{"item":"data-c","credit":"Taylor (gtmcknight): replica2, solved-game research, filters, qm tables; qt1/TwoComp/COMBO/QMS1 artifact authors credited in their preserved headers.","status":"completed_offline_classification","task_findings":["Offline held-out first-trade error: original COMBO pi 0.161723049; full frozen GBDT 0.158719332; paired delta -0.003003717.","Held-out logloss: original 0.356292059; full 0.343705075. Brier: original 0.112555275; full 0.110038148.","These are classification results on unchanged COMBO trajectories, not measured policy gains. No learned probability was inserted into live quoting.","Full model corrects 1352 net errors on 450109 test rows: false arb predictions fall by 1575 while missed arbs rise by 223. Error reduction is modest, not evidence of recovering the true flag.","Pi-only calibration does not significantly improve classification error (its paired 95% error-delta interval includes zero). Richer observable features provide the measured reduction.","Removing explicit callback history raises error by 0.000815358 (0.0815358 percentage points); removing fee/rival features raises it by 0.000295484 (0.0295484 points), with paired seed-bootstrap intervals above zero. The history ablation retains COMBO filter state."],"hashes":{"commit":"4812a6382311624e1506aca131ca35c378f930d9","commit_verified_by":"git rev-parse HEAD after checkout","artifact_sha256":{"qt1":"8a6e303b662e1a5ea9bb47f31acedd3988ef65b3ddb66d074517aa59ad41ca40","TC":"d3c849db2ac31e567b57e52b80b86e8c24c842e5b4125b9d75c78e3cb07115b9","COMBO":"d360bea6c8c982717b7b43a8069a87013169c89ee3e4da25622db4d880b936d4","QMS1":"17d249ab0c33fc44931d9b105cf3f9e3846a560bd518b67310c2a56defacdecc"},"files":{"qmref.py":{"sha256":"1dfc9b7a3c2237ff7a93dd9f0af54b9bdab16f2ce0a7eb00784cb4432598db92","bytes":7739,"source_url":"https://api.imd.fun/artifacts/8a6e303b662e1a5ea9bb47f31acedd3988ef65b3ddb66d074517aa59ad41ca40"},"twocomp.py":{"sha256":"9d3c1833321493fb51bbde4e1b4dabc2f8978bd6d774ee2af8dc8aed13f0f0af","bytes":11202,"source_url":"https://api.imd.fun/artifacts/d3c849db2ac31e567b57e52b80b86e8c24c842e5b4125b9d75c78e3cb07115b9"},"combo.py":{"sha256":"c5b711a0b4fb4783e8879a484bf87a427171f4a11aa5eb5ec3513c24ead1a796","bytes":5929,"source_url":"https://api.imd.fun/artifacts/d360bea6c8c982717b7b43a8069a87013169c89ee3e4da25622db4d880b936d4"},"blob.bin":{"sha256":"3ac7cb8811ecfb6a17f499aa4e9b70dd55f7fb6432b3aec3b904e11a3dece38e","bytes":10167,"source_url":"https://api.imd.fun/artifacts/17d249ab0c33fc44931d9b105cf3f9e3846a560bd518b67310c2a56defacdecc"},"Strategy.sol":{"sha256":"c68887cff61180e12ba2b1ceb333edf76a830f91f63b5d775a8806fb9ec08aca","bytes":42990,"source_url":"https://api.imd.fun/artifacts/8a6e303b662e1a5ea9bb47f31acedd3988ef65b3ddb66d074517aa59ad41ca40"},"replica2.py":{"sha256":"6e37b4e61e1ffa5ffb1e43912ddb8659105989aa986eb48fcd1d7608f8dca685","bytes":7503,"source_url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/4812a6382311624e1506aca131ca35c378f930d9/research/solved/replica2.py"},"gates.py":{"sha256":"603280c1f595ca163ffaf2ba1ebdc7945fdb08dabae67fc53c922d39e1cce604","bytes":792,"source_url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/4812a6382311624e1506aca131ca35c378f930d9/research/solved/gates.py"},"evalpol.py":{"sha256":"6e67667bfed7798bdd3384652a6efc6ba0b1fce653740d0f9af2bcbcb8e6a27c","bytes":4610,"source_url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/4812a6382311624e1506aca131ca35c378f930d9/research/solved/evalpol.py"},"real.py":{"sha256":"1305b5333731c046c3e6736cc2578eff4635c88132