{"key":"d-arb-by-order","approach":{"measurement":"Descriptive fill attribution of incumbent N only; no candidate, tuning, promotion, or strategy change.","incumbent_url":"https://api.imd.fun/artifacts/db22aff7167a4798251ed3d20432509f30e511e4958799aba5b17b41bb39b4b6","incumbent_code_sha256":"bf0d179f3e47d28819f6fa7e0d25548c2ced43374c216d36460b00a26a26ffae","simulator_commit":"b244530ee2a6247b5f930830eeca9dd950da0e7e","config":"sample_config(ChallengeConfig(), ParameterVariance(), seed=seed); 2000 steps; serial in one OS process, no sandbox or workers.","instrumentation":"Scratch subclass of PredictionMarket; original SimulationEngine and original matching/accounting methods. execute_arbitrage records current probability and its change from the preceding step (initial probability on step 0). _fill_order reads competitor_best_quotes immediately before calling the original method and attributes returned participant fills. No random draws added.","distance":"Live same-side competitor quote immediately before each individual fill, after any earlier fills on that step. For bids: competitor_bid-order_tick; for asks: order_tick-competitor_ask. Negative=inside, zero=at, one=behind_1, >=2=behind_2plus. no_quote is separate when that side has no competitor orders; it is not assigned an invented tick.","jump":"abs(p_after_process_step-p_before_process_step)>=0.03; based on true probability movement, not a latent process jump indicator.","table_units":"Edges in simulator score units. Each mean and SE uses all runs of that spread, including zero for runs without that class. Quantity is filled YES-share quantity, separately mean per run and total over that spread. bid=participant BUY; ask=participant SELL. net_edge_mean=arb+retail; net_edge_se computed from per-run sums, preserving covariance.","standard_errors":"Sample standard deviation of per-run class edge divided by sqrt(runs). Descriptive uncertainty only; no candidate difference or causal improvement was measured.","coverage":"Both aggressors, both sides, four requested distance classes plus no_quote, and both step classes. Behind rows are zero because price priority exhausts better live competitor orders first; this does not measure distance when the order was submitted.","evidence":"Tables and check values are locally executed measurements, not externally published scores; source URLs and exact quotes are in sources.json."},"tried":{"code_hash_verified":true,"baseline_comparison":{"seeds":"310000..310009","all_fields_exactly_equal":true,"per_seed_total_edge":{"310000":17.1160017503,"310001":20.0171113118,"310002":2.3903113616,"310003":241.4847522404,"310004":8.5209352976,"310005":0.037767639,"310006":4.5896171672,"310007":333.6631744596,"310008":31.143690212,"310009":32.1980250109},"seed_310000_expected_rounded":17.116002,"max_total_edge_difference":0},"all_500_runs":{"successful":true,"edge_reconciliation_tolerance":1e-06,"quantity_reconciliation_tolerance":1e-06,"fill_count_reconciled_exactly":true,"each_aggressor_edge_reconciled":true,"recomputed_all_cell_means_and_standard_errors":true,"finalist_seeds_used":false},"no_candidate_built":true},"complete":true,"first_seed":500000,"last_seed":500499,"n":500,"incumbent_mean":116.082447,"retail_mean":142.885347,"arb_mean":-26.8029,"by_spread":{"1":{"runs":134,"incumbent_mean":45.040271,"retail_mean":70.859088,"arb_mean":-25.818817,"table":{"columns":["side","distance","step","arb_edge_mean","arb_edge_se","arb_quantity_mean","arb_quantity_total","retail_edge_mean","retail_edge_se","retail_quantity_mean","retail_quantity_total","net_edge_mean","net_edge_se"],"rows":[["bid","inside","jump",-0.093376,0.04162,2.259179,302.73,0.001773,0.001222,0.020224,2.71,-0.091603,0.041663],["bid","inside","nonjump",-5.459797,0.607091,2559.553731,342980.2,1.442357,0.172052,282.144179,37807.32,-4.01744,0.52805],["bid","at","jump",-0.316918,0.130016,13.504478,1809.6,0.016795,0.015901,0.459701,61.6,-0.300123,0.13072],["bid","at","nonjump",-0.469373,0.097391,277.579776,37195.69,2.155434,0.380009,225.308433,30191.33,1.686061,0.363719],["bid","behind_1","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","behind_1","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","behind_2plus","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","behind_2plus","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","no_quote","jump",-2.033662,1.202977,266.954403,35771.89,0.0,0.0,0.0,0.0,-2.033662,1.202977],["bid","no_quote","nonjump",-5.41144,1.223842,10152.372463,1360417.91,0.140735,0.0375,65.853358,8824.35,