HackaTriathlon

Demo

About

Biometric Prediction Markets on Arc A real-time prediction market platform where spectators bet on live athlete performance using verifiable physiological data. Heart rate, HRV, steps, pace, and cadence stream from a BLE chest strap and iPhone sensors through a relay app to an onchain settlement layer — turning every race, workout, or competition into a live trading floor. Use cases: Crowd-sourced sports analytics, live event engagement, performance forecasting, and biometric research. Spectators take positions on whether an athlete's heart rate crosses a threshold, HRV drops under stress, or pace holds over an interval. Markets resolve deterministically from sensor data — no oracle trust assumptions. Architecture: A Polar-compatible BLE chest strap transmits heart rate measurements including RR intervals to an iOS relay app (Swift, CoreBluetooth, CoreMotion). The app computes sliding-window HRV metrics (SDNN, RMSSD) on-device and uploads telemetry batches to a Node.js backend. The backend stores samples in Supabase, publishes snapshots to Chainlink CRE, and resolves settlement queries. A Solidity prediction market contract on Arc handles market creation, binary AMM trading, and USDC-collateral payouts. The frontend (TypeScript, Vite, SVG) renders live telemetry, a real-time HRV distribution bell curve, and trading interfaces — all settled with sub-second deterministic finality.

How it's made

Technical Architecture The data pipeline begins at a Bluetooth Low Energy chest strap broadcasting Heart Rate Measurement characteristics (0x2A37), including beat-to-beat RR intervals at 1/1024s resolution. An iOS relay app (Swift, CoreBluetooth) parses the BLE payload, extracts BPM and RR intervals, computes RMSSD and SDNN over a 30-second sliding window using Gaussian statistics, and queues samples with elapsed-millisecond timestamps into a durable on-device store. CoreMotion's CMPedometer runs in parallel, capturing steps, cadence, and pace. A background upload worker batches samples over HTTPS to a Node.js backend with exponential backoff and network-aware retry. The backend (TypeScript, Vite, ws) ingests telemetry into Supabase, maintains a session registry, and publishes snapshots to a Chainlink CRE telemetry receiver contract. Settlement resolution queries the nearest sample to a target elapsed time and evaluates threshold comparisons for any signal type. On Arc (chain 5042002, EVM-compatible), a Solidity 0.8.24 PredictionMarket contract implements a constant-product binary AMM with USDC collateral (6-decimal ERC-20 at 0x3600...0000). Market lifecycle: create, seed liquidity, take YES/NO positions, close, request settlement, fulfill via operator, claim. Gas paid natively in USDC at ~$0.01/tx with deterministic sub-second finality. The frontend renders live SVG telemetry charts and a vertical Gaussian KDE distribution of RR intervals that reshapes in real time as data streams in.