{"family":"compact analytic displacement/uncertainty; clean-room restart","status":"completed replica structural screening; not qualification","hypothesis":"A moment-matched ambiguous-arb posterior can recover trade-implied fair price; smooth displaced-side fee relief makes corrective arbitrage cheap while the opposite side remains about 35 bps from the estimated price.","source_urls_sha256":[{"url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/main/research/filter-brief.md","sha256":"0ef2200226b68aa58de6aad268442ba09ada710c54b3fb0ec35c77b491f78a4b"},{"url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/main/research/replica.py","sha256":"d079712fa1f4850ad8337e656e7ee8dbdde806cd4762d597fae50f547ef9c7c4"},{"url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/main/research/oracle_table.json","sha256":"3e9d1fc8874c6d796d4c26948d486b587f2a6b5610b20fba99923a60954cf4b5"},{"url":"https://raw.githubusercontent.com/gtmcknight/amm-sim/main/research/setup.sh","sha256":"dc7fa6a1e59f5b043d3954cf68d538318b7affaf5c0db3ca69a1822680f300d4"},{"url":"https://raw.githubusercontent.com/benedictbrady/amm-challenge/6b6921f8d463314aa7c96234e1be25a25245f704/amm_sim_rs/src/lib.rs","sha256":"9fa375f51a8e429933795654112b0d6c59764f64f5783784dfa8969f9e1e074c"},{"url":"https://raw.githubusercontent.com/benedictbrady/amm-challenge/6b6921f8d463314aa7c96234e1be25a25245f704/amm_competition/competition/config.py","sha256":"e4b13d9f271fe27e184157904e6e9f5079db7c49ec5c0e1a31b52daeea2c1f9c"},{"url":"https://raw.githubusercontent.com/benedictbrady/amm-challenge/6b6921f8d463314aa7c96234e1be25a25245f704/amm_competition/competition/match.py","sha256":"78915a1aa36341e61c29f333fb18b31a319f70ce3e8bc2f737ba8b003aea5c17"}],"model":{"parameters":{"base_bps":35,"corrective_rebate_bps":7,"transition_knee_bps":36,"transition_width_bps":11,"opposite_offset_slope":0.28,"diffusion_multiplier":1,"retail_impact_scale_bps":20,"retail_first_intensity":0.7,"router_precision_multiplier":0.35,"uncertainty_width_multiplier":0,"student_t3_arb_likelihood":0},"input_boundary":"initialize(x,y); after_swap(sell_x,amount_x,amount_y,timestamp,post_x,post_y). No fair price, rival reserves, arb labels, or sampled run parameters. Amounts are accepted; the filter uses the equivalent post-reserve displacement.","state":"m=log-price in bps, v=variance in bps^2, previous log spot, last timestamp, previous bid/ask fees, q=innovation variance; arb probability and callback count are diagnostic fields.","filter":"First timestamp trade mixes an arb-implied pin with a truncated-normal retail hypothesis. Repeated timestamps force arb probability zero. Silent gaps use an O(1) moment approximation. Retail likelihood and weak router observation are heuristic.","quote":"Logistic displacement offsets with -7bps corrective-side limit; exact formula in model.py. Fees clipped to [0,.1].","all_constants":"All numeric constants, floors, initialization and updates are literal in the embedded source. No binary weights or downloads."},"protocol":{"engine":"supplied pure-Python replica, unmodified SHA-256 recorded; each arm alone against its fixed-30-bps rival","steps_per_seed":10000,"training_seeds":[20000,20127],"evaluation_seeds":[10000,10255],"ablation_seeds":[10000,10063],"forbidden_confirmation_seeds_used":false,"configuration":"K=1e6; initial x=100,y=10000,p=100; cfg omitted. random.Random(10000000+seed) draws size_mean U[19,21], lambda U[.6,1], sigma U[.000882,.001008], in that order. Price RNG=random.Random(seed); retail RNG=random.Random(seed+1); GBM drift=-sigma^2/2, dt=1; retail lognormal sigma=1.2, buy probability=.5. Full exact simulator code included.","fit":"12 predeclared structural candidates on all 128 training seeds, then 8 diffusion/tail combinations around the best first-stage candidate, also on all 128. Select highest mean edge; freeze before any held-out evaluation. No evaluation-dependent tuning.","oracle":"Exact supplied 17 ask +17 bid knots, linear interpolation/clamped displacement at [-8,8]*30bps, plus .1463 true arb-label bump; true p used only in this offline arm. All arms initialize at 35bps. Fixed arm stays at 35bps.","estimated_table":"Extra diagnostic: same table driven by posterior mean, with the observable soft arb probability replacing the true-label bump. Its returned fees are stored back in the estimator. No retuning.","pairing":"Same seed and random streams/configuration across arms; score is trade edge in Y, not submission minus rival edge. Oracle deficit is oracle edge minus analytic edge. SE=sample SD of per-seed differences/sqrt(n); 95% CI=mean +/-1.96*SE.","rms":"sqrt(total squared log-price errors in bps / total own afterSwap callbacks); computed after updating the model. Event-weighted and endogenous to the arm, not time-weighted or a pre-trade prediction error.","real_engine":"Setup/StarterStrategy compilation validation passed. No real-engine score run: Rust API requires EVM bytecode; this item delivers Python. No CH calibration, Solidity gas measurement, RS2 comparison, leaderboard entry, or 4000-seed qualification claimed."},"metrics":{"analytic":{"n":256,"mean":524.3579562342918,"se":5.763635475047293,"ci95":[513.0612307031992,535.6546817653845],"retail_edge":549.7308810001758,"arb_edge":-25.372924765884015,"retail_volume_y":72408.61246468822,"arb_volume_y":19570.223827214202,"rms_bps_after_own_callbacks":17.02475566379311,"rms_observations":2524814},"oracle":{"n":256,"mean":535.216487745992,"se":5.834284086752761,"ci95":[523.7812909359566,546.6516845560275],"retail_edge":566.7133336082994,"arb_edge":-31.496845862307328,"retail_volume_y":76274.98573644714,"arb_volume_y":23965.729903296793,"rms_bps_after_own_callbacks":null,"rms_observations":0},"fixed":{"n":256,"mean":349.2464435291094,"se":3.3514472898394065,"ci95":[342.67760684102416,355.81528021719464],"retail_edge":564.7301763259644,"arb_edge":-215.48373279685492,"retail_volume_y":70558.56404947276,"arb_volume_y":35413.881632622804,"rms_bps_after_own_callbacks":null,"rms_observations":0},"oracle_deficit":{"n":256,"mean":10.858531511700285,"se":0.18121482729713945,"ci95":[10.503350450197892,11.213712573202677]},"gain_over_fixed35":{"n":256,"mean":175.1115127051824,"se":2.9946185178919547,"ci95":[169.24206041011416,180.98096500025065]}},"per_seed":{"columns":["seed","analytic_retail_edge","analytic_arb_edge","analytic_retail_volume_y","analytic_arb_volume_y","oracle_retail_edge","oracle_arb_edge","oracle_retail_volume_y","oracle_arb_volume_y","fixed_retail_edge","fixed_arb_edge","fixed_retail_volume_y","fixed_arb_volume_y"],"rows":[[10000,456.008,-15.608,60247,17944,465.427,-19.744,62676,20750,457.475,-176.963,56637,29944],[10001,580.242,-18.486,84825,20701,598.35,-25.765,88626,26003,598.519,-193.675,81701,39289],[10002,537.997,-21.01,65821,19931,555.45,-27.493,69313,23992,554.906,-232.349,64429,34624],[10003,581.911,-18.249,78705,20871,599.415,-25.769,82803,25936,600.633,-221.004,76714,38056],[10004,530.144,-17.436,69600,19025,547.561,-22.982,73739,23809,541.816,-199.388,68425,34492],[10005,531.978,-17.523,68986,18969,550.975,-23.254,73537,23415,553.665,-211.731,68161,35378],[10006,495.768,-16.235,72120,18786,510.142,-22.512,75295,22988,505.325,-162.727,69549,33721],[10007,536.352,-17.444,73778,19240,551.177,-23.967,77242,23735,546.387,-193.997,71613,34766],[10008,670.743,-20.416,91413,21643,692.069,-27.959,96183,27361,685.869,-248.992,88943,42955],[10009,658.568,-19.246,86441,20930,681.555,-27.833,92582,27195,686.881,-255.421,86463,42947],[10010,484.55,-17.415,72036,19254,501.593,-22.039,76559,22846,495.196,-152.3,70013,34185],[10011,379.684,-14.037,56815,15969,388.935,-17.412,58837,18499,383.899,-117.735,54557,27122],[10012,464.75,-17.91,71530,19264,481.25,-22.238,75283,22752,477.781,-145.206,68511,33384],[10013,409.249,-13.907,55328,16104,420.484,-17.642,58204,18876,419.364,-154.088,53199,27322],[10014,715.078,-25.107,83032,21009,735.239,-32.389,87673,26631,734.988,-328.101,81111,40179],[10015,635.392,-20.967