{ "approach": "Measurement only: B unchanged; partition every participant fill by post-process-step true probability used for edge scoring, crossed with its 0-based fill step. Twenty cross-bucket rows followed by five probability and four step marginals. Default ChallengeConfig() and ParameterVariance(), 2000 steps; one serial, unsandboxed run_batch call per seed. Separate OS processes run disjoint seed ranges. No candidate and no oracle.", "tried": [ { "name": "incumbent integrity", "sha256": "baef955ec10728839f7c3357016ec9eb0550e7ec689bbd5171b1cdd5c9e37273", "matches_required": true }, { "name": "instrumentation noninterference", "seeds": { "start": 233000, "end_inclusive": 233019 }, "n": 20, "all_fields_exactly_identical": true, "fields_compared": "Full dataclasses.asdict(SimulationResult), including total_edge, retail_edge, arb_edge, failed, regime, all inventory/cash/wealth/trading statistics and error. Python exact dictionary equality, no tolerance. Baseline scoring method runs first; observer only reads its existing fills and probability into external counters.", "pairs": [ { "seed": 233000, "baseline_total_edge": -0.2048099781, "instrumented_total_edge": -0.2048099781, "baseline_retail_edge": 0.0221898588, "instrumented_retail_edge": 0.0221898588, "baseline_arb_edge": -0.2269998369, "instrumented_arb_edge": -0.2269998369, "failed": false }, { "seed": 233001, "baseline_total_edge": 30.053870923, "instrumented_total_edge": 30.053870923, "baseline_retail_edge": 41.2782229357, "instrumented_retail_edge": 41.2782229357, "baseline_arb_edge": -11.2243520126, "instrumented_arb_edge": -11.2243520126, "failed": false }, { "seed": 233002, "baseline_total_edge": 17.335944413, "instrumented_total_edge": 17.335944413, "baseline_retail_edge": 26.4727395883, "instrumented_retail_edge": 26.4727395883, "baseline_arb_edge": -9.1367951755, "instrumented_arb_edge": -9.1367951755, "failed": false }, { "seed": 233003, "baseline_total_edge": 17.7698609906, "instrumented_total_edge": 17.7698609906, "baseline_retail_edge": 24.8535957783, "instrumented_retail_edge": 24.8535957783, "baseline_arb_edge": -7.0837347882, "instrumented_arb_edge": -7.0837347882, "failed": false }, { "seed": 233004, "baseline_total_edge": 174.7209562388, "instrumented_total_edge": 174.7209562388, "baseline_retail_edge": 199.0922207295, "instrumented_retail_edge": 199.0922207295, "baseline_arb_edge": -24.3712644918, "instrumented_arb_edge": -24.3712644918, "failed": false }, { "seed": 233005, "baseline_total_edge": 132.9377392096, "instrumented_total_edge": 132.9377392096, "baseline_retail_edge": 172.849310573, "instrumented_retail_edge": 172.849310573, "baseline_arb_edge": -39.9115713631, "instrumented_arb_edge": -39.9115713631, "failed": false }, { "seed": 233006, "baseline_total_edge": 111.4781247896, "instrumented_total_edge": 111.4781247896, "baseline_retail_edge": 120.7859831424, "instrumented_retail_edge": 120.7859831424, "baseline_arb_edge": -9.3078583502, "instrumented_arb_edge": -9.3078583502, "failed": false }, { "seed": 233007, "baseline_total_edge": 77.8513826679, "instrumented_total_edge": 77.8513826679, "baseline_retail_edge": 99.3125544373, "instrumented_retail_edge": 99.3125544373, "baseline_arb_edge": -21.4611717686, "instrumented_arb_edge": -21.4611717686, "failed": false }, { "seed": 233008, "baseline_total_edge": 155.504247461, "instrumented_total_edge": 155.504247461, "baseline_retail_edge": 205.4538705463, "instrumented_retail_edge": 205.4538705463, "baseline_arb_edge": -49.9496230855, "instrumented_arb_edge": -49.9496230855, "failed": false }, { "seed": 233009, "baseline_total_edge": 9.0664815404, "instrumented_total_edge": 9.0664815404, "baseline_retail_edge": 16.268889358, "instrumented_retail_edge": 16.268889358, "baseline_arb_edge": -7.2024078184, "instrumented_arb_edge": -7.2024078184, "failed": false }, { "seed": 233010, "baseline_total_edge": 96.1829724423, "instrumented_total_edge": 96.1829724423, "baseline_retail_edge": 118.5530936918, "instrumented_retail_edge": 118.5530936918, "baseline_arb_edge": -22.3701212491, "instrumented_arb_edge": -22.3701212491, "failed": false }, { "seed": 233011, "baseline_total_edge": 41.2859247545, "instrumented_total_edge": 41.2859247545, "baseline_retail_edge": 45.2101640007, "instrumented_retail_edge": 45.2101640007, "baseline_arb_edge": -3.9242392466, "instrumented_arb_edge": -3.9242392466, "failed": false }, { "seed": 233012, "baseline_total_edge": 