a39c22f5780d78aef2c81e","bytes":9586,"source_url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/4812a6382311624e1506aca131ca35c378f930d9/research/solved/real.py"},"latt.py":{"sha256":"925c0fe996b19fd51fb79a85d5e4dba4bc2f5453d0a156cc1e480f65a6a00cc8","bytes":8893,"source_url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/4812a6382311624e1506aca131ca35c378f930d9/research/solved/latt.py"},"pol_mid.json":{"sha256":"efbb5ee102db4d312f64565d9bf7232b6becbe02b624164f744a402bbbf2983b","bytes":780061,"source_url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/4812a6382311624e1506aca131ca35c378f930d9/research/solved/pol_mid.json"},"qm_k1e-5.json":{"sha256":"5ff53e42ca87390e5f65d1caa3f7a480e67a8212fb0670274dfd562b05ba1afd","bytes":7892018,"source_url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/4812a6382311624e1506aca131ca35c378f930d9/research/solved/qm_k1e-5.json"},"dp.py":{"sha256":"ca4e7c378d23c46c7def4716611135c471a831d15234b984fe30d58d9eec5734","bytes":15595,"source_url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/4812a6382311624e1506aca131ca35c378f930d9/research/solved/dp.py"},"qmdp.py":{"sha256":"63b8c35048718c37eb0a5dae9e70849fa3af257c335e4a2e3a625c833f1dc1eb","bytes":8609,"source_url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/4812a6382311624e1506aca131ca35c378f930d9/research/solved/qmdp.py"},"calib_out_2048.json":{"sha256":"fdf24c7aa2b604d770336a5783276da8ed4d50bca9dbcc02e98b4da58f608d26","bytes":2948144,"source_url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/4812a6382311624e1506aca131ca35c378f930d9/research/solved/calib_out_2048.json"}},"frozen_spec_sha256":"151a51ac0d4fe6cb22e94ab754860d12b96a678937329e802439e0f43722daed","source_freeze":{"description":"Sources and spec frozen before test metric evaluation; no model selection on test or report seeds","frozen_spec_sha256":"151a51ac0d4fe6cb22e94ab754860d12b96a678937329e802439e0f43722daed","hashes":{"audit.py":"ec9ab6b2d9faa5d259c4f3c3e304e4a59e33d8cdb421306cf609a7868b0bae37","collect.py":"f86ea5557480dc7cc0171583ddd5f2554375f7fc038172704ec4767b608cdfe3","driver.py":"239fa0f6845b1f0c0f03f7915d7c37f3c697faee2b4a9fe38913584b30cc6eba","score.py":"015e1392ba8778adc1d92724e79c7fbe5a7efddc063593ac891d5e0365aebfe5","summarize.py":"a707175dcd2173126caedf810fe2fa2fab0091ce7de68402e30285202cb5e8c8","trees.js":"ac7658345091caf1e7026ee2b9e73deee760139c2ea39313c2dbd3200b24d08a"}},"model_freeze":{"before_test_metrics_and_remaining_report_runs":true,"models":{"belief_size_timing":{"margin_sha256":"54927d4837814dbacf1ca40222ef5474f563e3d9adbf465d513c30c156a40208","model_sha256":"960050c476e470dbcd87bd3aed3e950708d327e013419d636e0962f7b018b5d1","trees":192},"full":{"margin_sha256":"3c8797c363034eddf52220a3f4c04872595d2e1df94efde959b52b718b1d1889","model_sha256":"7ff0c18d1bc531f445c19beb57f45643c087bdc6034957cd4bc7c1d7c86cd371","trees":192},"no_fees_rival":{"margin_sha256":"8a75c2ea51bfd2fda928cdd9b1e7e390ec97a62132d0a099d6a4f7a8a2f286c8","model_sha256":"806ad185ae2576b8f449305677ed6dc54b4d5cc45a7fcf1c65bec3aa62d18af7","trees":192},"no_history":{"margin_sha256":"0d86d6d3edfdcb5700a0f8aa47df64f22baab3a8ca7896703863aaec565f9479","model_sha256":"079de0e418b8347af016f3dd62341e8504cb08f6a202c968516dd8c1e921dae4","trees":192},"pi_only":{"margin_sha256":"e26a9f428496db0e16b55a16e7beb9026ce1d9acf2ee33b467b09de71fe4f9e6","model_sha256":"f284a965d13a0aa4dea4582eee8491f3a8b2a7807642aebda52c6317c3c2d565","trees":192}},"policy_source_sha256":{"combo.py":"c5b711a0b4fb4783e8879a484bf87a427171f4a11aa5eb5ec3513c24ead1a796","qmref.py":"1dfc9b7a3c2237ff7a93dd9f0af54b9bdab16f2ce0a7eb00784cb4432598db92","replica2.py":"6e37b4e61e1ffa5ffb1e43912ddb8659105989aa986eb48fcd1d7608f8dca685","twocomp.py":"9d3c1833321493fb51bbde4e1b4dabc2f8978bd6d774ee2af8dc8aed13f0f0af"},"quant_json_sha256":"a457d774b8206185e8532f59801579514ea9341c7fa25c23535b7527f02b39a4","utc":"2026-11-10T04:00:10.299400+00:00"},"metrics_by_seed_sha256":"a4d87d4eb3ed5e1ee70ec7b9bea9596badd4729a08b2376189007726c42676d8"},"gates_output":"seed   