-5.270704,1.220256],["ask","inside","jump",-0.848591,0.685506,20.620896,2763.2,0.001579,0.000995,0.024627,3.3,-0.847012,0.68552],["ask","inside","nonjump",-3.369846,0.620977,1998.193582,267757.94,45.875299,7.035427,6046.85903,810279.11,42.505453,6.894457],["ask","at","jump",-7.590213,7.331193,161.752687,21674.86,0.0003,0.0003,0.033358,4.47,-7.589913,7.331193],["ask","at","nonjump",-0.211931,0.051591,169.382537,22697.26,21.214435,5.180349,1375.789403,184355.78,21.002504,5.175392],["ask","behind_1","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","behind_1","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","behind_2plus","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","behind_2plus","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","no_quote","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","no_quote","nonjump",-0.013672,0.002426,7.605672,1019.16,0.010381,0.003712,2.958507,396.44,-0.00329,0.003511]]}},"2":{"runs":119,"incumbent_mean":133.315031,"retail_mean":163.21405,"arb_mean":-29.899019,"table":{"columns":["side","distance","step","arb_edge_mean","arb_edge_se","arb_quantity_mean","arb_quantity_total","retail_edge_mean","retail_edge_se","retail_quantity_mean","retail_quantity_total","net_edge_mean","net_edge_se"],"rows":[["bid","inside","jump",-1.083467,0.276107,23.224034,2763.66,0.024076,0.011864,0.317227,37.75,-1.059391,0.276772],["bid","inside","nonjump",-9.192369,0.867866,3722.953445,443031.46,9.291833,0.640534,917.281597,109156.51,0.099464,0.84278],["bid","at","jump",-0.802614,0.193334,9.877395,1175.41,0.019654,0.016386,0.180756,21.51,-0.78296,0.194702],["bid","at","nonjump",-1.054442,0.162846,407.22,48459.18,8.557526,0.735351,488.201513,58095.98,7.503084,0.709065],["bid","behind_1","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","behind_1","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","behind_2plus","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","behind_2plus","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","no_quote","jump",-0.583495,0.583495,91.729916,10915.86,0.0,0.0,0.0,0.0,-0.583495,0.583495],["bid","no_quote","nonjump",-6.044314,1.588995,6693.206303,796491.55,1.071622,0.378912,193.924538,23077.02,-4.972693,1.60722],["ask","inside","jump",-3.519272,1.276004,60.203361,7164.2,0.047928,0.022349,0.68916,82.01,-3.471344,1.276944],["ask","inside","nonjump",-3.705263,0.491788,1279.838067,152300.73,115.97003,17.178728,6954.992269,827644.08,112.264768,17.137314],["ask","at","jump",-3.248394,1.940384,23.329496,2776.21,0.018765,0.007638,0.341092,40.59,-3.229629,1.939765],["ask","at","nonjump",-0.620141,0.15194,205.826218,24493.32,28.172234,3.064737,1311.87437,156113.05,27.552093,3.029373],["ask","behind_1","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","behind_1","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","behind_2plus","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","behind_2plus","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","no_quote","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","no_quote","nonjump",-0.045248,0.011075,13.040756,1551.85,0.040382,0.015151,6.597311,785.08,-0.004866,0.016374]]}},"3":{"runs":119,"incumbent_mean":106.968237,"retail_mean":137.000889,"arb_mean":-30.032652,"table":{"columns":["side","distance","step","arb_edge_mean","arb_edge_se","arb_quantity_mean","arb_quantity_total","retail_edge_mean","retail_edge_se","retail_quantity_mean","retail_quantity_total","net_edge_mean","net_edge_se"],"rows":[["bid","inside","jump",-1.901639,0.286487,44.111092,5249.22,0.211957,0.067018,2.32479,276.65,-1.689682,0.294253],["bid","inside","nonjump",-7.428956,0.76989,2921.694874,347681.69,18.565621,1.00881,1195.160924,142224.15,11.136665,1.004893],["bid","at","jump",-1.254331,0.315298,26.022353,3096.66,0.038207,0.016905,0.518655,61.72,-1.216124,0.315922],["bid","at","nonjump",-0.246728,0.080526,77.52042,9224.93,9.918102,0.629625,367.612101,43745.84,9.671374,0.61565],["bid","behind_1","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","behind_1","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","behind_2plus","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","behind_2plus","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","no_quote","jump",0.0,0.0,0.0,0.0,0.009166,0.009166,0.014454,1.72,0.009166,0.009166],["bid","no_quote","nonjump",-4.163997,1.329598,2563.528151,305059.85,1.101101,0.494372,137.047899,16308.7,-3.062896,1.356175],["ask","inside","jump",-7.048892,1.974199,86.767227,10325.3,0.17048,0.053738,2.257227,268.61,-6.878413