,87518,21648,660.937,-28.588,93503,27734,651.996,-235.547,85750,41231],[10016,613.906,-61.008,68060,19494,635.452,-66.12,71951,23389,633.912,-302.56,66433,33853],[10017,504.694,-17.51,69297,19419,518.835,-24.331,73103,23875,518.084,-184.118,67617,35188],[10018,491.777,-17.418,67194,19341,504.76,-22.588,70468,23390,501.311,-175.59,64985,33841],[10019,466.792,-14.956,59419,17496,474.959,-19.487,62125,20407,469.755,-187.508,57063,29555],[10020,504.342,-16.334,64847,18762,521.044,-22.895,68459,23206,513.342,-197.563,63343,32655],[10021,419.8,-14.435,55152,16946,431.315,-17.904,57695,19503,427.609,-170.917,52305,26874],[10022,451.416,-15.826,59629,17696,461.869,-19.816,61812,20259,456.645,-175.18,57126,29516],[10023,640.698,-20.44,83339,21158,664.299,-29.001,87929,26226,666.651,-262.755,81262,39414],[10024,487.061,-17.698,69622,19581,506.965,-23.288,73872,23743,497.405,-167.357,67353,33705],[10025,453.702,-15.78,59113,17369,468.134,-20.468,62443,20480,463.897,-183.386,57560,28940],[10026,352.41,-13.659,51938,15959,364.628,-17.24,54624,18425,356.823,-114.115,49668,25399],[10027,559.044,-17.645,70452,19487,574.393,-24.693,73912,24197,575.528,-228.393,68490,36129],[10028,540.401,-20.424,64844,18311,554.833,-27.06,68636,23007,557.595,-233.901,64249,34104],[10029,450.521,-15.71,70135,18218,464.817,-20.234,72760,21296,452.576,-130.366,66710,30476],[10030,600.856,-19.864,85121,21569,625.557,-27.738,89760,27047,616.298,-210.45,83322,40268],[10031,349.421,-16.007,55568,17952,361.741,-18.18,58012,19671,350.206,-100.149,52775,26386],[10032,646.109,-19.521,82582,20510,664.712,-26.775,86665,26072,667.13,-263.189,81578,40803],[10033,545.46,-19.211,78499,20708,561.374,-24.91,82155,25100,555.543,-175.169,77059,37553],[10034,570.924,-18.513,80499,20137,590.855,-24.319,84708,24619,585.446,-201.492,78120,36310],[10035,645.78,-21.295,90284,22173,664.602,-28.452,95096,27580,668.89,-237.637,87641,42782],[10036,554.977,-18.497,78792,20046,571.127,-24.51,82296,24462,568.196,-195.256,76377,36564],[10037,453.294,-16.461,64245,18416,466.127,-20.941,67464,21411,463.83,-157.497,61552,31753],[10038,626.064,-18.986,86572,20906,647.592,-26.987,92527,26775,646.332,-220.016,86096,41297],[10039,606.86,-22.288,75255,20072,619.965,-29.379,79367,24324,616.897,-253.767,72914,36594],[10040,616.654,-18.228,87065,20516,636.925,-26.947,91584,27150,642.561,-217.22,85106,41457],[10041,632.749,-17.572,80731,19954,657.778,-25.506,86754,25890,659.3,-249.493,81447,39985],[10042,719.944,-19.365,88962,21647,740.969,-29.76,94008,28685,745.786,-303.002,88308,44462],[10043,647.27,-19.813,84725,20596,666.799,-27.541,89318,26457,663.684,-253.104,82396,40297],[10044,428.676,-15.211,59039,17539,441.488,-19.598,61837,20568,439.845,-155.223,57081,29692],[10045,393.305,-16.639,59146,18520,408.241,-20.144,61898,21025,402.12,-125.071,56342,28964],[10046,442.773,-18.757,66903,20177,457.449,-23.255,69989,23412,451.571,-140.358,63816,32888],[10047,604.052,-19.67,86795,21529,627.58,-27.95,92298,27513,627.531,-201.354,86104,42191],[10048,508.524,-17.283,68618,19177,519.671,-22.462,71937,22579,516.613,-190.604,66398,33341],[10049,391.429,-13.175,55427,16019,403.299,-17.384,57914,18882,398.623,-135.528,53413,26904],[10050,378.561,-14.965,55402,17460,397.257,-18.706,58404,19888,388.098,-128.059,53566,26813],[10051,428.336,-14.962,57683,17402,444.659,-19.933,61020,20861,438.147,-156.849,56648,29503],[10052,440.064,-16.829,67216,19151,452.518,-20.976,69599,22038,444.335,-141.07,63198,30494],[10053,545.865,-17.648,70240,18860,561.555,-23.538,74075,23256,562.694,-226.144,67892,34651],[10054,455.918,-19.366,61298,17851,468.329,-24.006,64443,21140,466.446,-176.372,59227,30041],[10055,727.108,-21.211,91120,23087,741.362,-30.619,96634,29556,744.998,-292.416,89944,44630],[10056,591.97,-17.33,80275,19745,608.56,-25.096,84401,25291,614.225,-216.593,79321,39597],[10057,515.064,-15.274,66382,17900,532.812,-21.981,69677,22302,530.33,-208.609,64241,33169],[10058,449.247,-14.734,56616,16790,459.603,-19.531,59355,20169,456.697,-187.641,54009,28579],[10059,517.34,-19.157,69072,20245,534.176,-24.742,72848,24404,529.954,-197.123,67405,35008],[10060,2077.481,-1161.047,87766,25385,2097.785,-1169.007,92365,30249,2088.804,-1669.445,85905,43947],[10061,569.726,-16.698,79451,19270,587.03,-24.134,83912,24669,580.403,-201.463,77150,37136],[10062,610.955,-18.128,86296,20483,636.673,-25.699,90987,26017,633.238,-217.121,84357,39871],[10063,434.765,-14.547,61886,16924,448.657,-19.324,64900,20338,446.048,-154.797,59099,29615],[10064,677.267,-22.255,90361,21240,691.764,-29.652,95375,27644,695.304,-252.709,89153,43780],[10065,594.766,-18.123,79144,19970,611.992,-25.647,83054,25086,618.338,-236.157,76601,38050],[10066,493.003,-16.932,70681,18862,508.708,-22.505,74210,22714,507.354,-172.042,68484,34123],[10067,646.406,-20.451,76336,20261,662.203,-27.19,80613,25592,667.445,-281.96,75537,39160],[10068,511.082,-16.677,58021,18618,527.022,-22.058,61435,22259,525.851,-240.24,56364,31392],[10069,627.307,-18.717,77021,19425,649.626,-26.456,81739,25254,646.027,-266.884,76169,37837],[10070,422.552,-16.254,56377,18510,434.484,-20.845,59110,21204,430.378,-161.079,54363,29412],[10071,444.004,-15.081,59622,16882,457.175,-19.859,63530,20212,452.772,-168.965,58035,29920],[10072,632.617,-38.222,74065,18862,651.224,-45.34,78127,23887,645.883,-288.891,72267,34999],[10073,572.994,-17.154,77251,20007,592.591,-24.863,81978,25489,589.989,-211.168,76104,37560],[10074,537.417,-17.259,70937,18808,556.803,-23.718,74945,23395,555.819,-214.504,69477,34812],[10075,570.988,-18.53,76801,20332,587.224,-24.796,80848,24748,585.585,-218.207,74004,37200],[10076,540.347,-19.843,75477,20317,555.446,-24.585,79544,24536,561.122,-198.177,73002,36429],[10077,436.178,-15.263,64651,17525,450.093,-19.748,67501,20810,444.631,-143.206,61995,29781],[10078,631.391,-20.16,86280,20596,649.53,-27.451,90459,25916,656.086,-239.167,83678,40338],[10079,656.57,-20.527,84966,21364,679.757,-29.15,89866,27639,682.59,-264.613,83966,42674],[10080,496.544,-19.643,66815,20195,511.782,-23.71,70169,23522,510.361,-190.906,65200,33656],[10081,576.938,-18.993,72700,19762,593.496,-26.082,76711,24566,596.221,-236.546,71278,36424],[10082,576.371,-17.399,72803,18923,591.764,-25.017,76932,24102,592.928,-234.015,72028,36997],[10083,470.924,-16.879,70184,18995,489.645,-22.435,73591,23243,485.595,-147.965,68428,33351],[10084,483.844,-15.596,65944,17817,497.381,-21.175,69539,21897,499.47,-181.291,64307,32591],[10085,460.137,-15.156,60441,17727,476.495,-20.78,63971,21521,472.971,-181.248,58767,30378],[10086,580.723,-19.51,80882,20934,599.911,-26.071,84818,25517,599.133,-209.94,78969,39253],[10087,576.313,-17.759,83400,20267,596.537,-24.567,87790,24983,591.955,-197.171,80838,37374],[10088,451.214,-17.466,64328,19057,464.665,-22.186,67514,22278,460.213,-154.779,61918,32009],[10089,417.918,-15.914,57590,17719,435.164,-20.261,61437,20855,428.917,-151.659,55953,29086],[10090,598.105,-20.624,84540,21826,615.694,-28.01,88385,26948,618.005,-218.252,81230,39945],[10091,489.302,-18.163,64138,19187,510.75,-23.893,68349,22956,507.573,-198.652,63126,32608],[10092,563.199,-16.512,71998,19171,581.482,-23.743,75795,23933,582.427,-231.895,70186,36026],[10093,675.9,-26.71,88759,21253,699.154,-35.758,93751,27633,701.704,-271.95,86986,42762],[10094,487.661,-18.283,65770,19979,504.362,-23.729,69560,23973,502.803,-181.651,64522,33409],[10095,606.203,-19.891,81603,21130,624.982,-27.036,86225,26286,622.367,-226.91,79814,40018],[10096,498.0,-15.095,64085,17810,513.588,-21.501,67389,22189,516.608,-198.092,62969,32690],[10097,519.212,-17.142,69454,17933,534.439,-21.656,73627,21850,534.865,-198.794,67948,33496],[10098,512.686,-17.414,74026,19699,525.881,-22.623,77057,23409,520.043,-169.738,70853,34214],[10099,474.769,-15.337,68918,18075,491.428,-20.971,72537,22151,487.117,-152.716,67213,32606],[10100,458.973,-21.424,59746,18226,469.166,-25.058,62528,21210,466.055,-182.235,57357,30818],[10101,545.715,-16.273,72910,18600,560.533,-23.579,76583,23933,557