perfect-info      gaussian+plain    gaussian+qm       lattice+qm\n10000  448.261189  443.455598  443.522608  443.663520\n10001  576.779199  571.463409  571.623551  571.977293\n10002  531.886497  525.168028  525.865859  526.411631\n","gates":{"P":"PASS: all 12 values exactly match README at 6dp, before authored collector/trainer work; gates/replica unchanged","qt1_check":[{"seed":10000,"edge":443.1017411146794},{"seed":10001,"edge":571.454483381923},{"seed":10002,"edge":525.908441161234}],"Q":"PASS: Unchanged COMBO paired to immediate base TwoComp and actual qt1 on complete choose/report ranges. No new live policy; no choose optimization or report selection.","F":"not applicable: no authored contract; no fee parity or gas claims"},"seed_coverage":{"data":{"test":{"arb_labels":246993,"features_sha256":"405c709c448f25d8dac0e008758ef58578613947e6abf74d45fae70c48411ddb","initial_callbacks":64,"labels_sha256":"da9d9130c6145c65bcd89e2f73dcc7e87fae0e2dff45efc1bdc1f3532232b659","retail_labels":203116,"rows":450109,"seed_count":64,"seeds":[72192,72255],"timestamp_zero_rows":32,"per_seed_row_range":[6201,7702]},"train":{"arb_labels":739117,"features_sha256":"1ba45e581ae1e7d0037df5693f6b0d976201c9041b92c4922b6f349efb9e3ea7","initial_callbacks":192,"labels_sha256":"f0ff6facdb1ff2904314f041e97028d9ff9db2700b2eaf5725ae52715a618b0f","retail_labels":608797,"rows":1347914,"seed_count":192,"seeds":[72000,72191],"timestamp_zero_rows":106,"per_seed_row_range":[6269,7695]}},"collection_callbacks":2803113,"collection_first_trades":1798023,"choose":[64000,65023],"report":[10000,12047],"steps_per_run":10000,"config":"replica2.run defaults, cfg=None, qcap=None; fresh policy instance per seed","data_copies":"one materialized feature/label copy per seed; first 3 seeds additionally replayed for invariance/edge checks; seed72000 replayed for exact-pi audit","paired_policies":["COMBO","TwoComp","actual qt1 QMRef"],"no_seed_overlap_between_train_test_choose_report":true},"frozen_spec":{"ablations":["pi_only","belief_size_timing","no_history","no_fees_rival","full"],"bootstrap":{"interval":"2.5%,97.5% percentiles of paired metric differences","replicates":2000,"rng_seed":831947,"unit":"simulation seed"},"credit":"Taylor (gtmcknight), pinned replica, COMBO/TwoComp/QMRef","feature_dtype":"little-endian float64; no row header; one vector per first callback of each timestamp including t=0","features":["pi","logit_pi","jump_sd","abs_jump_sd","prior_sd_bps","ret_sd_bps","jump_bps","ret_jump_sd","noarb_mass","size_law_z","move_bps","signed_move_bps","log_ax","log_ay","buy","dt","silent_steps","time","initial_callback","bid_bps","ask_bps","fee_skew_bps","quote_width_bps","rival_bps","rival_vs_sp_bps","rival_band_low_bps","rival_band_high_bps","synced","prior_count","component_weight","component_gap_bps","component_sd_bps","last_pin_age","last_pin_gap_bps","prev_side","prev_move_bps","prev_dt","prev_step_fills","own_calls","buy_fraction","history_move_mean","history_move_sd","history_dt_mean","history_pi_mean","ew_move_10","ew_move_100","ew_dt_10","ew_dt_100","ew_pi_10","ew_pi_100","ew_buy_10","ew_buy_100"],"groups":{"belief":[0,1,2,3,4,5,6,7,8,29,30,31],"fees_rival":[19,20,21,22,23,24,25,26,27,28],"history":[32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51],"size_side":[9,10,11,12,13,14],"timing":[15,16,17,18]},"history":"all