L0product

Prediction Game (Sui)

Atomic settlement proof: rugby prediction game exercising PTBs, zkLogin, oracles, and Dynamic NFTs in one shippable product.

900

Priority Score

Pain × Demand × Edge × Trend × Conversion

Customer Journey

Why should I care?

Five cards that sell the dream

1Why

One tap, one transaction, settled.

What if predicting a rugby match was as simple as liking a post?

The friction: EVM prediction: install wallet, approve tokens, place bet, wait for oracle, claim winnings — 4-5 transactions, 2-5 minutes, 5 failure points.

The desire: Sui PTB composes predict + stake + settle + payout into one atomic operation. Minutes compress to seconds. Gas decisions eliminated.

The proof: The value stream map shows the compression — 4-5 EVM txs collapse to 1-2 Sui PTBs.

Value stream compression

Picture

A split screen: left shows 5 blockchain transaction confirmations stacking up with gas fee popups. Right shows a single tap on a rugby match card and a green checkmark. Cinematic, 16:9

2Evidence

No wallet, no gas, no friction.

What if onboarding to crypto felt like signing into Gmail?

zkLogin: OAuth credentials become a Sui address via zero-knowledge proof. Sponsored transactions: platform pays gas, user never holds SUI. First prediction in under 60 seconds from first visit.

EVM flow	Sui flow
Install wallet extension	Click "Sign in with Google"
Save seed phrase	Nothing
Buy gas token	Nothing
Approve token spend	Nothing

Frictionless onboarding stories

Picture

A phone showing Google sign-in flowing directly into a rugby prediction confirmation. No wallet popup, no gas estimation, no extension install banner. Cinematic, 16:9

3Platform

Hot potato prevents the bugs.

What if the programming language itself made double-settlement impossible?

Move's type system makes double-settlement impossible at compile time.OracleResult has no store, copy, or drop — must be consumed atomically.

Clock guard prevents early settlement. The compiler catches what auditors miss on EVM.

This isn't a feature. It's a constraint that eliminates an entire class of bugs.

Oracle safety stories

Picture

A diagram showing an OracleResult object with no abilities — it must be consumed in the same transaction that created it. No way to store, copy, or reuse it. Cinematic, 16:9

4Loop

Contracts, oracle, frontend, dogfood.

What's the fastest path to proving atomic settlement works?

Sprint	Ships	Proves
1 (2 wks)	Core contracts + zkLogin + sponsored tx	Prediction created on testnet via zkLogin
2 (2 wks)	Oracle + settlement + receipts	Full settlement cycle in single PTB
3 (1 wk)	Frontend polish + history + onboarding	Complete user journey browse to settle
4 (1 wk)	Admin tools + leaderboard + dogfood	50 predictions from real users

Build order and dependencies

Picture

A 4-track Gantt chart showing parallel workstreams converging at 'First Settlement' milestone. Dark background, crimson milestones. Cinematic, 16:9

5People

Rugby fans who back convictions.

Can sports be the trojan horse for crypto onboarding?

Who	Their job
Rugby fan (crypto-curious)	Back convictions without wallet install or gas
Crypto native	Prove prediction skill with on-chain receipts
Builder (dogfood)	Validate Sui settlement patterns for other products

Kill signal: Zero predictions after 30 days = thesis wrong. Dynamic NFT receipts evolve with results — proof of conviction, not just a bet.

Who we test with

Picture

A group watching rugby on a big screen, phones out, each showing a prediction card with their pick. One phone shows a 'Won' receipt NFT glowing green. Cinematic, 16:9

1 / 5

Same five positions. Different seat.

The customer sees frictionless prediction. The builder sees settlement stack proof. The outer game and inner game are the same loop — ship a game, prove a stack.

Feature Dev Journey

How does this get built?

Five cards that sell the process

1Job

The job: prove the stack works.

Can you ship one game that uses every piece of the settlement stack?

Situation: 4-5 EVM transactions per prediction. Wallet install, gas management, manual claims. Every step is a failure point and a user loss.

Intention: One product exercising PTBs, zkLogin, oracles, sponsored tx, and Dynamic NFTs. If it works here, the patterns transfer.

Obstacle: The oracle. No sports data feed exists on Sui. Contracts and frontend are solved problems. The oracle is the novel dependency.

Intent contract

Picture

A rugby stadium at night, overlaid with a transparent Sui PTB diagram showing predict → stake → settle → payout flowing as one atomic arrow. Cinematic, 16:9

2Stories

Every story has a number.

If the THEN doesn't name a data source and threshold, what are you actually testing?

Story	THEN assertion	Forbidden
S1: Browse	5+ fixtures in <3s, no wallet	Wallet required to view
S2: Predict	Single PTB via zkLogin, 0 gas	Partial execution
S3: Settle	1 PTB settles all, hot potato consumed	Multi-tx settlement
S5: Safety	Duplicate rejected, admin override	Double-settlement

Full story contracts

Picture

A test dashboard — rows of story IDs, each with a red or green indicator, thresholds visible. Dark background, terminal aesthetic. Cinematic, 16:9

3Build

60% pattern reuse, 40% new.