.756,-205.455,71017,34971],[10102,748.732,-24.338,93611,22418,771.585,-34.365,98773,29531,777.829,-311.208,93075,45918],[10103,749.063,-20.511,88402,21178,770.885,-29.34,93945,27808,778.216,-334.887,88092,44488],[10104,712.385,-20.537,92263,22260,733.446,-30.294,97837,29156,734.992,-274.784,91809,45363],[10105,480.243,-16.652,60993,19009,493.689,-21.44,63972,22505,486.307,-192.678,58569,31030],[10106,442.068,-17.07,57310,17213,450.277,-20.728,59645,19830,449.031,-176.783,55022,29145],[10107,496.771,-16.943,68995,19173,515.421,-23.452,72704,23672,515.565,-180.863,67025,34119],[10108,543.157,-19.956,72482,20296,561.94,-26.462,76029,24763,556.684,-214.946,69724,34820],[10109,763.942,-37.968,91391,21893,785.205,-46.374,97074,28432,788.977,-342.679,89896,44484],[10110,582.46,-19.299,86395,20939,598.906,-25.944,89949,25753,600.518,-195.226,82738,39131],[10111,410.959,-16.678,58557,18721,427.785,-21.485,61770,21909,424.4,-139.913,56925,30752],[10112,473.623,-17.593,62805,18720,486.235,-22.166,65699,22299,480.763,-176.705,61205,31765],[10113,439.292,-15.447,59544,17038,456.17,-19.636,62787,20343,452.662,-167.604,57689,29278],[10114,542.497,-17.428,78389,19779,558.955,-23.872,82134,24579,560.107,-184.686,75976,36594],[10115,668.269,-19.825,88651,21769,692.397,-29.244,94158,28539,694.253,-253.605,88071,43556],[10116,534.457,-16.628,72333,18896,549.635,-22.772,75363,23337,542.542,-195.039,70191,34804],[10117,420.567,-15.378,59120,17972,433.018,-19.967,62164,20894,430.552,-146.165,57638,30125],[10118,428.755,-16.475,58997,19006,445.835,-21.305,62259,21920,444.199,-160.56,57306,30519],[10119,423.605,-16.673,60538,18665,437.949,-20.998,63497,21587,432.349,-146.186,58094,30232],[10120,557.291,-19.085,79699,20762,579.01,-25.571,84471,25697,577.282,-191.084,78289,38282],[10121,544.746,-19.589,77010,20910,561.799,-25.11,81215,25202,558.454,-186.881,74402,36444],[10122,597.623,-19.896,83403,21206,614.965,-27.026,87701,25884,616.661,-214.352,81627,39829],[10123,510.164,-18.244,68743,20399,530.572,-23.856,72650,24285,526.748,-196.311,66812,34876],[10124,537.77,-18.806,71012,20208,558.501,-24.662,75113,24610,555.165,-213.838,69660,35678],[10125,469.462,-18.163,69554,20145,477.65,-22.333,72282,23039,468.071,-148.16,65955,33042],[10126,465.884,-16.133,62922,17102,481.208,-20.514,66048,20326,475.157,-179.34,60349,29772],[10127,501.616,-17.908,69533,19636,518.802,-23.043,73636,23522,518.513,-185.793,67902,33734],[10128,752.568,-29.123,93059,22426,769.386,-36.864,98182,29572,769.682,-306.953,92522,46165],[10129,540.846,-17.907,75477,19819,559.903,-24.6,78740,24589,560.604,-199.38,73283,36433],[10130,469.61,-16.751,65064,18362,483.334,-21.616,69078,22131,482.091,-167.457,63741,31856],[10131,506.126,-17.476,64847,18739,522.592,-22.908,68306,22362,516.361,-202.32,63141,32356],[10132,511.231,-15.599,63196,18053,525.562,-21.96,66053,21957,525.13,-214.831,61114,32246],[10133,632.368,-17.369,81376,20125,657.091,-25.593,86473,25991,657.286,-257.781,80413,38826],[10134,546.352,-19.295,73696,20681,562.463,-26.231,78212,25530,556.971,-199.992,72211,37287],[10135,469.664,-18.026,67052,19616,482.073,-22.529,70066,22997,480.821,-158.492,64990,33055],[10136,626.58,-19.832,86826,20872,646.973,-26.637,90949,25917,650.188,-236.136,83991,40729],[10137,633.275,-18.419,84129,20804,648.675,-25.977,87950,26091,651.264,-247.186,81076,40625],[10138,509.074,-17.905,65360,19262,524.732,-23.369,68392,23183,520.778,-209.642,62751,32779],[10139,473.76,-16.42,65877,18059,489.015,-21.453,69184,22078,485.72,-171.193,64019,32143],[10140,561.295,-16.828,76514,19424,576.402,-23.546,79959,24331,580.819,-209.688,74324,36886],[10141,598.383,-26.074,75950,19375,613.557,-32.903,79716,24308,610.049,-240.953,73905,36135],[10142,635.188,-20.896,88463,22371,657.0,-28.751,93782,28066,661.086,-229.798,87232,42934],[10143,480.268,-15.828,63399,17973,491.998,-20.545,66033,21128,491.661,-187.016,61053,31620],[10144,447.502,-16.057,63805,17673,462.012,-21.198,67258,21702,456.693,-148.815,61992,30818],[10145,487.726,-15.82,64876,18476,503.595,-21.995,68117,22496,501.0,-188.066,62600,31886],[10146,1003.32,-256.9,82069,22717,1030.387,-264.603,87524,28423,1032.752,-621.953,81982,42204],[10147,474.483,-16.234,65993,18651,491.652,-22.064,69300,22773,489.72,-168.28,64464,33444],[10148,702.414,-20.207,85415,21084,720.957,-28.612,90018,27081,722.408,-295.339,84337,41863],[10149,533.003,-22.851,73490,21663,552.5,-28.871,77920,25708,547.107,-197.973,71824,36483],[10150,593.812,-18.405,79945,20344,615.023,-25.775,84704,25602,608.553,-223.321,78329,37716],[10151,478.256,-15.998,66437,18221,495.791,-21.236,69775,22187,491.964,-176.974,64018,31869],[10152,426.458,-15.546,58427,18222,438.221,-20.109,61647,21430,427.633,-147.884,56190,29899],[10153,558.83,-16.916,77919,19364,579.662,-23.726,82124,24365,575.614,-205.467,75119,36378],[10154,494.404,-15.536,69224,18116,511.993,-21.052,72437,22046,509.217,-182.53,66485,32034],[10155,481.231,-15.456,66186,17535,499.14,-20.57,69513,21127,493.989,-178.929,63785,30778],[10156,553.22,-25.738,66727,19161,565.416,-31.309,69940,22959,561.713,-243.943,64285,32853],[10157,540.209,-16.767,72728,18131,559.428,-21.442,76479,22667,551.411,-201.322,70752,33870],[10158,447.813,-16.711,59535,18787,459.489,-21.626,62653,22146,461.132,-174.092,58067,31630],[10159,447.457,-16.935,61806,19050,461.217,-21.824,64757,22390,456.423,-160.041,59615,31392],[10160,429.919,-17.1,59537,18835,443.91,-21.299,62739,22033,439.953,-155.178,57643,30425],[10161,466.634,-16.997,63229,18858,481.192,-22.569,66903,23046,479.726,-173.398,61635,33101],[10162,725.256,-19.022,94362,20966,750.711,-28.991,99413,28287,759.532,-295.565,93297,45870],[10163,676.474,-20.145,91771,21869,698.116,-28.984,96491,28434,698.547,-255.827,89466,43895],[10164,777.534,-106.758,80490,22057,797.898,-113.654,84637,27140,797.009,-407.587,78551,40241],[10165,448.782,-16.344,65857,17990,463.383,-21.592,69230,21618,458.186,-144.229,64096,31749],[10166,428.171,-15.343,57293,17220,439.552,-20.016,60229,20327,436.834,-160.833,55485,29812],[10167,504.225,-19.088,64871,18579,518.319,-24.076,68553,22548,515.282,-201.486,63012,32226],[10168,725.129,-111.265,68018,21294,743.472,-116.231,72373,25180,738.729,-409.021,66401,35371],[10169,577.348,-16.938,81291,19318,596.623,-23.741,86227,24857,595.979,-202.572,79586,38354],[10170,395.772,-16.4,60721,18158,409.691,-19.992,63580,20750,406.547,-123.304,58067,28925],[10171,577.496,-17.123,70334,19545,592.95,-24.534,73966,24505,596.02,-249.306,68930,36337],[10172,699.658,-23.242,89689,21738,718.27,-31.333,96021,27825,724.973,-279.448,89564,44345],[10173,405.12,-14.64,54491,16844,416.837,-18.356,57173,19102,412.414,-153.735,52342,27162],[10174,547.795,-18.036,80094,20470,569.149,-25.105,84193,25323,566.196,-184.104,77112,37344],[10175,515.313,-16.782,69889,18902,527.592,-22.406,73421,22900,525.945,-188.655,68161,33742],[10176,448.92,-15.68,63910,18152,470.672,-21.515,68039,22221,467.419,-158.468,63152,31440],[10177,702.991,-59.243,79951,20549,723.548,-67.397,84007,25567,723.138,-332.74,79024,38670],[10178,676.017,-22.097,87565,21843,693.637,-29.431,92440,26773,694.699,-274.177,85336,41811],[10179,388.597,-16.405,55042,17688,402.559,-20.131,58040,20551,397.22,-134.906,53484,28595],[10180,594.225,-20.517,79196,21789,613.76,-28.01,82864,27032,618.868,-237.734,77159,39671],[10181,466.258,-17.354,66909,18996,486.568,-22.591,71358,23215,480.452,-161.845,65059,32559],[10182,583.0,-19.4,81174,21410,601.238,-26.067,85531,26443,600.013,-203.93,79420,39849],[10183,644.153,-23.177,81790,20967,664.194,-31.088,86635,26557,664.696,-263.356,80689,40857],[10184,535.556,-26.159,62442,18910,554.683,-32.112,66433,22992,551.033,-245.269,61622,32395],[10185,554.809,-24.936,74437,21122,575.64,-30.012,78506,24521,566.275,-216.007,72278,34875],[10186,586.273,-20.426,69301,20446,601.642,-26.183,73215,24722,595.71,-258.042,67549,35767],[10187,561.026,-28.012,72436,19196,587.266,-35.608,76597,23534,587.225,-235.444,71820,34667],[10188,543.807,-18