{"key":"b5-probe/probe-2","status":"complete_negative_result","recommendation":"Reject this probe; retain unchanged thr09 base. No engine or board submission.","idea":"After an ambiguous first callback, move just the quote facing the midpoint of the arb and retail price hypotheses toward that midpoint, capped at 6 log-bps. Offline next-callback rollouts optimize immediate expected edge plus mutual information, including elapsed time and absence of a callback. A frozen depth-three rule decides when to probe; the original thr09 belief, arb threshold, phase and rival-update logic are unchanged.","settings":{"base":"thrpol:ThrLattice thr=0.9 qm=qm_k1e-5.json, as supplied; no base tuning","candidate":"probe_policy:Probe setting=4 qm=qm_k1e-5.json","classifier_threshold_unchanged":0.9,"candidate_settings_max":4,"candidate_entropy_weights_Y_per_nat":[0.01,0.04,0.12,0.3],"selected_setting":4,"probe_log_quote_cap_bps":6.0,"unchanged_classifier":true,"price_hypotheses":"Arb: exact lp reading using fees in force. Retail: mean and variance of the same lattice prior cut to the old no-arb band, before mixing.","initial_ambiguity_gate":"First callback of step and 0.1 < pi < 0.9; only quote action is gated, not arb/retail classification.","frozen_rule":"Probe iff 0.1513898256464548 < pi <= 0.7323873326122237 and abs(lp_arb - mean_retail)*10000 > 22.75491225974058, subject to the initial gate and a nonzero feasible one-sided action.","quote_action":"Let mid=(lp_arb+mean_retail)/2 and ls=log(own_spot). Choose ask if mid>ls, else bid. Move that log quote boundary toward mid by at most 6e-4, clip to nonnegative fee and 10% fee cap, leave the other fee unchanged. Feed the actually returned fees into the original next-callback belief update.","no_blend_or_new_classifier":true,"freeze_utc":"2026-11-10T09:05:57.537214+00:00","frozen_file_sha256":{"probe_rule.py":"05a22c117805ac1827ac15bd710a064e018ab74ce167287c1c966e028911a08b","frozen.py":"6cdcc5ce9633a2dac36ab78e3cedbb88c74eaaf87d77898bd75a5ca24edd770f","probe_policy.py":"8db57a93fbb0bd24461760536cf493bdf7354d0dc78ca6f07a13e7babef53430","probe_base.py":"15626e7a3fedd549bd0b2ecc785c3c75f4b512aad0755e7c980b3a1594bfcbdf"},"selection":"The first three weights learned exactly no action and were not separately simulated. Setting 4 is the only active probe. It was retained as a frozen falsification experiment despite negative tuning and confirmation; no holdout-driven changes."},"offline_training":{"replica_seeds":[62000,62127],"replica_runs":128,"snapshots":512,"samples_per_run":4,"snapshot_sampling":"Deterministic reservoir, seed 1 separately per run; only afterSwap-derived values, through Blind.","hypotheses":2,"actions":2,"actions_description":["unchanged base quotes","single feasible midpoint-directed side perturbation"],"rollouts_per_hypothesis_action_snapshot":512,"local_continuations":1048576,"common_random_numbers_across_actions_and_hypotheses":true,"rollout_random_seed_rule":"710000 + snapshot index (synthetic local Monte Carlo, not additional replica seeds)","local_parameters":{"sigma_bps":9.45,"future_lambda":0.8,"current_step_remainder_lambda":0.4,"size_mean_Y":20.0,"size_log_sd":1.2,"future_steps_max":6},"transition":"Constant product K=1e6; rival arbitrage then own arbitrage on future steps; exact two-pool routing formulas for each sampled retail order; stop at first own callback or censor after horizon.","observation":"(elapsed steps 0..6, buy/sell direction, abs log post-spot movement in 4-bps bins with 0.25/0.5/0.25 neighbor smoothing), plus a no-callback censor symbol at horizon. Time and silence are observable; true branch