What already exists that you haven't wired yet?

Layer	Exists	Build
Move contracts	14 packages, 2,500+ lines	7 new modules, ~1,500 lines
Frontend	dApp Kit, shadcn, Tailwind	4 new pages, prediction components
Oracle	Nothing	Agent-operated HTTP oracle
Gas station	Nothing	Sponsored tx backend

Dependency map

Picture

Architecture diagram split into layers — contracts, oracle, frontend, infra — with reuse percentages. Crimson highlights on new code. Cinematic, 16:9

4Sprints

Four sprints to first settlement.

Which sprint proves the thesis — and which just ships features?

Sprint	Features	Acceptance
1	#1-4, #14, #15	Prediction created on testnet via zkLogin
2	#5-9	Full settlement cycle in single PTB
3	#10-13, #16-17	Complete user journey from browse to settle
4	#18, #12	50 predictions from real users

Build order

Picture

A timeline with four sprint blocks — each block lights up as features ship. Progress bar filling. Dark background, crimson milestones. Cinematic, 16:9

5Validation

50 predictions or we stop.

What's the smallest experiment that proves the settlement stack works?

Kill signal: Zero predictions on testnet after 30 days, OR PTB settlement requires more than one transaction. Either means the thesis is wrong for this use case.

Gate	Evidence
L1	Core contracts deployed on testnet
L2	Full predict → settle → payout in single PTB
L3	50+ predictions, zero double-settlements
L4	Mainnet audit passed, monitoring live

Commissioning gates

Picture

A kill switch on a dark panel — crimson countdown timer, the number 50 prominent. Stark, minimal, cinematic, 16:9

1 / 5

The pitch is the shape. The flow diagrams prove the thinking. The VV stories validate the value.

Flow Diagrams VV Stories

Problem

Situation

On-chain prediction markets require 4-5 separate transactions on EVM — wallet install, token approval, prediction, oracle settlement, user claim. Each step loses users. Each transaction costs gas. Each is a failure point.

Intention

A prediction game where the entire lifecycle — browse, predict, settle, pay — happens in 1-2 atomic transactions with zero wallet setup and zero gas decisions.

Obstacle

No sports data oracle exists on Sui. zkLogin and sponsored transactions haven't been integrated into a real product. The settlement PTB composition pattern is unproven at the application level.

Hardest Thing

The oracle. Move contracts and frontend are engineering problems with known solutions. A reliable sports data feed that submits results on-chain with attestation — and handles conflicts, delays, and edge cases — is the novel dependency.

Scorecard

Priority (5P)

4/5

Pain

3/5

Demand

5/5

Edge

5/5

Trend

3/5

Convert

Readiness (5R)

Principles4 / 5

Performance2 / 5

Platform3 / 5

Process4 / 5

Players2 / 5

What Exists

Component	State	Gap
Move contracts (14 packages)	Partial	Existing DePIN/identity packages on testnet. New prediction_game package needed (7 modules).
Frontend (Next.js 15 + dApp Kit)	Working	Stack exists. New prediction app needed in NX monorepo.
Oracle	Missing	No sports data oracle on Sui. Custom agent-operated oracle required.
Gas station	Missing	Sponsored transaction backend needed.
zkLogin	Missing	Not yet integrated into any deployed product.

Frozen Scope

5 feature IDs from Onchain JTBD:

ID	Capability
WCAP-007	Gas abstraction — sponsored transactions
WCAP-008	Seedless onboarding (zkLogin)
WCAP-011	NFT management — Dynamic prediction receipts
WCAP-019	Atomic multi-operation settlement (PTB)
WCAP-020	Oracle data integration

Relationships

PRD	Contributes
Onchain JTBD	The capability register this PRD advances (WCAP-007, -008, -011, -019, -020).
Sui Patterns	Hot potato, Dynamic Display, clock guard patterns used in contracts.
Sui Standards	Display, sponsored txs, kiosk standards.
Sui Protocols	Oracle + Settlement protocol layers.
Sui Business Dev	Strategic context and 5Ps analysis.

Kill Signal

Zero predictions on testnet after 30 days of deployment, OR PTB settlement requires more than one transaction. Either means the thesis is wrong for this use case. Kill date: 2026-06-30.

Questions

What would this game need to prove before you'd trust the same settlement patterns for real money?

If the oracle is the hardest thing, why are we building contracts first — is that courage or avoidance?
The hot potato pattern prevents double-settlement at compile time — but what class of bugs does it NOT prevent?
If rugby fans complete 50 predictions on testnet, does that prove the UX works, or just that they're tolerant?