.074,76183,19746,557.755,-24.355,79826,24323,554.19,-188.895,74180,37239],[10189,728.844,-22.889,89880,22103,749.153,-31.577,94951,28257,757.269,-316.957,88996,43697],[10190,670.703,-21.415,82152,20652,691.011,-29.318,86593,26388,694.897,-292.319,81128,40644],[10191,677.138,-24.598,85358,20381,694.801,-32.327,89216,26095,702.144,-286.331,83371,40868],[10192,675.946,-21.331,88831,21412,693.746,-27.877,93250,26495,700.45,-272.734,86509,42304],[10193,587.07,-18.555,81358,21005,608.683,-25.337,86070,26115,605.184,-210.504,79808,38975],[10194,518.792,-20.046,77343,21400,537.517,-25.39,80970,25249,531.883,-170.687,74223,36208],[10195,564.312,-18.814,76892,20527,580.041,-25.518,81176,25364,581.899,-209.109,75073,37388],[10196,595.417,-20.904,82498,21663,614.756,-27.631,86662,26572,612.626,-219.5,80233,39824],[10197,451.146,-17.221,62092,19263,464.405,-21.055,65132,21998,459.418,-163.118,59588,31450],[10198,565.551,-21.222,71423,20098,585.831,-27.433,75027,24608,584.188,-237.738,69687,35321],[10199,496.894,-23.824,61213,17874,511.404,-27.705,64109,21491,508.745,-213.108,59192,30921],[10200,571.209,-17.824,78372,19641,588.217,-25.143,82791,24875,583.47,-203.269,77101,37012],[10201,488.051,-16.077,71714,18399,503.453,-21.045,74911,22273,500.727,-158.427,68893,33258],[10202,578.92,-20.533,77273,20004,596.304,-26.18,81157,24139,589.25,-218.264,75617,37293],[10203,625.853,-20.652,82433,21919,646.177,-27.723,87694,27106,643.833,-242.316,81002,40141],[10204,464.964,-17.384,64023,19205,479.652,-22.95,67522,23083,475.856,-162.681,62647,32597],[10205,425.359,-15.274,61478,17688,440.522,-19.881,64878,20969,432.966,-145.198,59112,29089],[10206,530.955,-17.09,74915,19160,546.251,-22.618,78877,23134,549.035,-184.434,73439,35637],[10207,647.087,-20.333,84460,21671,666.887,-29.564,88922,27905,670.938,-255.723,83248,42831],[10208,688.226,-68.008,66631,19910,702.626,-73.036,70364,23760,695.502,-372.642,65399,33868],[10209,564.378,-19.862,70867,19001,580.351,-26.303,74508,23714,579.134,-231.251,69306,35400],[10210,433.316,-14.706,60350,17033,446.538,-19.392,63716,20521,447.961,-155.337,58780,30199],[10211,448.8,-15.225,59997,17104,461.133,-19.437,63041,20305,452.663,-167.866,57751,28976],[10212,632.479,-20.045,83426,21214,657.74,-27.518,90156,26896,661.041,-242.013,83897,41580],[10213,587.372,-18.967,79758,20224,609.316,-25.438,84550,25221,607.879,-223.196,78220,37817],[10214,580.842,-19.182,86674,21327,605.252,-26.673,91528,26328,601.017,-188.743,84371,39905],[10215,467.782,-16.395,62686,18090,479.552,-22.138,65390,22097,480.173,-177.713,60700,32165],[10216,464.001,-16.993,57517,18162,476.585,-21.722,60782,21466,477.426,-195.484,56153,31274],[10217,683.473,-22.831,88293,22893,699.638,-31.048,93444,28922,699.948,-267.271,87702,44801],[10218,637.668,-18.518,84918,21062,661.313,-27.644,90199,27477,667.423,-246.518,84372,41512],[10219,536.247,-19.403,80553,20981,549.807,-24.872,83955,24903,547.165,-170.019,76588,37161],[10220,537.75,-28.802,60017,19137,551.031,-32.557,63406,21918,549.105,-255.415,58518,32104],[10221,551.544,-24.239,63333,19681,567.098,-30.133,66802,24014,571.455,-258.1,62560,34804],[10222,625.865,-19.796,78054,20597,648.794,-27.072,83210,26171,646.267,-264.686,77034,40012],[10223,672.87,-19.346,81797,20529,695.754,-31.707,86475,26605,692.444,-294.53,80259,39569],[10224,445.636,-17.832,62645,17750,459.017,-22.132,65273,20352,451.636,-161.604,59507,29044],[10225,489.569,-16.549,69542,18788,506.789,-22.693,73050,23003,501.186,-170.357,67418,33332],[10226,714.546,-20.335,88539,21092,737.493,-29.71,92627,27723,739.147,-306.973,86442,41952],[10227,450.361,-15.573,61435,17874,466.573,-19.795,64908,20656,459.693,-168.232,59134,29431],[10228,548.399,-18.225,79442,19805,566.203,-23.952,83534,24072,559.632,-181.514,77504,36402],[10229,437.619,-15.568,67026,18147,454.285,-20.605,70285,21562,442.74,-135.76,63798,30748],[10230,644.104,-20.923,88568,21961,662.566,-28.606,93846,27758,665.578,-232.69,87536,43048],[10231,706.277,-21.075,85322,21424,726.703,-29.954,90642,28098,731.369,-303.74,85486,43364],[10232,511.021,-17.055,69210,18762,525.775,-23.151,72285,22898,524.786,-196.438,66921,33425],[10233,454.642,-16.48,64728,18696,470.396,-21.671,67777,22529,466.239,-150.931,63125,32166],[10234,548.561,-20.759,70158,19664,566.195,-26.297,74716,24415,571.087,-219.788,69393,35677],[10235,676.768,-21.882,79835,20353,693.17,-30.039,85023,25803,702.759,-307.9,79650,40523],[10236,451.557,-16.754,66995,18164,466.333,-20.686,70210,21228,466.941,-148.569,64918,31569],[10237,657.323,-19.864,83180,21298,673.891,-28.195,87369,27179,682.757,-270.237,81928,42688],[10238,470.238,-23.12,56189,19441,481.584,-26.761,58865,21767,477.716,-205.733,54799,30492],[10239,604.093,-19.083,78162,20421,631.293,-27.397,83419,26226,629.336,-245.701,77535,38833],[10240,621.755,-31.798,79485,20582,639.54,-35.592,83499,25211,632.843,-260.814,76249,37092],[10241,417.37,-17.954,57810,17668,427.616,-20.825,60363,19996,419.938,-151.705,55028,27512],[10242,433.374,-15.042,57013,16527,444.784,-19.273,59468,19573,441.17,-167.289,55176,29145],[10243,672.795,-58.843,69975,20669,687.694,-66.328,73657,24733,679.009,-339.479,68132,35410],[10244,446.166,-16.2,61238,18442,462.189,-21.339,64500,21799,459.46,-164.065,59275,31081],[10245,605.956,-18.959,79333,20267,626.347,-26.239,84022,25643,630.154,-241.705,78443,39383],[10246,571.132,-19.415,79282,21148,588.897,-25.561,83030,25591,588.712,-216.97,76191,37350],[10247,511.537,-18.473,73938,19963,528.627,-23.708,77279,23846,523.018,-172.959,71052,35391],[10248,509.235,-18.574,75754,20005,522.875,-23.379,79583,23742,517.509,-160.8,73207,35339],[10249,546.015,-17.594,74287,19768,561.249,-24.275,77630,24632,566.546,-207.904,72130,36422],[10250,410.399,-14.897,58514,16952,426.173,-19.119,61403,19933,415.03,-142.978,55996,27237],[10251,396.462,-16.616,57537,18609,406.063,-19.842,59863,20963,398.4,-131.711,54936,28122],[10252,406.015,-15.707,57198,17384,419.62,-19.85,60196,20226,408.806,-142.367,55149,28424],[10253,585.124,-17.983,84674,20455,606.779,-25.252,89485,25797,598.534,-197.796,82205,38391],[10254,507.302,-17.275,62518,19188,527.215,-24.739,66520,23957,522.451,-216.513,61502,33380],[10255,606.516,-18.283,82915,20674,627.803,-25.985,87228,26135,631.798,-228.29,81124,39774]],"rounding":"retail/arb edges 0.001 Y; volumes rounded to whole Y (max error .5 Y). Per-seed total edge is retail+arb. Aggregate metrics and RMS use unrounded values"},"estimated_table_diagnostic":{"metrics":{"n":256,"mean":526.7555024399737,"se":5.831062481054699,"ci95":[515.3266199771065,538.184384902841],"retail_edge":562.9816597884741,"arb_edge":-36.226157348500394,"retail_volume_y":74713.2223949701,"arb_volume_y":24950.685262385345,"rms_bps_after_own_callbacks":16.87768864629013,"rms_observations":2621726},"oracle_deficit":{"n":256,"mean":8.460985306018356,"se":0.08756348553931848,"ci95":[8.289360874361293,8.63260973767542]},"gain_over_analytic":{"n":256,"mean":2.3975462056819303,"se":0.15957234460154635,"ci95":[2.0847844102628996,2.710308001100961]},"per_seed_edge":[437.487,565.635,521.612,564.51,518.1,521.849,478.827,519.076,652.972,645.637,469.298,363.822,450.304,395.571,694.816,619.855,556.828,488.178,473.739,448.585,489.997,404.181,432.396,626.943,471.091,439.614,340.517,542.372,521.539,436.081,586.076,332.695,629.629,528.211,558.486,625.776,539.465,438.095,611.658,581.962,602.672,624.04,704.796,630.312,412.215,378.639,424.209,590.353,488.811,378.28,368.624,417.318,422.924,528.976,438.387,704.064,576.882,502.91,431.257,500.271,920.455,554.274,598.866,420.949,652.861,577.357,477.616,629.417,496.057,613.99,405.228,428.154,596.614,559.308,523.764,553.754,521.934,421.487,613.617,643.526,479.007,560.444,558.863,458.775,468.948,446.757,565.793,561.544,432.926,408.026,579.523,477.964,551.26,652.792,472.814,587.795,484.458,504.297,496.088,462.063,435.088,529.647,729.19,733.7,696.018,465.388,422.538,483.718,526.077,730.281,567.575,397.254,457.451,428.442,528.454,653.023,517.093,404.787,415.56,408.243,542.025,527.904,580.089,498.094,521.76,447.994,449.938,488.282,726.233,528.571,453.12,489.49,496.521,623.481,528.029,452.929,610.123,614.814,492.097,459.648,546.858,574.948,619.811,464.721,432.724,474.52,754.182,462.046,684.538,512.989,576.504,467.903,409.533,548.967,483.268,469.607,526.224,528.845,430.321,430.378,415.746,451.348,714.698,659.289,677.062,432.68,412.095,486.292,616.347,563.053,381.188,562.728,678.155,390.625,533.583,497.768,439.094,648.469,655.319,373.603,576.805,454.54,567.595,624.422,513.83,534.836,567.312,543.175,524.944,710.956,653.652,655.759,657.654,574.18,502.573,546.981,577.316,433.829,550.814,477.141,555.419,474.473,559.438,610.518,450.438,411.382,514.756,630.19,620.337,544.452,420.007,432.818,620.319,573.638,567.889,450.847,446.608,660.101,625.873,516.24,511.351,529.616,612.355,653.869,428.967,476.0,700.179,437.526,532.986,423.919,623.046,687.377,495.713,441.772,530.59,657.606,435.695,637.036,446.259,593.212,592.892,397.084,417.124,612.672,432.525,591.641,555.19,494.506,491.302,529.404,396.943,379.052,391.124,573.218,493.29,594.818],"seeds":"10000..10255 