own earlier callbacks; inferred pins only; ew alphas 2/(10+1), 2/(100+1)","labels":"separate uint8, 1=arb, 0=retail; no hidden feature inputs","leakage":"seed never feature; hidden init config, p,s2,j never feature; labels separate; feature schema frozen before collection","model":{"algorithm":"logistic histogram Newton gradient-boosted trees","base_margin":"logit of training prevalence; no COMBO logit offset","bins":64,"depth":4,"l2":10,"learning_rate":0.08,"max_leaf_value":3,"min_leaf_hessian":2,"min_leaf_rows":128,"sampling":"all rows/all features, deterministic feature-index then threshold tie breaks","selection":"fixed spec; no test early stopping or hyperparameter selection","threshold":0.5,"trees":192},"seed_split":{"test":[72192,72255],"train":[72000,72191]},"steps":10000},"classification":{"schema_notes":{"formula_authority":"Bundled collect.py Live.features and Live.update; frozen feature order is frozen_spec.features","rival_band_low_bps":"Stored legacy name: actually (ln(s2h/.997)-m)*10000, the upper fair-price band side","rival_band_high_bps":"Stored legacy name: actually (ln(s2h*.997)-m)*10000, the lower fair-price band side","prev_step_fills":"number of own callbacks in last observed timestamp, not unseen immediately preceding timestamp","ew_suffix":"10 and 100 are callback-span alphas 2/(span+1), not time-step windows","history_pi":"current first-trade pi on dt>=1; fixed nonarb zero for same-timestamp callbacks; all earlier own callbacks contribute","no_history_ablation":"Removes explicit rolling/last-fill history group; retains carried COMBO belief/filter/rival/fee state. Not a memoryless classifier.","raw_features":"No seed, true arb flag, true p, true s2, true j, sigma, lambda, size_mean or unseen fills/pins in feature vector"},"units":"row-pooled metrics; error is a fraction, logloss in natural-log nats, Brier=(p-label)^2; arb=1; threshold strictly >0.5","uncertainty":"2000 paired simulation-seed bootstrap replicates on 64 test seeds; pooled metrics recomputed under each seed sample; 95% percentile intervals, not row bootstrap","results":{"belief_size_timing":{"test":{"brier":0.11118037344883464,"confusion_TN_FP_FN_TP":[161811,41305,30912,216081],"error":0.1604433592751978,"logloss":0.3470349218940494,"n":450109,"positive":246993},"test_minus_full":{"brier":{"bootstrap_sd":5.805856810726392e-05,"ci95":[0.0010295110785894183,0.0012560281071614069],"delta":0.0011422252234733477},"error":{"bootstrap_sd":0.00022457820823403436,"ci95":[0.0013005342071249597,0.002198838258917628],"delta":0.0017240268468304287},"logloss":{"bootstrap_sd":0.0001427384178708113,"ci95":[0.003049225471516188,0.0036128265334274537],"delta":0.0033298465955379775}},"test_minus_original_pi":{"brier":{"bootstrap_sd":5.944473120948642e-05,"ci95":[-0.0014865891241301916,-0.00126020801574668],"delta":-0.0013749019457865719},"error":{"bootstrap_sd":0.00020983412854111847,"ci95":[-0.0017012011994370177,-0.000890344947719013],"delta":-0.0012796900306370207},"logloss":{"bootstrap_sd":0.000532279296945116,"ci95":[-0.010347127958725364,-0.008239303703272294],"delta":-0.009257136744593664}},"train":{"brier":0.11030222370005316,"confusion_TN_FP_FN_TP":[485393,123404,91513,647604],"error":0.15944414851392597,"logloss":0.34410066487484