is only an offline label.","objective":"E[edge accumulated until next callback/censor] + entropy_weight * I(current arb/retail branch; next observation). Mutual information equals expected posterior entropy reduction under this local model.","fitter":{"depth_max":3,"min_leaf":24,"split_quantiles":[0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9],"leaf_action":"probe iff summed modeled advantage positive"},"settings_modeled":[{"setting":1,"entropy_weight_Y_per_nat":0.01,"snapshots":512,"chosen":0,"mean_modeled_edge_delta":0.0,"mean_modeled_information_delta_nats":0.0,"tree":0},{"setting":2,"entropy_weight_Y_per_nat":0.04,"snapshots":512,"chosen":0,"mean_modeled_edge_delta":0.0,"mean_modeled_information_delta_nats":0.0,"tree":0},{"setting":3,"entropy_weight_Y_per_nat":0.12,"snapshots":512,"chosen":0,"mean_modeled_edge_delta":0.0,"mean_modeled_information_delta_nats":0.0,"tree":0},{"setting":4,"entropy_weight_Y_per_nat":0.3,"snapshots":512,"chosen":289,"mean_modeled_edge_delta":-0.0032577988571611347,"mean_modeled_information_delta_nats":0.019528934773718053,"tree":[0,0.7323873326122237,[1,22.75491225974058,0,[0,0.1513898256464548,0,1]],0]}],"limitations":["Retail hypothesis approximates the cut lattice by a Gaussian for offline rollouts only.","Fixed local rates and a fixed remainder-of-step model; no adaptation or true run parameters.","Finite horizon and coarse smoothed bins approximate next-observation information.","Immediate edge until the next callback is a surrogate, not the full continuation value. Actual paired performance, not modeled information, determines rejection."]},"tuning":{"seeds":[62000,62127],"vs_base":{"n":128,"mean_diff":-0.6981550182230767,"se":0.04059005687319346,"sum":-89.36384233255382,"sumsq":89.17243193921762,"worst5":[[62007,-1.9782937502482127],[62029,-1.7819667804993173],[62068,-1.695335927933911],[62047,-1.6883220218545603],[62106,-1.6839835298482058]]},"metrics":{"probe":{"n":128,"mean_edge":518.2887729189506,"first_trades":901484,"misread":272587,"first_trade_misread_rate":0.30237586024821295,"callbacks":1401183,"callback_squared_log_price_error_bps2":231355726.5557764,"callback_rms_log_price_error_bps":12.849691380983728,"mean_retail":546.4690961850636,"mean_arb":-28.180323266112953,"mean_rv":78184.53591133656,"mean_av":24946.06306320284,"rows_sha256":"67518d297d7cf9643b0adefe357eee1f2daeb36c48aa674edb79a95f3921a682"},"base":{"n":128,"mean_edge":518.9869279371736,"first_trades":895051,"misread":262675,"first_trade_misread_rate":0.29347489696117873,"callbacks":1387408,"callback_squared_log_price_error_bps2":245155391.8794986,"callback_rms_log_price_error_bps":13.292866021566669,"mean_retail":546.8069174639777,"mean_arb":-27.819989526804118,"mean_rv":77330.12642708373,"mean_av":24409.588335169257,"rows_sha256":"561d7f53ecd90e1a4c8324038cd33e1f222533acafdef943d27a375d4ee5a0e3"}}},"confirmation":{"seeds":[62128,62255],"vs_base":{"n":128,"mean_diff":-0.7273312670206349,"se":0.040487122242401356,"sum":-93.09840217864127,"sumsq":94.36032890299882,"worst5":[[62199,-1.7484526263434645],[62237,-1.724637871750076],[62198,-1.5826851880965478],[62133,-1.5761886560655967],[62177,-1.5262531662440324]]},"metrics":{"probe":{"n":128,"mean_edge":534.6457567584481,"first_trades":913731,"misread":269313,"first_trade_misread_rate":0.29473991798461474,"callbacks":1437948,"callback_squared_log_price_error_bps2":236283621.79256982,"callback_rms_log_price_error_bps":12.818736461657874,"mean_retail":561.6079344939176,"mean_arb":-26.962177735469453,"mean_rv":80782.35360647544,"mean_av":25276.273569833622,"rows_sha256":"fa54a3580673ee3b5c09d6fe881d9