in order; full splits reproducible by diagnostic command"},"ablation":{"change":"Force first own trade at each timestamp to be arb (probability 1); subsequent same-timestamp trades remain retail. No retuning.","metrics":{"n":64,"mean":377.614377715482,"se":10.444833544214195,"ci95":[357.1425039688222,398.08625146214183],"retail_edge":549.0598450501486,"arb_edge":-171.44546733466674,"retail_volume_y":71108.7110190657,"arb_volume_y":33834.865321329686,"rms_bps_after_own_callbacks":49.22627485386075,"rms_observations":564247},"ablation_minus_analytic":{"n":64,"mean":-138.23574758532882,"se":3.859055357087771,"ci95":[-145.79949608522085,-130.6719990854368]},"oracle_deficit":{"n":64,"mean":148.77544112750414,"se":3.94614037631918,"ci95":[141.04100598991855,156.50987626508973]},"columns":["seed","retail_edge","arb_edge","retail_volume_y","arb_volume_y"],"rows":[[10000,443.614,-136.437,57056,29922],[10001,586.969,-160.981,84817,37516],[10002,522.066,-146.753,63546,32428],[10003,586.857,-193.767,79810,38227],[10004,520.406,-154.698,68576,33778],[10005,540.546,-144.668,69222,32701],[10006,494.722,-138.788,71111,33904],[10007,533.518,-140.894,74323,33994],[10008,665.377,-200.072,90752,39737],[10009,641.661,-162.316,84060,37346],[10010,480.673,-133.908,71747,33190],[10011,382.159,-115.169,57215,29099],[10012,472.023,-133.512,73262,34476],[10013,416.852,-132.547,56774,29171],[10014,697.159,-165.123,77269,34137],[10015,643.636,-176.342,89688,39395],[10016,599.828,-247.572,65375,31831],[10017,501.336,-144.719,67754,33476],[10018,492.779,-125.235,66165,32634],[10019,460.626,-171.582,59390,31047],[10020,499.103,-166.015,64866,32899],[10021,384.374,-102.427,47929,24417],[10022,447.687,-132.458,57439,28958],[10023,641.171,-157.963,81434,36479],[10024,487.577,-154.08,70518,35389],[10025,451.986,-152.284,59288,28825],[10026,356.872,-110.715,52536,27982],[10027,550.61,-150.496,68811,32608],[10028,548.916,-211.825,66167,33785],[10029,442.023,-101.951,69260,31080],[10030,607.819,-161.472,86222,38812],[10031,344.964,-88.045,55953,28988],[10032,638.481,-195.424,80795,36569],[10033,538.092,-129.523,77539,35034],[10034,563.892,-170.807,78263,34835],[10035,649.835,-155.718,89229,38814],[10036,554.004,-160.183,77554,35604],[10037,444.914,-123.676,62754,31992],[10038,632.218,-176.952,89293,39471],[10039,601.573,-160.026,74707,35929],[10040,627.404,-165.101,86655,39321],[10041,638.216,-202.112,81776,37625],[10042,724.792,-243.957,89949,40697],[10043,631.55,-177.966,82330,36358],[10044,409.211,-121.047,54926,28481],[10045,394.72,-101.002,59451,30365],[10046,447.907,-126.547,68588,34521],[10047,608.573,-155.056,87999,38526],[10048,509.663,-153.529,68960,32614],[10049,390.963,-116.026,55096,27734],[10050,382.937,-129.065,56428,29928],[10051,429.067,-149.051,58079,30172],[10052,437.837,-118.493,66232,32803],[10053,538.475,-124.656,67367,31566],[10054,460.778,-167.924,62477,32551],[10055,705.163,-219.527,89688,40406],[10056,594.614,-165.928,80403,36869],[10057,513.366,-165.551,64057,32049],[10058,438.684,-138.509,54006,28438],[10059,521.297,-164.098,69853,34872],[10060,2049.834,-1346.51,83073,39525],[10061,569.097,-165.653,78657,36371],[10062,610.945,-136.427,85427,36875],[10063,435.821,-131.651,61011,30282]]},"training":{"columns":["stage","candidate","mean_edge","rms_bps","weights"],"rows":[["fit",0,503.738496,19.456351,[35.0,7.0,36.0,11.0,0.28,1.0,20.0,0.7,0.35,0.5]],["fit",1,503.352721,19.729946,[35,7,36,11,0.28,1,20,0.7,0,0.5]],["fit",2,510.947159,17.07456,[35,7,36,11,0.28,1,20,0.7,0.35,0]],["fit",3,501.689656,22.436238,[35,7,36,11,0.28,1,20,1.4,0.35,0.5]],["fit",4,502.570097,20.883231,[35,7,36,11,0.28,1,20,0.35,0.35,0.5]],["fit",5,504.678961,19.110799,[35,7,36,11,0.28,1,40,0.7,0.35,0.5]],["fit",6,499.735065,22.148449,[35,7,36,11,0.28,1,10,0.7,0.35,0.5]],["fit",7,476.802353,23.396287,[35,7,36,11,0.28,1,20,0.7,0.35,1.5]],["fit",8,503.823754,17.314275,[35,7,24,8,0.28,1,20,0.7,0.35,0.5]],["fit",9,501.68218,21.734413,[35,7,50,14,0.28,1,20,0.7,0.35,0.5]],["fit",10,507.625081,18.288704,[30,7,36,11,0.28,1,20,0.7,0.35,0.5]],["fit",11,499.364354,20.549974,[40,7,36,11,0.28,1,20,0.7,0.35,0.5]],["refine",0,506.703122,20.988854,[35,7,36,11,0.28,0.5,20,0.7,0.35,0,0]],["refine",1,510.947159,17.07456,[35,7,36,11,0.28,1.0,20,0.7,0.35,0,0]],["refine",2,510.773346,18.207949,[35,7,36,11,0.28,2.0,20,0.7,0.35,0,0]],["refine",3,508.489361,21.057341,[35,7,36,11,0.28,4.0,20,0.7,0.35,0,0]],["refine",4,493.901718,22.97397,[35,7,36,11,0.28,0.5,20,0.7,0.35,0,1]],["refine",5,486.776641,24.949173,[35,7,36,11,0.28,1.0,20,0.7,0.35,0,1]],["refine",6,475.165324,27.790599,[35,7,36,11,0.28,2.0,20,0.7,0.35,0,1]],["refine",7,458.366474,31.549524,[35,7,36,11,0.28,4.0,20,0.7,0.35,0,1]]]},"leakage_tests":{"commands":["python experiment.py leakage","python verify.py"],"results":[{"passed":true,"callbacks":256,"hidden_poison_invariance":true,"repeated_timestamp_arb_probability_zero":true},{"passed":true,"observable_signature":true,"only_math_imported":true,"arb_pin_identity_both_sides":true,"hidden_input_invariance_callbacks":1024,"repeated_timestamp_checks":255,"ambiguous_first_trade_observed":true,"fee_and_state_finite_checks":1024}],"separation":"Only Arm instrumentation receives truth; it calls model.after_swap before recording errors. The model imports math only. Hidden-value poisoning changes instrumentation truth but leaves fees and every model field identical.","artifact_replay":{"commands":["python -S verify.py","python experiment.py eval --start 10000 --count 1 --workers 1"],"result":"PASS: extracted embedded files, verified all six SHA-256 hashes, repeated both leakage tests, and matched all raw fields for analytic/oracle/fixed seed 10000 exactly. verify.py also passes with site packages disabled."}},"runnable":{"requirements":"Python 3.12 standard library only. No pip packages, compiler, network, wheel, or submodule needed for the delivered experiment. Setup-only tools are not runtime dependencies.","extract":"Load this JSON with Python json.load; write every entry of runnable.files verbatim into one empty directory, using its key as filename; run commands there.","commands":["python experiment.py fit --workers 8","python experiment.py refine --workers 8","python experiment.py leakage","python verify.py","python experiment.py eval --start 10000 --count 64 --workers 8","python experiment.py eval --start 10064 --count 64 --workers 8","python experiment.py eval --start 10128 --count 64 --workers 8","python experiment.py eval --start 10192 --count 64 --workers 8","python experiment.py ablation --start 10000 --count 64 --workers 8","python experiment.py diagnostic --start 10000 --count 256 --workers 8"],"resume":"Training checkpoints each candidate; evaluation checkpoints each 64 seeds. Packaged frozen.json permits evaluation without training.","files":{"model.py":"import math\n\n# State is exclusively initialized reserves and own observable trades.