Loops HouseNEAR on the Evidence

NEAR HACKATHON
Links
Team
1 member- ABOwner
ABDELDJALIL CHOUGUI
Overview
NEAR on the Evidence is a NEAR investment thesis built as a re-runnable test suite. One command pulls public data with no keys: NEAR mainnet RPC, the FastNEAR archival RPC, NearBlocks, DefiLlama, CoinGecko and GitHub. It stores the data in a dated snapshot with a source URL and fetch time on every file, then checks 9 pre-registered claims against kill criteria written down in advance. THESIS.md and a static dashboard (light/dark, works offline) are generated from that output. No number is typed by hand.
What the data says (snapshot 2026-11-11):
- Value capture has moved to NEAR Intents: $53.7M in Intents fees over the last year vs $1.17M in L1 gas fees, on $32.9B of volume.
- Monetary policy, checked on-chain: archival bisection of the protocol config finds the inflation cap cut from 5% to 2.5% (protocol v81) and the gas burn raised from 70% to 100% (v87). Popular dashboards still describe the old 70% burn. Realised net supply growth is 2.51%/yr.
- Bear case, reported as clearly as the bull case: burns plus Intents buybacks offset only 18.6% of issuance value, and daily active accounts are down 75% YoY. The usage-growth claim is FALSIFIED.
- The track prompt's own hypothesis ("Intents volume leads usage better than price") fails our lead-lag test.
An interactive scenario model (Python CLI plus dashboard sliders, kept identical by a parity test) shows how much Intents volume would be needed for value accrual to match issuance, under your own assumptions.
NEAR's role: the thesis uses NEAR's own RPC as the ground truth for monetary policy and security (validators, Nakamoto coefficient 9, 44% of supply staked), and tests whether chain-abstracted settlement through NEAR Intents is where value actually accrues.
Stack: Python standard library only, 28 tests, vendored Chart.js, GitHub Actions for a daily refresh to Pages.