{"sources": [
  {"url": "https://raw.githubusercontent.com/gtmcknight/amm-sim/e65314f88de945f6d055bfc2aab01c482ab6570e/research/solved/README.md", "quote": "- the router likelihood per price hypothesis, each hypothesis carrying its own rival spot;"},
  {"url": "https://raw.githubusercontent.com/gtmcknight/amm-sim/e65314f88de945f6d055bfc2aab01c482ab6570e/research/solved/README.md", "quote": "Not tried: choosing quotes against the whole lattice posterior instead of its mean and sd, per-run estimates of sigma and order rate inside the classifier, the router likelihood per price hypothesis."},
  {"url": "https://raw.githubusercontent.com/gtmcknight/amm-sim/e65314f88de945f6d055bfc2aab01c482ab6570e/research/solved/README.md", "quote": "Earlier scalar router reweighting lost badly: do not reuse it without deriving the observation event correctly."},
  {"url": "https://raw.githubusercontent.com/gtmcknight/amm-sim/e65314f88de945f6d055bfc2aab01c482ab6570e/research/solved/README.md", "quote": "- A policy sees what afterSwap sees. `run.py` enforces it; a result from any other harness needs the same blindness."},
  {"url": "https://raw.githubusercontent.com/gtmcknight/amm-sim/e65314f88de945f6d055bfc2aab01c482ab6570e/research/solved/README.md", "quote": "- Tune on replica seeds 60000 and up. Score once on 10000..12047 and report the `paired{...}` line `pair.py` gives against the nearest reference in `ref/`."},
  {"url": "https://raw.githubusercontent.com/gtmcknight/amm-sim/e65314f88de945f6d055bfc2aab01c482ab6570e/research/solved/real.py", "quote": "for _ in range(min(dt - 1, 12)):          # fully silent steps\n                v += SIG2\n                _, m, v = trunc(m, v, lo, hi)"},
  {"url": "https://api.imdnotdead.fun/artifacts/22819a1542e36a609ec3d02a160b4b7aaf345d26d7184ef6517abc35b85cfbfc", "quote": "{\"key\": \"fresh3-1\", \"idea\": \"Cost-weighted arb/retail call: keep the lattice posterior as is, but read the first own trade of a step as the arb (phase 0, no rival-spot update) only when its posterior arb probability exceeds 0.9 instead of 0.5, because mistaking a retail fill for the arb costs more than the reverse.\""},
  {"url": "https://api.imdnotdead.fun/artifacts/22819a1542e36a609ec3d02a160b4b7aaf345d26d7184ef6517abc35b85cfbfc", "quote": "\"holdout_paired\": {\"n\": 2048, \"mean_diff\": 0.3229, \"se\": 0.0107, \"sum\": 661.282652, \"sumsq\": 692.740899"}
]}