Loops HouseCapture without churn

NEAR HACKATHON
Links
Team
1 member- ALOwner
Aloys Jehwin J
Overview
An investment thesis on NEAR, tested with data anyone can rebuild.
Question: in February 2026 NEAR Intents switched on a fee share, keeping part of the fees partner apps charge. Did the swaps leave?
Answer: supported in part.
- NEAR's take per $100 swapped rose from 1.42 to 2.11 cents (1.49x).
- Volume would have had to fall 32.7% to cancel that gain.
- Against 7 cross-chain peers (THORChain, Chainflip, Maya, LI.FI, SOCKET, Rango, Bungee), volume moved -1.9% (90% interval -18.9% to +21.5%). The break-even loss is ruled out; a smaller loss is not.
- Since then revenue is 4.9x, and 77% of that growth is the higher cut. Partners did not raise their fees to pass the cost on.
How: a synthetic control against rule-selected peers, a permutation interval, the same test run on every peer (it finds an effect in 0 of 7), and 693 ways of running it (median +2.3%). Fixed decision rules in code, every threshold in one config file.
NEAR's own data: a sample of 1,048,000 swaps from NEAR's Explorer API, partner by partner. The 10 largest partners sent 0.94x their earlier volume after the switch and 1.29x now. Evidence against is reported too: without SwapKit, smaller fee-charging partners sent 0.48x while no-fee partners sent 0.93x, and Infinex and Rhea left.
Reproducible: Python standard library only. "python3 -m cwc verify" proves every committed number, chart and sentence matches a fresh rebuild from stored data, byte for byte, offline. The live dashboard can also pull today's data and redo the headline sum in the browser.