336,"n":1347914,"positive":739117}},"full":{"test":{"brier":0.11003814822536129,"confusion_TN_FP_FN_TP":[162786,40330,31111,215882],"error":0.15871933242836736,"logloss":0.3437050752985114,"n":450109,"positive":246993},"test_minus_original_pi":{"brier":{"bootstrap_sd":8.560271279739848e-05,"ci95":[-0.0026819588446745524,-0.0023456474973926754],"delta":-0.0025171271692599195},"error":{"bootstrap_sd":0.0002841194386265638,"ci95":[-0.0035486978835374255,-0.00244805317359243],"delta":-0.0030037168774674494},"logloss":{"bootstrap_sd":0.0005542294647628711,"ci95":[-0.013688127294753926,-0.011571171327184488],"delta":-0.012586983340131641}},"train":{"brier":0.10899723666510465,"confusion_TN_FP_FN_TP":[488280,120517,91656,647461],"error":0.15740841032884886,"logloss":0.3403532590966686,"n":1347914,"positive":739117}},"no_fees_rival":{"test":{"brier":0.11029355305342103,"confusion_TN_FP_FN_TP":[162757,40359,31215,215778],"error":0.15901481641113596,"logloss":0.34446581542510907,"n":450109,"positive":246993},"test_minus_full":{"brier":{"bootstrap_sd":1.5570686516404968e-05,"ci95":[0.00022564618900653937,0.00028533742267579403],"delta":0.0002554048280597465},"error":{"bootstrap_sd":0.00012071694797912168,"ci95":[6.643398454299346e-05,0.0005314624175946521],"delta":0.00029548398276860177},"logloss":{"bootstrap_sd":4.419384805881359e-05,"ci95":[0.0006782226140853793,0.0008478007051731071],"delta":0.0007607401265976588}},"test_minus_original_pi":{"brier":{"bootstrap_sd":8.257934048353558e-05,"ci95":[-0.002419884443032962,-0.0020972058213843668],"delta":-0.002261722341200173},"error":{"bootstrap_sd":0.0002829982858914537,"ci95":[-0.0032506348859822846,-0.00215687142492257],"delta":-0.0027082328946988476},"logloss":{"bootstrap_sd":0.0005538900177037048,"ci95":[-0.01294662093965728,-0.010798189073434558],"delta":-0.011826243213533982}},"train":{"brier":0.10924583792345698,"confusion_TN_FP_FN_TP":[488217,120580,91950,647167],"error":0.15767326402129514,"logloss":0.341120027370218,"n":1347914,"positive":739117}},"no_history":{"test":{"brier":0.11043160438556494,"confusion_TN_FP_FN_TP":[162462,40654,31154,215839],"error":0.15953469048608226,"logloss":0.3447443254880698,"n":450109,"positive":246993},"test_minus_full":{"brier":{"bootstrap_sd":6.096076685678408e-05,"ci95":[0.0002760552126797341,0.0005138000748816522],"delta":0.00039345616020365626},"error":{"bootstrap_sd":0.00022899109423700438,"ci95":[0.00037031885244991975,0.0012753949198148398],"delta":0.0008153580577149044},"logloss":{"bootstrap_sd":0.00015343009490010823,"ci95":[0.0007521913302466589,0.0013469382612371435],"delta":0.001039250189558416}},"test_minus_original_pi":{"brier":{"bootstrap_sd":7.260980647651636e-05,"ci95":[-0.002258854496576136,-0.001984222885782714],"delta":-0.0021236710090562633},"error":{"bootstrap_sd":0.00023088721370698834,"ci95":[-0.002636911422132505,-0.0017224601748976544],"delta":-0.002188358819752545},"logloss":{"bootstrap_sd":0.0005644406776994366,"ci95":[-0.012706173868111875,-0.010459295769513368],"delta":-0.011547733150573225}},"train":{"brier":0.10953027532424568,"confusion_TN_FP_FN_TP":[487328,121469,92113,647004],"error":0.15845372924385384,"logloss":0.3417557415443806,"n":1347914,"positive":739117}},"original_pi":{"test":{"brier":0.11255527539462121,"confusion_TN_FP_FN_TP":[161211,41905,30888,216105],"error":0.1617230493058348,"logloss":0.35629205863864305,"n":450109,"positive":246993},"train":{"brier":0.11173726592240651,"confusion_TN_FP_FN_TP":[483644,125153,91641,647476],"error":0.16083667058877643,"logloss":0