A retail attempt at sports-market arbitrage and prediction-market trading, run June–July 2026 with real money on the line. Every thesis was measured against pre-registered, frozen verdict rules — and every thesis failed or was stopped:
- Model A (live in-play hedge, Hard Rock + Kalshi): stopped 2026-07-07 — the hedge leg never filled; the market worked as designed.
- Model B (pre-game arb): stopped 2026-07-07 — zero passing locks, then the data feed died.
- Plan D (slow-window tennis vs Pinnacle fair, Kalshi + Polymarket):
failed its frozen Phase-0 gate 2026-07-08 — price deviations reverted at
coin-flip rates (0.496/0.538 vs the required 0.60) on both venues
independently. Record:
docs/status/plan_d_verdict_2026-07-08.md.
Nothing here is running anymore and none of it made money. It's published
as-is because the process was the interesting part: pre-committed kill
criteria, unattended measurement, and arithmetic verdicts that don't
negotiate. Start with AGENTS.md, docs/doc_index.md, and the status memos
in docs/status/.
Nothing in this repository is trading advice. The opposite, if anything.