diff --git a/FINDINGS.md b/FINDINGS.md index 6f0cac6b..a6360c2e 100644 --- a/FINDINGS.md +++ b/FINDINGS.md @@ -655,6 +655,19 @@ or nothing. The through-line of the week, now written as one sentence: **every taker edge died at the requote wall, and every instrument that measured the wall found the makers standing on top of it.** +Coda, same night: we simmed standing on the wall ourselves +(maker_quote_sim — quotes at oracle-fair−m, 25k tokens, ~31–50k +chain-graded fills per cell, every trade-through assumed to fill us +at queue-front). Best cell **+$0.37/fill**; every 4s-requote cell +−$7–9/fill. A stale quote buys below *stale* fair, and the flow that +reaches it is the informed flow — staleness IS the adverse selection. +The wall pays sub-second colocated speed and queue position, not a +fair-value formula at hobbyist latency. Maker pivot: killed at Stage +1, $0 spent, one evening. What survives the week: the concentrated +copy edge and its two pre-registered execution upgrades (#20/#21), +and the human-maker inventory-lean line — following the wall's +incumbents, not competing with them. + ## Repo layout - `insider.py` — the detector: z-score/p-value, timing/freshness/sizing signals, diff --git a/research/README.md b/research/README.md index ed71fcd8..88e01827 100644 --- a/research/README.md +++ b/research/README.md @@ -57,8 +57,11 @@ Layout: follow-on: inventory-lean signal sibling_sum_scan.py T4 — print-substrate sum-arb scan (artifact- dominated; needs standing-book data; parked) - maker_quote_sim.py T1 — crypto maker quoting at fair−m (stale-quote - latency model; re-run pending) + maker_quote_sim.py T1 — crypto maker quoting at fair−m: STAGE-1 + KILL 2026-07-23 (best cell +$0.37/fill at + m=4c/R=1s under queue-front optimism; every + R=4s cell −$7-9/fill — staleness IS the + adverse selection; ~31-50k graded fills/cell) requote.py crater→requote timing (feeds the bots' per-niche retry) forward.py scores frozen studies on new tape days → forward_ledger (payouts_for chain overlay mandatory; controls + sub5c