{"key":"fresh2-1","idea":"Refine the latent-price lattice to reduce boundary-cut and rebinning error while keeping the likelihood model and quote tables fixed.","settings_tried":[{"setting":{"cell_width_bps":1.25,"window_half_width_bps":150,"recenter_threshold_bps":100},"n":256,"mean_diff":-0.008219236142543718,"se":0.00936641096399404},{"setting":{"cell_width_bps":0.625,"window_half_width_bps":150,"recenter_threshold_bps":100},"n":256,"mean_diff":-0.0006338232217470985,"se":0.010254581020406337},{"setting":{"cell_width_bps":1.875,"window_half_width_bps":150,"recenter_threshold_bps":100},"n":256,"mean_diff":-0.0056827547913220755,"se":0.008883627191823689}],"confirm_paired":{"n":256,"mean_diff":-0.009635100074174341,"se":0.010103136002436147},"holdout_paired":{"n":2048,"mean_diff":0.0019057169269703633,"se":0.0040415591506397,"sum":3.902908266435304,"sumsq":68.4845873367989,"worst5":[[11218,-1.5760909422745044],[10842,-1.1865477163734113],[11451,-0.8554201311013685],[11250,-0.7384823826518527],[11676,-0.6479073767926593]]},"misread_rate":{"base":0.162248261260289,"mine":0.1621374117674669},"price_error_bps":{"base":13.64283973652866,"mine":13.640883953957534},"gas_estimate":{"extra_persistent_probability_cells":360,"extra_temporary_cells":360,"extra_persistent_slots_naive":360,"extra_multiply_adds_per_full_support_diffusion":55170,"extra_gas_per_one_step_callback_rough":3000000,"measured":false,"assumptions":"Naive one-slot-per-cell EVM implementation; 481x123 vs 121x33 full-support convolution, plus longer scans/rebinning. Silent steps repeat diffusion. No Solidity port; exceeds practical gas/storage limits."},"code_zb64":"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","code_sha256":"10615bf93b53d1dfcc16e4b692cbf3b8b126b198e6c084609550b0bd6f1091be","notes":"First 1.25-bps refinement lost 0.0082 on tuning seeds; all three refinements had nonpositive tuning means. Selected 0.625 bps by largest tuning mean, then confirmed once on 60256..60511; no retuning. Physical window and recenter threshold fixed. Code hashes frozen before holdout; four PROCS=8 run.py commands covered 10000..12047 exactly once. Baseline accelerator matched plain Python exactly on three tuning seeds; selected pure implementation parity checked on the same three seeds. Primary module needs standard Python and supplied solved modules/tables only; optional accelerator falls back without numpy/numba. Price metric is callback-weighted RMS log-price error; misread metric is first-trade-weighted. This numerical refinement is not a useful policy improvement and should not be ported. Next try analytic within-cell boundary truncation rather than adding cells. No engine or leaderboard claim."}