.35338495735216935,"n":1347914,"positive":739117}},"pi_only":{"test":{"brier":0.11251035695017365,"confusion_TN_FP_FN_TP":[162194,40922,31887,215106],"error":0.1617585962511303,"logloss":0.352947576707839,"n":450109,"positive":246993},"test_minus_full":{"brier":{"bootstrap_sd":7.636301301495158e-05,"ci95":[0.0023240207892734786,0.0026210331725284247],"delta":0.0024722087248123664},"error":{"bootstrap_sd":0.00028226287161622513,"ci95":[0.0024886200275927913,0.003609698687593002],"delta":0.0030392638227629276},"logloss":{"bootstrap_sd":0.00024914632629287846,"ci95":[0.008771810196850156,0.009747118863507077],"delta":0.009242501409327586}},"test_minus_original_pi":{"brier":{"bootstrap_sd":3.0524573024724914e-05,"ci95":[-0.00010099309912339843,1.6105418728537496e-05],"delta":-4.491844444755311e-05},"error":{"bootstrap_sd":0.00010424533073545496,"ci95":[-0.00017501273398762472,0.00023190625748670122],"delta":3.5546945295478194e-05},"logloss":{"bootstrap_sd":0.00042235263384840304,"ci95":[-0.004238117838097197,-0.0025451751227002876],"delta":-0.003344481930804055}},"train":{"brier":0.11169154091503147,"confusion_TN_FP_FN_TP":[486638,122159,94556,644561],"error":0.16077806150837518,"logloss":0.350511986665052,"n":1347914,"positive":739117}}},"initial_timestamp_zero_rows":138,"initial_pi_rule":"COMBO has no new-step pi at timestamp 0 (dt=0); use its fixed nonarb decision pi=0. Initial-callback and timestamp features expose this; later initial callbacks with dt>=1 use exact original pi.","model_selection":"All 5 variants fixed before fitting. All 192 trees used, no test early stopping, no test threshold fitting. No report seeds used for classification."},"policy":{"selection":"No new live policy. Only fixed, preexisting COMBO scored; classifications are not substituted into it.","choose":{"COMBO vs TwoComp":{"mean_diff":0.2012123291187956,"n":1024,"se":0.015331104507144632,"sum_d":206.0414250176467,"sum_d2":287.67759401247514,"worst5":[[64101,-1.6432879854905877],[64920,-1.3619684038204696],[64110,-1.2944815700027448],[64133,-1.1774341364948668],[64380,-1.111318779611679]]},"COMBO vs qt1":{"mean_diff":0.6096046465295315,"n":1024,"se":0.01940052280944413,"sum_d":624.2351580462403,"sum_d2":774.8145734774697,"worst5":[[64101,-2.7769830832371554],[64845,-1.1638684799705743],[64374,-1.023433823667915],[64795,-0.9831714544297938],[64555,-0.9461893679592208]]},"TwoComp vs qt1":{"mean_diff":0.4083923174107359,"n":1024,"se":0.01718737463861712,"sum_d":418.19373302859356,"sum_d2":480.24009356140374,"worst5":[[64205,-1.4142493224746886],[64374,-1.3881508787469556],[64101,-1.1336950977465676],[64590,-1.1226070784020976],[64698,-1.0536838335264065]]},"mean_edges":{"COMBO":536.3301001181576,"TwoComp":536.1288877890387,"qt1":535.720495471628},"worker_run_seconds":1140.179034321569},"report":{"COMBO vs TwoComp":{"mean_diff":0.1725880800835669,"n":2048,"se":0.01076265438861963,"sum_d":353.460388011145,"sum_d2":546.6118894605424,"worst5":[[11695,-1.8298040107538327],[10475,-1.6557344060586274],[10519,-1.4469295236498283],[11376,-1.3956585600539029],[11892,-1.361633324228137]]},"COMBO vs qt1":{"mean_diff":0.5973909633856562,"n":2048,"se":0.013387993974623986,"sum_d":1223.456693013824,"sum_d2":1482.2951576559892,"worst5":[[10146,-3.552496929485528],[11062,-2.3348491437070606],[11808,-1.8227381435300458],[10519,-1.782691465320454],[11376,-1.7502956875490554]]},"TwoComp vs