,1.974398],["ask","inside","nonjump",-2.977254,0.251581,754.545462,89790.91,78.158887,11.347188,3274.735126,389693.48,75.181633,11.269039],["ask","at","jump",-4.808544,2.690521,36.304622,4320.25,0.060906,0.051573,0.435126,51.78,-4.747638,2.691938],["ask","at","nonjump",-0.148705,0.020967,40.328739,4799.12,28.639721,3.132388,944.053613,112342.38,28.491016,3.129678],["ask","behind_1","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","behind_1","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","behind_2plus","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","behind_2plus","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","no_quote","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","no_quote","nonjump",-0.053605,0.010426,11.667647,1388.45,0.12674,0.044912,14.517899,1727.63,0.073135,0.041956]]}},"4":{"runs":128,"incumbent_mean":182.907174,"retail_mean":204.859202,"arb_mean":-21.952028,"table":{"columns":["side","distance","step","arb_edge_mean","arb_edge_se","arb_quantity_mean","arb_quantity_total","retail_edge_mean","retail_edge_se","retail_quantity_mean","retail_quantity_total","net_edge_mean","net_edge_se"],"rows":[["bid","inside","jump",-4.31187,0.590703,108.962344,13947.18,0.345613,0.079141,4.584062,586.76,-3.966257,0.591472],["bid","inside","nonjump",-5.613838,0.522473,2357.178047,301718.79,27.258754,1.5383,1427.992266,182783.01,21.644917,1.407012],["bid","at","jump",-1.051661,0.233063,22.788281,2916.9,0.029595,0.013136,1.707578,218.57,-1.022066,0.233964],["bid","at","nonjump",-0.090807,0.01423,27.304141,3494.93,12.266414,0.639898,357.108359,45709.87,12.175607,0.637332],["bid","behind_1","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","behind_1","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","behind_2plus","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","behind_2plus","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["bid","no_quote","jump",-0.145461,0.102418,56.175547,7190.47,0.0,0.0,0.0,0.0,-0.145461,0.102418],["bid","no_quote","nonjump",-2.118945,0.684135,2489.179531,318614.98,2.093935,0.98924,276.497109,35391.63,-0.025009,1.052731],["ask","inside","jump",-6.459735,3.69852,103.757891,13281.01,0.315387,0.118629,3.461406,443.06,-6.144348,3.702524],["ask","inside","nonjump",-1.398303,0.1141,303.968906,38908.02,121.509892,16.952852,3798.301016,486182.53,120.111589,16.950586],["ask","at","jump",-0.587943,0.183398,16.044375,2053.68,0.113113,0.044419,1.506797,192.87,-0.47483,0.19086],["ask","at","nonjump",-0.121146,0.021836,30.212734,3867.23,40.818016,4.727748,1024.257188,131104.92,40.69687,4.72669],["ask","behind_1","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","behind_1","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","behind_2plus","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","behind_2plus","nonjump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","no_quote","jump",0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0,0.0],["ask","no_quote","nonjump",-0.05232,0.012893,12.458672,1594.71,0.108482,0.029236,13.008672,1665.11,0.056163,0.029547]]}}},"finding":"Spread 2: the largest arbitrage-loss cell is inside bids on nonjump steps (-9.192369, SE 0.867866), but its retail edge leaves net +0.099464 (SE 0.842780); the largest net-loss cell is no-quote bids on nonjump steps (-4.972693, SE 1.607220). Spread 3: the largest arbitrage-loss cell is also inside bids on nonjump steps (-7.428956, SE 0.769890), yet it nets +11.136665; the largest net-loss cell is inside asks on jump steps (-6.878413, SE 1.974398). Across all distances, jump-step asks net -6.700973 (SE 2.311367) at spread 2 and -11.626050 (SE 3.321655) at spread 3, whereas nonjump inside asks net +112.264768 and +75.181633 respectively. This points to testing one change: increase the modeled upward-jump loss component used to size asks in spreads 2 and 3 before orders are placed, rather than conditioning on the unobservable future jump. These are realized-fill diagnostics, not measured recoverable gains or evidence about unfilled opportunities; live-quote behind classes have zero fills.","next":"Hand off one candidate hypothesis: an ask-only multiplier on the existing upward-jump loss term for spreads 2 and 3, using only information available at order placement; preserve the nonjump retail opportunity and compare paired total edge on disjoint tuning and holdout seeds through the designated candidate and measure items. No candidate was built or evaluated here, and no finalist seeds were touched.","seconds_per_simulation":0.286136}