Loops HouseNEAR Intents thesis: usage vs value capture

NEAR HACKATHON
Links
Team
1 member- EROwner
Erfan Mohebolkhames
Overview
NEAR's raw on-chain activity fell more than 90% from its peak while NEAR Intents grew into billions of dollars of cross-chain volume a month. This project asks whether the NEAR token captures that new usage, and answers with evidence anyone can re-run.
One command (python build.py) pulls public, keyless data (NearBlocks daily chain stats since 2020; DefiLlama NEAR Intents volume and fees, NEAR DEX volume, TVL), computes every metric, runs a seeded lead-lag permutation test, and renders six charts, a static dashboard (GitHub Pages) and THESIS.md. Every number in the thesis is filled from metrics.json, and the build fails on any unfilled number. python build.py --offline rebuilds the committed snapshot exactly; pytest checks determinism.
Findings (snapshot 2026-11-05): Intents moved $32.9B in the last 365 days, earns about $74.5M a year in fees and is 84% of NEAR DEX volume. The tested hypothesis (Intents volume leads NEAR's price better than raw activity) is NOT supported at weekly or 4-week horizons. Fee burn offsets 0.84% of issuance, and market cap is about 70x annualized Intents fees. THESIS.md sets out the evidence for, the evidence against, what public data cannot answer (where Intents fees go), and concrete results that would change the conclusion.
NEAR fit: the analysis is entirely about NEAR mainnet and NEAR Intents activity. No contract is deployed.