qt1":{"mean_diff":0.42480288330208926,"n":2048,"se":0.012201543529995695,"sum_d":869.9963050026788,"sum_d2":993.7102211536977,"worst5":[[10146,-2.7519340066928635],[11701,-2.426585095000405],[11062,-1.680672467095519],[11949,-1.2201722781235844],[10394,-1.2018312208341513]]},"mean_edges":{"COMBO":533.8602216298888,"TwoComp":533.6876335498051,"qt1":533.2628306665031},"worker_run_seconds":1716.9087908212095},"se_definition":"sample_sd(diff)/sqrt(n); sum_d and sum_d2 retain unrounded paired sums","engine":"not run; replica-only item; no contract/submission"},"checks":{"failed_checks":[],"not_executed":["Rust engine","Solidity compilation","contract fee parity","gas measurement; replica-only item"],"reference_hash_verification":{"published_extracted_hashes":"PASS, 5/5","unchanged_upstream_files":"PASS, 10/10"},"hidden_input_invariance":{"seeds":[72000,72001,72002],"callbacks":30081,"features_and_fees":"exactly equal on fixed callbacks under altered arb,p,s2,j and hidden initialization parameters; post-callback COMBO state dictionaries equal","direct_COMBO_edges":"exactly equal for each of first 3 collection seeds"},"python_prediction_driver":{"rows":100,"seed":72000,"max_abs_probability_diff":0.0,"python_driver_vs_JS_training_margin":"PASS"},"original_pi":{"seed":72000,"callbacks":10080,"original_pi_exact_matches":6756,"default_steps":10000,"edge":440.9608459529789},"trainer_fixture":{"balanced_separable_fixture":{"n":1024,"train_rows":768,"trees":32,"errors":0,"logloss":0.060863646288277146,"checkpoint_resume_byte_identical":true,"model_sha256":"176c65741b59c3e226162ca65f05d8eae3f30f9619793c419fdceef051e56fd7"}},"raw_feature_vs_binned_inference":{"tested":20581,"max_abs_margin_diff":0},"dependencies":"No packages installed. Python stdlib and preinstalled Node only; no network needed for bundled collection/training/scoring.","unmodified_references":"byte hashes of all preserved refs above; replica2 and gates unchanged"},"runtime":{"collection_wall_seconds":41.37361898692325,"collection_worker_seconds":160.9166907039471,"paired_worker_seconds":2857.0878251427785,"finished_utc":"2026-11-10T04:08:03.554575+00:00","elapsed_wall_seconds_since_initial_README_download":1791.2658479213715,"maximum_driver_child_seconds":470,"training":"checkpointed chunks of 32 trees (pi_only up to 128); elapsed chunk logs and summaries hashed in run_log_checksums; no timing-based experiments"},"limitations":["Classification uses unweighted logistic loss/error, not downstream edge-weighted mistakes; injecting probabilities into a policy can change subsequent observations and outcomes.","Classification evidence only: deployment, end-to-end learned policy gain, fee parity, gas and leaderboard performance are unmeasured.","Fixed-callback invariance cannot prove complete absence of every possible implementation leak; audited inputs/state and explicit feature signature constrain this risk.","One 256-seed data copy and 64 test clusters; seed-bootstrap uncertainty is conditional on this policy, replica and feature/model specification.","Test bootstrap conditions on fitted weights; training-set/model-fitting uncertainty is not included. The custom deterministic histogram GBDT is not an exhaustive classifier search. Quantiles are trained on a deterministic sample of training rows only.","Full model weights are regenerated by the bundled deterministic trainer, not embedded; model JSON and margin SHA256 values identify expected outputs.","The offline bundle contains inputs for collection, fitting and paired scoring. Repeating the separate P gates needs the pinned upstream solved tables (too large for this 64KB result).","Replica levels do not convert to leaderboard levels. No engine seeds used."],"reproducibility":{"encoding":"base64(zlib(UTF-8 JSON mapping relative paths to text; blob.bin maps to {base64:...