\nDEFAULT = [35., 7., 36., 11., .28, 1., 20., .7, .35, .5]\n\ndef bounded(x,a,b): return min(b,max(a,x))\ndef moments(m,v,lo,hi):\n    sd=math.sqrt(max(v,1e-10)); a=(lo-m)/sd; b=(hi-m)/sd\n    cdf=lambda z: .5*math.erfc(-z/math.sqrt(2))\n    pdf=lambda z: math.exp(-.5*z*z)/math.sqrt(2*math.pi)\n    mass=max(1e-12,cdf(b)-cdf(a))\n    if mass < 1e-8: return bounded(m,lo,hi), min(v,(hi-lo)**2/12),mass\n    pa,pb=pdf(a),pdf(b); shift=(pa-pb)/mass\n    return m+sd*shift, max(.01,v*(1+(a*pa-b*pb)/mass-shift*shift)),mass\n\nclass Analytic:\n    def __init__(self, params=None, ablate=False):\n        self.params=list(DEFAULT if params is None else params); self.ablate=ablate\n    def initialize(self,x=100.,y=10000.):\n        self.m=math.log(y/x)*1e4; self.v=1.; self.s=self.m\n        self.last=-1; self.fb=self.fa=.0035; self.q=9.45**2\n        self.arbprob=0.; self.count=0\n        return self.fb,self.fa\n    def after_swap(self,sell_x,ax,ay,t,x,y):\n        base,rebate,knee,width,slope,noise,scale,retail_rate,router_gain,uncert=self.params[:10]\n        post=math.log(y/x)*1e4; direction=-1 if sell_x else 1\n        lo=self.s+math.log1p(-self.fb)*1e4\n        hi=self.s-math.log1p(-self.fa)*1e4\n        dt=max(0,t-self.last); first=t!=self.last\n        # Aggregate moment projection for unobserved silent steps; bounded O(1).\n        if dt>1:\n            self.v+=self.q*noise*(dt-1)*.5\n            self.m,self.v,_=moments(self.m,self.v,lo,hi)\n            self.v+=self.q*noise*(dt-1)*.5\n        if first: self.v+=self.q*noise\n        oldm=self.m; priorv=self.v\n        pin=post+(math.log1p(-self.fb)*1e4 if sell_x else -math.log1p(-self.fa)*1e4)\n        impact=max(.001,abs(post-self.s))\n        # Under retail-first hypothesis the current timestamp had no own arb.\n        rm,rv,mass=moments(self.m,self.v,lo,hi) if first else (self.m,self.v,1.)\n        # Router provides only a weak, heteroscedastic observation.\n        obs=pin-direction*(30.+.5*impact)\n        obsvar=(30.+impact)**2/max(.001,router_gain)\n        gain=rv/(rv+obsvar)\n        rm+=gain*bounded(obs-rm,-3*math.sqrt(rv+obsvar),3*math.sqrt(rv+obsvar))\n        rv=max(.01,rv*(1-gain))\n        sd=math.sqrt(max(priorv,.01)); z=(pin-oldm)/sd\n        arbden=math.exp(-.5*z*z)/(sd*math.sqrt(2*math.pi)) if first else 0.\n        if first and len(self.params)>10 and self.params[10]:\n            arbden=2/(math.sqrt(3)*math.pi*sd)*(1+z*z/3)**-2\n        # Heavy-tailed retail impact density, positive on both sides.\n        lz=(math.log(impact/scale)+.5*1.2**2)/1.2\n        retden=retail_rate*.5*mass*math.exp(-.5*lz*lz)/(impact*1.2*math.sqrt(2*math.pi))\n        w=arbden/(arbden+retden+1e-30) if first else 0.\n        if self.ablate: w=1. if first else 0.\n        self.m=w*pin+(1-w)*rm\n        self.v=max(.01,w*(pin-self.m)**2+(1-w)*(rv+(rm-self.m)**2))\n        # Robust innovation-based volatility, only high-confidence own arb pins.\n        if first and w>.95 and dt>0:\n            innovation=(pin-oldm)**2/max(1,dt)\n            self.q=.995*self.q+.005*bounded(innovation,6.**2,14.**2)\n        self.arbprob=w; self.last=t; self.s=post; self.count+=1\n        d=post-self.m\n        spread=base+uncert*math.sqrt(self.v)\n        def offset(z):\n            logistic=1/(1+math.exp(bounded(-(z-knee)/width,-60,60)))\n            # Smooth approximation to positive part.\n            pos=.5*(math.sqrt(z*z+25.)-z)\n            return spread-(spread+rebate)*logistic+slope*pos\n        self.fb=bounded(-math.expm1((-offset(d)-d)/1e4),0.,.1)\n        self.fa=bounded(-math.expm1((d-offset(-d))/1e4),0.,.1)\n        return self.fb,self.fa\n","experiment.py":"import argparse, json, math, statistics, time, pathlib, concurrent.futures, hashlib, random\nimport replica\nfrom model import Analytic, DEFAULT\nROOT=pathlib.Path(__file__).resolve().parent\nTABLE=json.loads((ROOT/'oracle_table.json').read_text())\nclass Arm:\n    def __init__(self,kind,params=None):\n        self.kind=kind; self.model=Analytic(params,kind=='ablation'); self.err2=0.; self.n=0\n    def init(self,*hidden):\n        return self.model.initialize()\n    def on_trade(self,sell,ax,ay,t,x,y,arb,p,rival):\n        if self.kind=='fixed': return .0035,.0035\n        if self.kind=='oracle':\n            d=math.log((y/x)/p); z=max(0.,min(16.,d/.003+8)); i=min(15,int(z)); f=z-i\n            ask=TABLE['ask'][i]*(1-f)+TABLE['ask'][i+1]*f+TABLE['arb_flag']*arb\n            bid=TABLE['bid'][i]*(1-f)+TABLE['bid'][i+1]*f+TABLE['arb_flag']*arb\n            return 1-math.exp(-bid*.003-d),1-math.exp(d-ask*.003)\n        fees=self.model.after_swap(sell,ax,ay,t,x,y)\n        # This line is evaluation-only: truth never reaches the model.\n        self.err2+=(self.model.m-math.log(p)*1e4)**2; self.n+=1\n        if self.kind=='estimated_table':\n            d=math.log(y/x)-self.model.m/1e4; z=max(0.,min(16.,d/.003+8)); i=min(15,int(z)); f=z-i\n            ask=TABLE['ask'][i]*(1-f)+TABLE['ask'][i+1]*f+TABLE['arb_flag']*self.model.arbprob\n            bid=TABLE['bid'][i]*(1-f)+TABLE['bid'][i+1]*f+TABLE['arb_flag']*self.model.arbprob\n            self.model.fb=max(0.,min(.1,1-math.exp(-bid*.003-d)))\n            self.model.fa=max(0.,min(.1,1-math.exp(d-ask*.003)))\n            return self.model.fb,self.model.fa\n        return fees\n\ndef one(job):\n    seed,kind,params=job; arm=Arm(kind,params); r=replica.run(seed,arm,10000)\n    return [seed,r['edge'],r['retail'],r['arb'],r['rv'],r['av'],arm.err2,arm.n]\n\ndef batch(seeds,kind,params,workers=8):\n    with concurrent.futures.ProcessPoolExecutor(max_workers=workers) as pool:\n        return list(pool.map(one,[(s,kind,params) for s in seeds]))\n\ndef summary(rows):\n    scores=[r[1] for r in rows]; n=len(rows); se=statistics.stdev(scores)/math.sqrt(n) if n>1 else None\n    return {'n':n,'mean':statistics.mean(scores),'se':se,'ci95':[statistics.mean(scores)-1.96*se,statistics.mean(scores)+1.96*se] if se else None,'rms_bps_callback':math.sqrt(sum(r[6] for r in rows)/max(1,sum(r[7] for r in rows)))}\n\ndef main():\n    ap=argparse.ArgumentParser(); ap.add_argument('mode',choices=['fit','refine','eval','ablation','leakage','diagnostic']); ap.add_argument('--start',type=int,default=10000); ap.add_argument('--count',type=int,default=32); ap.add_argument('--workers',type=int,default=8); a=ap.parse_args()\n    if a.mode=='fit':\n        # Predeclared structural alternatives, no held-out seeds in selection.\n        candidates=[DEFAULT,\n          [35,7,36,11,.28,1,20,.7,0,.5],\n          [35,7,36,11,.28,1,20,.7,.35,0],\n          [35,7,36,11,.28,1,20,1.4,.35,.5],\n          [35,7,36,11,.28,1,20,.35,.35,.5],\n          [35,7,36,11,.28,1,40,.7,.35,.5],\n          [35,7,36,11,.28,1,10,.7,.35,.5],\n          [35,7,36,11,.28,1,20,.7,.35,1.5],\n          [35,7,24,8,.28,1,20,.7,.35,.5],\n          [35,7,50,14,.28,1,20,.7,.35,.5],\n          [30,7,36,11,.28,1,20,.7,.35,.5],\n          [40,7,36,11,.28,1,20,.7,.35,.5]]\n        path=ROOT/'fit.json'; results=json.loads(path.read_text()) if path.exists() else []\n        for i,params in enumerate(candidates):\n            if i<len(results): continue\n            begin=time.monotonic(); rows=batch(range(20000,20128),'analytic',params,a.workers)\n            r={'index':i,'params':params,'summary':summary(rows),'seconds':time.monotonic()-begin}\n            results.append(r); path.write_text(json.dumps(results)); print(json.dumps(r),flush=True)\n        best=max(results,key=lambda r:r['summary']['mean'])\n        (ROOT/'frozen.json').write_text(json.dumps(best)); return\n    if a.mode=='refine':\n        initial=json.loads((ROOT/'fit.json').read_text())\n        best=max(initial,key=lambda r:r['summary']['mean']); base=best['params']\n        candidates=[]\n        for tail in (0,1):\n            for noise in (.5,1.,2.,4.):\n                params=list(base)+[tail]; params[5]=noise; candidates.append(params)\n        path=ROOT/'refine.json'; results=json.loads(path.read_text()) if path.exists() else []\n        for i,params in enumerate(candidates):\n            if i<len(results): continue\n            begin=time.monotonic(); rows=batch(range(20000,20128),'analytic',params,a.workers)\n            r={'index':i,'params':params,'summary':summary(rows),'seconds':time.monotonic()-begin}\n            results.append(r); path.write_text(json.dumps(results)); print(json.dumps(r),flush=True)\n        best=max(initial+results,key=lambda r:r['summary']['mean'])\n        (ROOT/'frozen.json').write_text(json.dumps(best)); return\n    if a.mode=='leakage':\n        # Synthetic observable tape replay with hidden arguments poisoned.