Pyth Insight
Team: @Defilion
Submitted: March 23, 2026
──────────────────────────────────────
Answer Capsule
Pyth Insight uses Pyth Price Feeds (live + historical Hermes API) to empirically test whether published confidence intervals hold up statistically across 1,687+ feeds, Pyth Benchmarks OHLCV data to compute realized volatility and a novel CI/RV alignment score, and Pyth Hermes BTC/USD oracle attestations as a verifiable entropy source for a provably fair prediction game.
──────────────────────────────────────
What It Does
Most DeFi protocols use Pyth’s confidence intervals as liquidation guards without ever checking whether those CIs are statistically accurate.
Pyth Insight is the first platform to test this empirically fetching 72+ historical Hermes snapshots per asset and measuring whether the published ±1σ band actually captures 68.3% of subsequent price moves.
It also powers a provably fair prediction game seeded by live Pyth Entropy, a Volatility Intelligence dashboard using Pyth Benchmarks OHLCV data, and an AI analyst with all 1,687+ live feeds injected as real-time context.
──────────────────────────────────────
Pyth Features Used
Check all that apply:
☐ Price Feeds (on-chain or off-chain)
☐ Entropy (randomness)
Both
──────────────────────────────────────
Links
Live Demo: https://pyth-insight.vercel.app
Source Code: GitHub - DE-LIGHTPRO/pyth-insight: AI oracle intelligence platform for Pyth Playground Hackathon 2026 · GitHub
Video Walkthrough: https://youtu.be/NHZgKFj0ERw
──────────────────────────────────────
Screenshots / Media
──────────────────────────────────────
Tech Stack
Framework/Language: Next.js 15 App Router, TypeScript (strict mode), Tailwind CSS, Recharts
Blockchain (if applicable): Base mainnet — Pyth Entropy contract 0x98046Bd286715D3B0BC227Dd7a956b83D8978603
Agent Framework (if applicable): N/A
Deployment: Vercel
Pyth APIs used:
Pyth Hermes REST — live prices + CI data for all ~1,687 feeds (/v2/updates/price/latest)
Pyth Hermes Historical — 72+ snapshots per asset for CI calibration (/v2/updates/price/{timestamp})
Pyth Benchmarks — OHLCV for realized volatility + sparklines (benchmarks.pyth.network/v1/shims/tradingview/history)
Pyth Entropy — full RevealedWithCallback eth_getLogs pipeline + Fortuna REST; falls back to live Hermes BTC/USD oracle attestation packed into a 32-byte verifiable seed
@pythnetwork/hermes-client SDK
What makes it original:
The CI/RV Alignment metric compares live Hermes CI width against realized volatility from Pyth Benchmarks directly showing which assets Pyth is over- or under-confident about.
The CI Calibration page tests the statistical validity of Pyth’s ±1σ claim with real historical data. Both are unique and directly useful for DeFi protocol risk engineers.
──────────────────────────────────────
Content Contributions (Required)
Public Post (Dev.to): I stress-tested Pyth Oracle's confidence intervals and built a provably fair game seeded by live oracle attestations - DEV Community
Technical Contribution (GitHub Gist): Using Pyth Hermes oracle attestations as a verifiable entropy source — seed packing recipe` · GitHub
Bonus — X Platform Post: x dot com/Delight/status/2036045110903705689
Bonus — Wikipedia Contribution: en.wikipedia dot org/wiki/Blockchain_oracle (Pyth Network section)
──────────────────────────────────────
Licensing
This project is licensed under Apache 2.0 (LICENSE file in the GitHub repo above).
──────────────────────────────────────
Eligibility Confirmation
I am 18+ years old
I am not located in an OFAC-sanctioned jurisdiction
I confirm this is an original work created during the hackathon period
I have read and agree to the Terms & Conditions