Lead GitHub discovery with the Fee Reality benchmark — the negative result, reproducible, as the README's opening hook

[eaitbrahim]

Jesse has eight years of SEO, tutorials, and GitHub traction. keel's fastest hook is the finding itself: no shipped rule family is net-positive at the taker fee actually paid. Publishing negative results is already constitution line 8 — this is the brand's core result made the front door, not marketing invented for reach.

The asset

• A README section leading with the fee-drag benchmark: standard TA rule families (RSI/MACD/turtle-class) shown twice — fee-blind, then priced honestly (taker fee + measured spread at the clip size actually traded) — collapsing to sub-1.0 net PF.
• A ~10-second terminal capture (GIF/asciinema-style) of a real run, not a mockup.

Hard constraints — these are the issue

• Every number reproducible from a recorded run in the repo (fix(experiments): resolve the sweep harnesses' paths, so a recorded run can be re-run #265 made recorded runs re-runnable). The asset renders the ledger's numbers or it does not ship. No hand-typed figures.
• No competitor naming in the asset. Generic fee-drag math — "RSI at 0bp illusion vs 60bp reality," never what another product costs you. Trademark posture: nominative use on the compare page is one thing; an animated callout is another.
• Dated captions; per-venue fee basis named in the frame.
• Generation tooling committed (the terminal-shot pattern already exists in the site repo and can be transposed), so the GIF regenerates when the measurement does.

Done when

• README leads with the benchmark before the feature list
• The GIF/frame numbers each link their experiment record
• Site counterpart (interactive widget) cross-linked: keeltrading.com 136