\n        params=json.loads((ROOT/'frozen.json').read_text())['params']; u=Arm('analytic',params); v=Arm('analytic',params)\n        assert u.init(.001,.8,20)==v.init(float('nan'),float('nan'),float('nan'))\n        x,y=100.,10000.; t=0\n        for i in range(256):\n            t+=1 if i%3 else 0; sell=i%2==0; ax=.01+(i%11)*.005\n            if sell: nx=x+ax*(1-u.model.fb); ny=1e6/nx; ay=y-ny\n            else: nx=x-ax; ny=1e6/nx; ay=(ny-y)/(1-u.model.fa)\n            fu=u.on_trade(sell,ax,ay,t,nx,ny,False,100.,100.)\n            fv=v.on_trade(sell,ax,ay,t,nx,ny,True,999.,float('nan'))\n            assert fu==fv and vars(u.model)==vars(v.model)\n            if i and t==u.model.last and i%3==0: assert u.model.arbprob==0\n            x,y=nx,ny\n        out={'passed':True,'callbacks':256,'hidden_poison_invariance':True,'repeated_timestamp_arb_probability_zero':True}\n        (ROOT/'leakage.json').write_text(json.dumps(out)); print(json.dumps(out)); return\n    params=json.loads((ROOT/'frozen.json').read_text())['params']\n    assert 10000<=a.start and a.start+a.count<=10256\n    kinds=['estimated_table'] if a.mode=='diagnostic' else (['ablation'] if a.mode=='ablation' else ['analytic','oracle','fixed'])\n    begin=time.monotonic(); out={k:batch(range(a.start,a.start+a.count),k,params,a.workers) for k in kinds}\n    out['seconds']=time.monotonic()-begin\n    (ROOT/f'{a.mode}_{a.start}_{a.count}.json').write_text(json.dumps(out))\n    print(json.dumps({k:summary(out[k]) for k in kinds}),flush=True)\nif __name__=='__main__': main()\n","verify.py":"import ast, json, math, pathlib, random\nfrom model import Analytic\nfrom experiment import Arm, ROOT\nparams=json.loads((ROOT/'frozen.json').read_text())['params']\n# The deployable model imports math only and has no path to the simulator.\ntree=ast.parse((ROOT/'model.py').read_text())\nimports=[n.names[0].name for n in ast.walk(tree) if isinstance(n,ast.Import)]\nassert imports==['math']\nassert not any(isinstance(n,ast.ImportFrom) for n in ast.walk(tree))\nassert list(__import__('inspect').signature(Analytic.after_swap).parameters)==['self','sell_x','ax','ay','t','x','y']\n# Verify both signs of trade-implied pin using the precise arb target reserves.\nfor sell in [False,True]:\n a=Analytic(params); a.initialize(); p=100.8 if not sell else 99.2\n g=1-(a.fb if sell else a.fa)\n spot=p/g if sell else p*g; x=math.sqrt(1e6/spot); y=1e6/x\n implied=math.log(y/x)*1e4+(math.log(g)*1e4 if sell else -math.log(g)*1e4)\n assert abs(implied-math.log(p)*1e4)<1e-8\n# Two identical observable tapes with drastically different hidden values.\nu,v=Arm('analytic',params),Arm('analytic',params)\nassert u.init(.000882,.6,19)==v.init(.1,999.,999.)\nx,y=100.,10000.; timestamp=0; repeated=0; ambiguous=False\nfor i in range(1024):\n dt=[0,1,1,2,0,10,100,1][i%8]; timestamp+=dt; sell=i%2==0; ax=.001+(.007*(i%50))\n if sell: nx=x+ax*(1-u.model.fb); ny=1e6/nx; ay=y-ny\n else: nx=x-ax; ny=1e6/nx; ay=(ny-y)/(1-u.model.fa)\n priorlast=u.model.last\n fu=u.on_trade(sell,ax,ay,timestamp,nx,ny,False,100.,90.)\n fv=v.on_trade(sell,ax,ay,timestamp,nx,ny,True,10000.,float('nan'))\n assert fu==fv and vars(u.model)==vars(v.model)\n assert all(math.isfinite(f) and 0<=f<=.1 for f in fu)\n assert math.isfinite(u.model.m) and math.isfinite(u.model.v) and u.model.v>=0\n if timestamp==priorlast:\n  assert u.model.arbprob==0; repeated+=1\n else: ambiguous |= 0<u.model.arbprob<1\n x,y=nx,ny\nassert ambiguous\nout={'passed':True,'observable_signature':True,'only_math_imported':True,'arb_pin_identity_both_sides':True,'hidden_input_invariance_callbacks':1024,'repeated_timestamp_checks':repeated,'ambiguous_first_trade_observed':ambiguous,'fee_and_state_finite_checks':1024}\n(ROOT/'verify.json').write_text(json.dumps(out)); print(json.dumps(out))\n","replica.py":"\"\"\"Pure-Python replica of benedictbrady/amm-challenge @6b6921f8 (engine.rs, cfmm.rs,\narbitrageur.rs, router.rs, retail.rs), written from the source for ceiling experiments.\nRNG streams differ from the Rust engine, so absolute scores move by the seed set;\ncompare policies paired on the same seeds.\n\nA policy gets a callback after every trade on its own pool (arb or retail), like afterSwap,\nplus an oracle view (true fair price p and the 30 bps pool's spot s2 after the pending\naction of the same event), and returns (bid_fee, ask_fee).\"\"\"\nimport math, random\n\nK = 100.0 * 10000.0\nMAXF = 0.1\nNF = 0.003  # normalizer fee\n\n\ndef clampf(f):\n    return 0.0 if f < 0.0 else (MAXF if f > MAXF else f)\n\n\ndef run(seed, policy, n_steps=10000, cfg=None):\n    cr = random.Random(10_000_000 + seed)\n    size_mean = cr.uniform(19.0, 21.0)\n    lam = cr.uniform(0.6, 1.0)\n    sigma = cr.uniform(0.000882, 0.001008)\n    if cfg:\n        size_mean, lam, sigma = cfg\n    pr = random.Random(seed)\n    rr = random.Random(seed + 1)\n    drift = -0.5 * sigma * sigma\n    mu_ln = math.log(size_mean) - 0.5 * 1.2 * 1.2\n    emlam = math.exp(-lam)\n    p = 100.0\n    x1, y1 = 100.0, 10000.0\n    x2, y2 = 100.0, 10000.0\n    fb, fa = policy.init(sigma, lam, size_mean)\n    fb, fa = clampf(fb), clampf(fa)\n    e1 = e2 = 0.0\n    ret1 = arb1 = 0.0  # our retail / arb edge\n    rv1 = av1 = 0.0\n    sum_fb = sum_fa = 0.0\n    g2 = 1.0 - NF\n    sqrt = math.sqrt\n    for t in range(n_steps):\n        p *= math.exp(drift + sigma * pr.gauss(0.0, 1.0))\n        # --- arbitrage, normalizer first (pools are independent), then ours\n        s2 = y2 / x2\n        if s2 < p:\n            nx = sqrt(K / (g2 * p)); ax = x2 - nx\n            if ax > 0:\n                ax = min(ax, x2 * 0.99)\n                nrx = x2 - ax; nry = K / nrx; net_y = nry - y2; ty = net_y / g2\n                prof = ax * p - ty\n                if prof > 0:\n                    x2, y2 = nrx, nry; e2 -= prof\n        elif s2 > p:\n            xv = sqrt(K * g2 / p); ax = (xv - x2) / g2\n            if ax > 0:\n                nrx = x2 + ax * g2; nry = K / nrx; yo = y2 - nry\n                prof = yo - ax * p\n                if yo > 0 and prof > 0:\n                    x2, y2 = nrx, nry; e2 -= prof\n        s1 = y1 / x1\n        if s1 < p:\n            g = 1.0 - fa\n            if g > 0:\n                nx = sqrt(K / (g * p)); ax = x1 - nx\n                if ax > 0:\n                    ax = min(ax, x1 * 0.99)\n                    nrx = x1 - ax; nry = K / nrx; net_y = nry - y1; ty = net_y / g\n                    prof = ax * p - ty\n                    if prof > 0:\n                        x1, y1 = nrx, nry; e1 -= prof; arb1 -= prof; av1 += ty\n                        fb, fa = policy.on_trade(False, ax, ty, t, x1, y1, True, p, y2 / x2)\n                        fb, fa = clampf(fb), clampf(fa)\n        elif s1 > p:\n            g = 1.0 - fb\n            if g > 0:\n                xv = sqrt(K * g / p); ax = (xv - x1) / g\n                if ax > 0:\n                    nrx = x1 + ax * g; nry = K / nrx; yo = y1 - nry\n                    prof = yo - ax * p\n                    if yo > 0 and prof > 0:\n                        x1, y1 = nrx, nry; e1 -= prof; arb1 -= prof; av1 += yo\n                        fb, fa = policy.on_trade(True, ax, yo, t, x1, y1, True, p, y2 / x2)\n                        fb, fa = clampf(fb), clampf(fa)\n        # --- retail\n        n = 0; q = rr.random()\n        