}))","bundle_zlib_sha256":"07ac8b2f9d302f40ff0b94e2c673301acf218bc9c49d2b830e65453857100aa1","bundle_zlib_base64":"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","model_spec_and_expected_outputs":{"hash_reference":"hashes.model_freeze.models","spec_reference":"frozen_spec.model","regeneration":"driver.py train NAME STOP; SHA256 must match model_freeze model and margin files"},"unpack_python":"import json,zlib,base64,pathlib; r=json.load(open('artifacts/goal/batches/e02-observable-data/items/data-c/result.json')); d=json.loads(zlib.decompress(base64.b64decode(r['reproducibility']['bundle_zlib_base64']))); root=pathlib.Path('test/scratch/data-c-repro'); [( (root/k).parent.mkdir(parents=True,exist_ok=True),(root/k).write_bytes(base64.b64decode(v['base64']) if isinstance(v,dict) else v.encode()) ) for k,v in d.items()]","versions":{"python":"3.13.5","node":"v24.21.0","OS":"Linux x86_64","packages":"none"},"command_notes":"Initial prep/training were invoked as node trees.js directly; driver.py executes the same program with a 470s child timeout. All replica jobs ran python3 -I with explicit local imports.","commands":["git clone https://github.com/gtmcknight/amm-sim test/scratch/data-c/upstream; git -C test/scratch/data-c/upstream checkout 4812a6382311624e1506aca131ca35c378f930d9; git -C test/scratch/data-c/upstream rev-parse HEAD","python3 -I test/scratch/data-c/upstream/research/solved/gates.py","Fetch four artifact URLs listed in hashes; assert SHA256 equals URL suffix; extract fenced qmref.py, twocomp.py, combo.py, Strategy.sol with trailing newlines; decode QMS1 base64; verify published hashes in hashes.files.","cd test/scratch/data-c-repro (after running unpack_python with python3 -I)","python3 -I driver.py collect freeze (must match frozen_spec_sha256)","python3 -I driver.py collect 72000 72004; python3 -I driver.py collect 72004 72256 (observed collection chunks; 4 workers)","python3 -I driver.py prep","For each name in frozen_spec.ablations, run python3 -I driver.py train NAME STOP for STOP=32,64,96,128,160,192; each command bounded to 470s. Observed pi_only checkpoints=64,192.","Check model JSON hashes against model_spec_and_expected_outputs before evaluating test. Record frozen model hashes.","python3 -I driver.py evaluate; python3 -I driver.py verify; python3 -I driver.py metrics","python3 -I audit.py","python3 -I test/scratch/data-c/test_trainer.py (original fixture location; bundled trainer_fixture_test.py also works standalone)","node --check trees.js; Python ast.parse syntax checks for authored scripts; final JSON parse/byte/path checks","For choose chunks [64000,64256),[64256,64512),[64512,64768),[64768,65024), run python3 -I driver.py score START STOP.","For report chunks [10000,10256),[10256,10512),[10512,10768),[10768,11280),[11280,11792),[11792,12048), run python3 -I driver.py score START STOP. The required 3-seed qt1 checks are cached, not rerun during report.","python3 -I driver.py paired","python3 -I driver.py predict < feature_vectors.jsonl (after fitting; one 52-number feature vector per line)","Data regeneration SHA256 uses concatenated per-seed little-endian float64 feature vectors and separate uint8 labels in ascending seed order; exact train/test checksums in seed_coverage.data."]},"run_log_checksums":{"log_count":43,"sorted_filename_hash_manifest_sha256":"fe178dc5bfd02ccb6e0d495b5019996b7994cc90ac2ddf9b9b57f69343d43f7d"}}