locations and preserve its source files locally. This reanalysis is research evidence; a live insurance oracle is not configured. ERA5-Land dataset ↗
Download long-term point history once, then reuse it in the Pricing lab or as a same-provider, same-point reference. This is not a homogenized climate-change analysis.
Requires at least 30 complete matching seasons before the analysed season. Missing seasons are disclosed, never filled.
Select a saved source version to inspect rainfall.
Queued weather downloads
Loading saved requests…
EVIDENCE BEFORE A VERDICT
A missing value is not a dry month.
Choose an Ethiopian point and up to 12 months. Incomplete months, conflicting source files and provisional data are kept separate from a drought decision.
Recent download attempts
No download attempts recorded.
Inspect an on-chain observation
The workspace above does not finalize Chainlink reports. An actual deployed observation can be inspected separately below.
PRIVATE PARTICIPATION
Apply, review and verify access
1. Prove wallet control. 2. Submit an application. 3. Complete external review. 4. Verify the on-chain approval.
Sign in to access your private applications. A wallet signature does not prove identity or land ownership.
Sessions expire after 30 minutes and are kept only in this page's memory.
Your applications
New application
AUTHORIZED OPERATOR
Participant review queue
Application details
Separate contract approval
Off-chain review has no effect on the contract. Review the unsigned transaction, sign with your operator wallet and verify its receipt.
No private key is sent to the application.
Private audit history
SIGNED APPROVALS
Operator signs. Participant submits.
An authorized operator can sign a reviewed approval without sending a transaction. The applicant then submits that signature to AccessRegistry. This is separate from the wallet sign-in message.
Sign in above and select an application to begin.
Evidence before approval
Calibration & basis risk
Nonbinding research
Turn saved rainfall evidence into a reviewable pricing package, investigate missed droughts, and compare locations. These tools do not approve terms, issue coverage, change policies or reduce collateral.
Presentation arithmetic
Reference counts only
1,013 triggers out of 10,000 reproduce the slides’ financial arithmetic. This does not recreate their historical climate run.
Loading reference…
01 · Saved ERA5-Land evidence
Prepare a calibration
Select complete history from one research point. Multiple saved snapshots can form one baseline; conflicting versions are rejected.
Original slide data and approved geography are still required to claim exact reproduction.
02 · Paired farmer observations
Measure basis risk
Compare each index observation with an explicit farm-loss label. Missing loss labels are never treated as “no loss”. Uploaded declarations are not independently verified.
The population share of missed losses and the rate among loss cases have different denominators.
03 · Explicit scenario assumptions
Explore spatial noise
A new illustrative experiment: fitted index rainfall plus independent Gaussian local noise. The local drought label is a rainfall proxy, not measured crop damage.
Results will be labelled synthetic and are not forced to match the slides’ 2.9%.
04 · Geographic diversification research
Compare aligned locations
The same historical year is sampled at every location, preserving observed dependence. Compare against a maximum-dependence benchmark with the same historical marginal losses.
No live multi-region policy is created. Full contractual collateral remains unchanged.
DEPLOYMENT & OPERATIONS
Readiness, without assumptions
Technical readiness
Loading…
Before policy sales
Code/configuration checks cannot certify legal permission, actuarial suitability or the quality of an independent security audit.
Identified insurer and permitted operating structure
Reviewed future-season product terms and pricing
Deployed CRE workflow and verified forwarder identity
Independent contract and oracle review
Actual collateral, funded operations and support
Contract manifest
Administration uses your multisignature wallet
Use the included CLI to generate unsigned Safe transaction batches for participant approval, workflow binding, observation/product registration and activation. Production administrator keys stay in your wallet. The local demonstration uses published test keys.
python -m services.cli admin --help
There is no administrator endpoint that writes a fabricated weather observation. The independent resolver has a separately disclosed adjudication role.