cc7fe360d027363282277451150ec28c349"},"base":{"n":128,"mean_edge":535.3730880254687,"first_trades":907698,"misread":260070,"first_trade_misread_rate":0.28651599981491643,"callbacks":1424868,"callback_squared_log_price_error_bps2":250292746.83837867,"callback_rms_log_price_error_bps":13.253690224251036,"mean_retail":561.9793859305945,"mean_arb":-26.60629790512572,"mean_rv":79925.98836244653,"mean_av":24759.540295940464,"rows_sha256":"4a919b5c585ba9edb480149bb0a1c010b02933718a8794b0be1ee92e3e472452"}}},"holdout":{"seeds":[10000,12047],"candidate_n":2048,"base_n":2048,"lattice_reference_n":2048,"candidate_evaluations_per_seed":1,"base_evaluations_per_seed":1,"chunks_per_arm":8,"seeds_per_chunk":256,"harness":"Unmodified run.py and replica2.py; optional callback-equivalent acceleration of policy only. pair.py called on both full aggregate comparisons. No full-simulator fast runner used."},"holdout_vs_base":{"n":2048,"mean_diff":-0.6680195730342616,"se":0.010654377723072823,"sum":-1368.1040855741678,"sumsq":1389.8074526269356,"worst5":[[10387,-3.0034528036728716],[10475,-2.2780456473446975],[11429,-2.175556434660166],[11628,-2.126169014748598],[11753,-2.087235020475646]]},"holdout_vs_lattice_qm":{"n":2048,"mean_diff":-0.3451276532237704,"se":0.01415442415015115,"sum":-706.8214338022818,"sumsq":1083.8525666233618,"worst5":[[10942,-3.3280795822504388],[10475,-3.002978471509209],[10578,-2.889722166980846],[11651,-2.718799156278351],[11811,-2.5396776677179105]]},"base_vs_lattice_qm":{"n":2048,"mean_diff":0.3228919198104912,"se":0.01069159987528695,"sum":661.282651771886,"sumsq":692.7408990952625,"worst5":[[11487,-1.9115248303883163],[10942,-1.6076924222966227],[10982,-1.4457581244053017],[11380,-1.2365145499310302],[10990,-1.1504964526710637]]},"holdout_diagnostics":{"probe":{"n":2048,"mean_edge":533.4757967209376,"first_trades":14571168,"misread":4330172,"first_trade_misread_rate":0.2971739808366769,"callbacks":22848808,"callback_squared_log_price_error_bps2":3766417378.8932896,"callback_rms_log_price_error_bps":12.83903694011598,"mean_retail":564.2513388749287,"mean_arb":-30.775542153991104,"mean_rv":80231.10659262181,"mean_av":25231.94082893397,"rows_sha256":"7e721b1016627f3c30e8447c4cc879f550d35fd26d04f188d85d0d5d302d2411"},"base":{"n":2048,"mean_edge":534.1438162939719,"first_trades":14472181,"misread":4180058,"first_trade_misread_rate":0.28883400504733875,"callbacks":22636622,"callback_squared_log_price_error_bps2":3990446776.5684752,"callback_rms_log_price_error_bps":13.277152756808208,"mean_retail":564.5685069120709,"mean_arb":-30.42469061809902,"mean_rv":79370.90588207457,"mean_av":24708.59534425337,"rows_sha256":"2d74845c78c1c1c7a215f450b66f3ed75c19975c8e62b6ec93da5773443c009b"},"lattice_qm":{"n":2048,"mean_edge":533.8209243741613,"first_trades":14315685,"misread":2322695,"first_trade_misread_rate":0.162248261260289,"callbacks":22263118,"callback_squared_log_price_error_bps2":4143769057.6884427,"callback_rms_log_price_error_bps":13.64283973652866,"mean_retail":563.8254861326118,"mean_arb":-30.004561758450425,"mean_rv":78272.73176182061,"mean_av":24093.289877718693,"rows_sha256":"02e2e352ea83b848b2f40b4094ee8fafc5869f56aba0f3d759bd2b66323a8a3a"}},"interpretation":{"callback_rms_improvement_vs_base_bps":0.4381158166922283,"mean_retail_edge_diff_vs_base":-0.31716803714220987,"mean_arb_edge_diff_vs_base":-0.3508515358920832,"conclusion":"Obtaining cleaner price estimates did not pay for the probe. Misread rate is worse, and both retail and arb edge deteriorate. Negative result for this specific capped midpoint rule and one-step surrogate, not proof against all