while q > emlam:\n            n += 1; q *= rr.random()\n        for _ in range(n):\n            size = rr.lognormvariate(mu_ln, 1.2)\n            buy = rr.random() < 0.5\n            if buy:  # trader buys X with Y\n                ga = 1.0 - fa\n                a1 = sqrt(x1 * ga * y1); a2 = sqrt(x2 * g2 * y2); r = a1 / a2\n                d = ga + r * g2\n                u1 = (r * (y2 + g2 * size) - y1) / d if d != 0 else size / 2\n                u1 = 0.0 if u1 < 0 else (size if u1 > size else u1)\n                u2 = size - u1\n                if u2 > 1e-4:\n                    nry = y2 + u2 * g2; nrx = K / nry; xo = x2 - nrx\n                    if xo > 0:\n                        e2 += u2 - xo * p; x2, y2 = nrx, nry\n                if u1 > 1e-4:\n                    nry = y1 + u1 * ga; nrx = K / nry; xo = x1 - nrx\n                    if xo > 0:\n                        ed = u1 - xo * p; e1 += ed; ret1 += ed; rv1 += u1\n                        x1, y1 = nrx, nry\n                        fb, fa = policy.on_trade(False, xo, u1, t, x1, y1, False, p, y2 / x2)\n                        fb, fa = clampf(fb), clampf(fa)\n            else:  # trader sells X\n                tx = size / p\n                gb = 1.0 - fb\n                b1 = sqrt(y1 * gb * x1); b2 = sqrt(y2 * g2 * x2); r = b1 / b2\n                d = gb + r * g2\n                v1 = (r * (x2 + g2 * tx) - x1) / d if d != 0 else tx / 2\n                v1 = 0.0 if v1 < 0 else (tx if v1 > tx else v1)\n                v2 = tx - v1\n                if v2 > 1e-4:\n                    nrx = x2 + v2 * g2; nry = K / nrx; yo = y2 - nry\n                    if yo > 0:\n                        e2 += v2 * p - yo; x2, y2 = nrx, nry\n                if v1 > 1e-4:\n                    nrx = x1 + v1 * gb; nry = K / nrx; yo = y1 - nry\n                    if yo > 0:\n                        ed = v1 * p - yo; e1 += ed; ret1 += ed; rv1 += yo\n                        x1, y1 = nrx, nry\n                        fb, fa = policy.on_trade(True, v1, yo, t, x1, y1, False, p, y2 / x2)\n                        fb, fa = clampf(fb), clampf(fa)\n        sum_fb += fb; sum_fa += fa\n    return dict(edge=e1, norm=e2, retail=ret1, arb=arb1, rv=rv1, av=av1,\n                fb=sum_fb / n_steps * 1e4, fa=sum_fa / n_steps * 1e4,\n                sigma=sigma, lam=lam, size=size_mean)\n\n\nclass Fixed:\n    def __init__(self, f): self.f = f\n    def init(self, *a): return self.f, self.f\n    def on_trade(self, *a): return self.f, self.f\n\n\nclass OracleQuote:\n    \"\"\"Quote ask p(1+m) and bid p(1-m) around the TRUE fair price after every own trade.\"\"\"\n    def __init__(self, m): self.m = m\n    def init(self, *a): return 0.003, 0.003\n    def on_trade(self, is_buy, ax, ay, t, x, y, arb, p, s2):\n        s = y / x\n        fa = 1.0 - s / (p * (1 + self.m))\n        fb = 1.0 - p * (1 - self.m) / s\n        return fb, fa\n\n\nclass OracleCompete:\n    \"\"\"Know p and the 30 bps pool's spot s2. Ask quote: undercut the rival's ask by u,\n    but never below p(1+mlo); bid quote: outbid the rival's bid by u, never above p(1-mlo).\"\"\"\n    def __init__(self, mlo, u, mhi=1.0):\n        self.mlo, self.u, self.mhi = mlo, u, mhi\n    def init(self, *a): return 0.003, 0.003\n    def on_trade(self, is_buy, ax, ay, t, x, y, arb, p, s2):\n        s = y / x\n        ra = s2 / (1 - NF); rb = s2 * (1 - NF)\n        qa = max(p * (1 + self.mlo), min(ra * (1 - self.u), p * (1 + self.mhi)))\n        qb = min(p * (1 - self.mlo), max(rb * (1 + self.u), p * (1 - self.mhi)))\n        return 1.0 - qb / s, 1.0 - s / qa\n","oracle_table.json":"{\n \"units\": \"offset from the true price p in units of 30 bps; ask above p, bid below p; knots at own displacement ln(spot/p) = -8..+8 x 30 bps; add arb_flag to both offsets right after an own arb trade\",\n \"fees\": \"askFee = 1 - exp(ln(spot/p) - ask*0.003); bidFee = 1 - exp(-bid*0.003 - ln(spot/p)); clamp to [0, 0.10]\",\n \"ask\": [\n  -0.8836,\n  -0.9345,\n  -0.8159,\n  -0.7282,\n  -0.6342,\n  -0.5652,\n  -0.2453,\n  0.7252,\n  1.1735,\n  1.4444,\n  1.6154,\n  2.7552,\n  2.9767,\n  3.3159,\n  4.4695,\n  5.071,\n  5.7838\n ],\n \"bid\": [\n  5.6921,\n  5.1279,\n  4.4766,\n  4.0233,\n  3.3293,\n  2.4659,\n  1.7209,\n  1.4329,\n  1.2125,\n  0.7478,\n  -0.2275,\n  -0.5564,\n  -0.6444,\n  -0.7169,\n  -0.7926,\n  -0.8617,\n  -0.8452\n ],\n \"arb_flag\": 0.1463,\n \"replica_score_1000_seeds\": 532.45,\n \"same_table_p_from_own_arbs_only\": 507.61,\n \"retrained_for_p_from_own_arbs\": 519.24\n}","frozen.json":"{\"index\": 2, \"params\": [35, 7, 36, 11, 0.28, 1, 20, 0.7, 0.35, 0], \"summary\": {\"n\": 128, \"mean\": 510.947159039017, \"se\": 7.189586053451346, \"ci95\": [496.85557037425235, 525.0387477037816], \"rms_bps_callback\": 17.074560192515463}, \"seconds\": 24.54600156401284}"},"sha256":{"model.py":"9dae7b09d55b3ec504c6185f684a717ec4ac034e132af2b5834571f2f423a53a","experiment.py":"1060611a60aaf1843853e1892a1faade7ab1bddd974f5afbe5bd4967c5365af1","verify.py":"ec7676f69991cd4fe174058a157877586bc6e795444a1dadf91144da03a59d47","replica.py":"d079712fa1f4850ad8337e656e7ee8dbdde806cd4762d597fae50f547ef9c7c4","oracle_table.json":"3e9d1fc8874c6d796d4c26948d486b587f2a6b5610b20fba99923a60954cf4b5","frozen.json":"9ea337e4e63154b1a331bb854de624be1b9c05b474ce0f8bb4cfc4bb044a3b57"}},"runtime":{"wall_seconds_since_research_download":863.9688286781311,"python":"3.12.3","workers":8,"fit_job_seconds":253.83166729169898,"refine_job_seconds":182.9850594336167,"evaluation_job_seconds":109.12463344004937,"ablation_job_seconds":30.162919827969745,"diagnostic_job_seconds":95.45361944590695,"completed_evaluation_seeds":256,"completed_ablation_seeds":64,"max_allowed_command_seconds":480,"setup":"pinned checkout and supplied setup.sh completed; wheel checksum and StarterStrategy validation passed"},"port_state_cost_estimate":{"core_scalar_state":7,"unpacked_storage_bytes":224,"diagnostic_scalars":2,"fee_parameters":11,"work_per_callback":"O(1), no grid, no history buffer, no elapsed-time loop. Logs, exp/expm1, sqrt, erfc need bounded fixed-point implementations. Student t3 branch is optional; selected branch recorded above.","gas_measured":null,"gas_target":245000,"qualification":"No Solidity implementation or gas guarantee. Arithmetic and storage costs require actual benchmarking. Fees-to-Python parity not tested."},"limitations":["Replica RNG/config generation is not MatchRunner._build_configs (which uses NumPy RNG), and replica is not the Rust engine. No translation of these scores to board points.","Silent-step moment projection is approximate; unobserved retail absence and precise router split selection are not modeled. Retail impact likelihood is a heuristic.","Innovation-based q is selected-trade biased; uncertainty calibration is not guaranteed. RMS measured after own events can understate error before next trade.","Oracle comparison includes both policy shape differences and perfect price/arb-label information; the estimated-table diagnostic helps separate these but is still endogenous.","The 128 training seeds are repeatedly reused for selection; held-out results are the only performance estimate. This is not the requested later +4.6 vs RS2 on 4000 fresh real-engine seeds."],"dead_ends_with_numbers":[{"variant":"uncertainty width 1.5 vs 0, otherwise same first-stage constants","training_mean":[476.802353046638,510.947159039017],"conclusion":"Direct widening substantially reduced edge; selected width and remaining experiments are reported without hiding negative results."},{"variant":"t3 arb likelihood; diffusion .5,1,2,4","training_means":[493.901718,486.776641,475.165324,458.366474],"conclusion":"All below selected Gaussian mean 510.947159."}],"freeze_sha256":"aaba4afc739094704e70b3effc0e54d17c8827e25d65a1a64e9e728046a204d7","conclusion":"Not qualified. Analytic oracle deficit 10.859 (95% CI 10.503..11.214); estimated-table deficit 8.461 exceeds the 2-point filter target. No engine/RS2 claim."}
