ONE TRADE, FOUR FORMS
Every risk-transfer industry performs the same operation: a risk seller and a risk buyer agree on an uncertain event, and value moves when it resolves. People trade risk at random through games, for profit through predictions, and as protection through insurance. The industries differ only in where the truth comes from and who participates — not in the trade. RAIN serves all three on one primitive: trustless resolution, instant pre-collateralized settlement, and an underwriting vault that prices spread against variance regardless of what the distribution is called.
GAMES · LIVE PREDICTIONS · LIVE VIA RAIN MARKETS PARAMETRIC INSURANCE · SUPPORTED PERSONAL INSURANCE · ROADMAP OUTPERFORM MARKETS · LIVE (ORACLE v1, NOT PROVABLY FAIR YET)
1GAMES — RISK BOUGHT AT RANDOM · LIVE
Blackjack and roulette run today on Arbitrum One over RAIN Channels: ~300 ms per action, zero gas, provable fairness — every card and spin a mutual commit-reveal, every state dual-signed, every dispute objectively decidable on-chain. The games are the protocol's live proof under the most adversarial conditions possible: if it can settle a $520k blackjack hand in 300 ms with fraud proofs, it can settle any risk.
Who it's for: gaming operators — bond into HouseBond, run a house node, plug in new games via IRulesVerifier. Live games · SDK & builder guide · play the demo ↗
2PREDICTIONS — RISK TRADED FOR PROFIT · LIVE VIA RAIN MARKETS
The Rain prediction-markets side (built by the Rain markets team) runs permissionless markets on world events: creation, trading, order book, liquidity, split/merge — with an AI + dispute + appeal resolution layer that serves as the protocol's oracle for real-world outcomes. It ships as rain-sdk-v2 on npm and is wrapped unmodified inside the unified SDK's rain.predictions module.
Who it's for: trading firms and market builders — trustless transaction builders for every market action, api-backed reads for indexing and order books (labeled as such). Unified SDK
3PARAMETRIC INSURANCE — RISK SOLD AWAY AS PROTECTION · SUPPORTED
A parametric cover is a prediction-market position bought as protection. The event is public-data-verifiable — flight delayed ≥ 3 h, rainfall > X mm, price < Y — so no claims process exists: the data either printed past the threshold or it did not. Everything needed is live today: the resolution layer (Rain markets' dispute system or signed data feeds), underwriting capacity (the risk vault), and instant pre-collateralized settlement.
Worked example — flight-delay cover
| ROLE | WHO | WHAT HAPPENS |
|---|---|---|
| Risk buyer | Traveler with a €400 non-refundable connection | Buys €400 of the YES side of “flight AB123 arrives ≥3 h late” at the market price (say 10% → €40 premium). Intent: protection, not profit. |
| Risk seller | Market counterparties + the risk vault (underwriting capacity) | Sell the risk for the ≈€36 expected margin; exposure is bounded by the vault's exposure rules, payout collateralized before the position opens. |
| Resolution | The prediction-markets resolution layer / signed flight data | Reads the public arrival record. Disputed outcomes escalate through the bonded dispute + appeal game. |
| Settlement | The protocol | If late: the traveler is paid €400 the moment the market resolves — no adjuster, no cheque, no delay. If on time: underwriters keep the premium. |
The precise line. What is supported today is parametric cover on public-data events — a capability the live components already serve. Personal/individual insurance — private loss events, per-claim assessment, quote-based underwriting, long-lived reserve management — is explicitly Stage 2 / roadmap. We state the line plainly rather than blurring it.
Who it's for: insurance companies — distribute parametric products on protocol rails, or deploy a dedicated instance settling in your own token (token-agnostic deployments). Whitepaper §7P
4RELATIVE-PERFORMANCE MARKETS — “WHICH ASSET OUTPERFORMS” · LIVE (OUTPERFORM)
A fourth form of the same trade, live on reference deployment #2: 5-minute pari-mutuel rounds on BTC / ETH / SOL / BNB. Players buy shares of the coin they expect to move most over the round; the price curve pays dividends to earlier holders as later entries arrive; the winning coin's pool is split by shares, minus a disclosed rake. Rounds are keeper-settled on-chain in the CryptoRace v1.1 contract (0x340991442B5BFCD80a46D709748be28b788414B7, Sourcify exact_match).
Honest scope: open/close prices come from a house-signed off-chain oracle v1 (median of OKX + Bybit REST tickers, EIP-712-attested via OffchainOracleAdapter 0x47b33BbE635682E7164f0A4f33D02e25592D006e). This is not provably fair yet: the contract enforces signer, race binding, one attestation per race and a 60-second deadline (late/no attestation ⇒ round voids, everyone refunded), so the house cannot settle late with cherry-picked prices — but it can choose not to settle. The Pyth path is planned: a Pyth adapter is deployed and re-enabled by setOracle through a 48-hour timelock.
Who it's for: trading firms and anyone who wants a relative-performance instrument with a 5-minute horizon. Trade it live ↗ · @rain/markets quickstart
+RANDOMNESS AS A PROTOCOL SERVICE
The games vertical forced the protocol to solve trustless randomness at interactive speed: two-party commit-reveal hash chains — unbiased if either party is honest — with on-chain fraud proofs and a Pyth Entropy session seed as defense-in-depth. Any builder who needs cryptographically fair random outcomes can consume the same machinery. Randomness research · Channels spec