information-seeking quotes.","promotion_bar_plus_0_3_met":false},"checks":{"pinned_fetch":{"files":26,"sha256_mismatches":0},"zero_change":{"seeds":[62000,62127],"n":128,"all_row_fields_bit_identical":true,"paired":{"n":128,"mean_diff":0.0,"se":0.0,"sum":0.0,"sumsq":0.0,"worst5":[[62000,0.0],[62001,0.0],[62002,0.0],[62003,0.0],[62004,0.0]]}},"base_fast_callback_parity":{"seeds":[62000,62003],"callbacks":41161,"max_fee_error_bps":0.0,"max_log_price_difference":0.0,"first_and_is_arb_match":true,"before_scoring":true},"probe_fast_callback_parity":{"settings":[0,4],"seeds":[62000,62003],"callbacks":82771,"active_callbacks":41610,"active_probes":6731,"max_fee_error_bps":0.0,"max_log_price_difference":0.0,"first_is_arb_probe_counts_match":true,"before_scoring":true},"blindness":{"seed":62004,"callbacks":10945,"equal_outputs_with_all_hidden_labels_and_parameters_perturbed":true,"needs_flag":false,"callback_parameters":["self","is_buy","ax","ay","t","x","y"],"probes":1736,"eligible":3608},"stdlib_training_replay":{"snapshots":512,"rollout_value_max_abs_difference":0.0,"frozen_trees_equal":true},"stdlib_snapshot_replay":{"seed":62000,"snapshots":4,"all_fields_equal":true},"source_quotes_refetched_and_found":5,"holdout_unique_and_complete_both_arms":true,"stdlib_runtime_dependencies":[],"network_required_for_policy_or_reproduction":false,"shared_data_prerequisite":"Pinned solved pricing tables and reference rows are required inputs, deliberately excluded from the source bundle under the no-tables/no-traces instruction.","stdlib_scoring_replay":{"n":2,"seeds":[62000,62001],"callbacks":19762,"all_row_fields_equal":true}},"gas_estimate":{"measured":false,"solidity_port_present":false,"estimated_incremental_gas_range":[10000,40000],"estimate_basis":"Heuristic only: a maximum 121-cell extra moments scan plus scalar log/exp/sqrt and at most three decision comparisons. Depends on table/filter packing and math libraries; not an EVM benchmark or a gas guarantee.","operations":{"extra_moments_scan_max_cells":121,"moments_support_passes":2,"additional_log_per_eligible":2,"additional_exp_per_eligible":1,"additional_sqrt_per_eligible":1,"tree_comparisons_max":3,"diffusions_added":0,"online_rollouts":0,"online_classifier_changes":0},"persistent_slots_incremental":0,"instrumentation_counters_not_required":2,"full_base_naive_unpacked_scalar_slots":135,"naive_slots_basis":"121 lattice probability cells plus 14 scalar filter/diagnostic fields. No verified 32-slot packing or complete <=245000 gas port is claimed.","full_strategy_gas_estimate":null},"evidence_provenance":{"external_sources":"sources.json; pinned SHA256SUMS verified","measured_numbers":"Local run.py callback-blind replica runs; paired differences computed by evidence.py (included) and checked against pair.py rounded output. Diagnostics are pooled counts, RMS weighted by all callbacks. No imputation, board score conversion, or Rust-engine claim.","worst5_format":"[seed, candidate_edge - reference_edge]","row_fingerprints":"SHA256 of seed-sorted rows serialized with sorted keys and compact JSON separators; traces not bundled."},"code_format":"base64(zlib(USTAR)); deterministic flat source files, no pricing tables, traces, dependencies or caches","code_sha256":"bef576d1a8f43675a1bbdcd9087e49fd0996c75d15062ebb50925351b3b660c9","code_zb64":"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","source_files":["README_probe.md","check_blindness.py","evalpol.py","evidence.py","frozen.py","latt.py","offline.py","pair.py","par.py","probe_base.py","probe_policy.py","probe_rule.py","real.py","replica2.py","reproduce.py","run.py","thrpol.py"]}