107.5395607446, "instrumented_total_edge": 107.5395607446, "baseline_retail_edge": 119.613107084, "instrumented_retail_edge": 119.613107084, "baseline_arb_edge": -12.0735463391, "instrumented_arb_edge": -12.0735463391, "failed": false }, { "seed": 233013, "baseline_total_edge": 256.8520662651, "instrumented_total_edge": 256.8520662651, "baseline_retail_edge": 270.7325343464, "instrumented_retail_edge": 270.7325343464, "baseline_arb_edge": -13.8804680804, "instrumented_arb_edge": -13.8804680804, "failed": false }, { "seed": 233014, "baseline_total_edge": 182.5015741228, "instrumented_total_edge": 182.5015741228, "baseline_retail_edge": 199.6616187627, "instrumented_retail_edge": 199.6616187627, "baseline_arb_edge": -17.1600446402, "instrumented_arb_edge": -17.1600446402, "failed": false }, { "seed": 233015, "baseline_total_edge": 86.9442204355, "instrumented_total_edge": 86.9442204355, "baseline_retail_edge": 95.7140580844, "instrumented_retail_edge": 95.7140580844, "baseline_arb_edge": -8.7698376491, "instrumented_arb_edge": -8.7698376491, "failed": false }, { "seed": 233016, "baseline_total_edge": 194.8480105734, "instrumented_total_edge": 194.8480105734, "baseline_retail_edge": 254.9136939251, "instrumented_retail_edge": 254.9136939251, "baseline_arb_edge": -60.0656833515, "instrumented_arb_edge": -60.0656833515, "failed": false }, { "seed": 233017, "baseline_total_edge": 15.7952475394, "instrumented_total_edge": 15.7952475394, "baseline_retail_edge": 26.7904000078, "instrumented_retail_edge": 26.7904000078, "baseline_arb_edge": -10.9951524662, "instrumented_arb_edge": -10.9951524662, "failed": false }, { "seed": 233018, "baseline_total_edge": 25.9176568532, "instrumented_total_edge": 25.9176568532, "baseline_retail_edge": 31.7186576999, "instrumented_retail_edge": 31.7186576999, "baseline_arb_edge": -5.801000847, "instrumented_arb_edge": -5.801000847, "failed": false }, { "seed": 233019, "baseline_total_edge": 218.5352858175, "instrumented_total_edge": 218.5352858175, "baseline_retail_edge": 235.8251835302, "instrumented_retail_edge": 235.8251835302, "baseline_arb_edge": -17.2898977126, "instrumented_arb_edge": -17.2898977126, "failed": false } ] }, { "name": "20-seed timing gate", "mean_baseline_seconds": 1.3720036451937632, "mean_instrumented_seconds": 1.4524015531060286, "projected_1000_serial_seconds": 1452.4015531060286, "decision": "Projection below 1800 seconds; run all 1000 allowed seeds. First 20 instrumented runs reused in measurement, without double-counting." } ], "next": [ { "change": "Test halving main quote sizes (keep tiny probes) during steps 1800-1999 when B estimates a 5-80 cent probability. Apply the endgame gate only to this estimated-probability range so below-5-cent retail capture remains available.", "edge_at_stake": { "scope": "(5_to_20c or 20_to_80c) x 1800-1999", "mean": -1.5718757049801613, "se": 0.6077257043319831, "retail_mean": 0.801206658357048, "retail_se": 0.08102777724151644, "arb_mean": -2.3730823633372093, "arb_se": 0.6117593159530497, "gross_arb_loss_per_simulation": 2.3730823633372093 }, "interpretation": "Diagnostic selection uses true probability at fill time; B only has an earlier estimate, so this is a proposed test, not measured policy improvement. Net edge and SE for the union are recomputed from each seed, not by adding standard errors. Gross arbitrage loss is an upper bound on savings at unchanged fills." }, { "change": "Test uncertainty-sensitive stock acquisition and bid sizing near B estimates below 5 cents, separately from low-price asks, to reduce inventory acquisition losses while preserving retail sales.", "edge_at_stake": { "scope": "below_5c x all_steps", "arb_loss_per_simulation": 11.490496264251973, "arb_se": 0.8946531678662225, "retail_edge_per_simulation": 85.68106119884104 }, "interpretation": "Bucket-level evidence supports a sizing experiment; it does not identify which side or individual order caused loss. Gross loss is an upper bound, not a forecast gain." } ], "code": "\"\"\"Run from repository root. Stdlib only; requires the pinned simulator source and B.\nUsage: python3 measure.py SIMULATOR_ROOT B.py pilot|block OUTPUT [START END]\nEvery simulation uses serial run_batch(..., n_simulations=1, workers=1).\n\"\"\"\nimport sys, json, hashlib, time, math, statistics\nfrom pathlib import Path\nfrom dataclasses import asdict\n\nPIN = 'baef955ec10728839f7c3357016ec9eb0550e7ec689bbd5171b1cdd5c9e37273'\nPROB = ['below_5c', '5_to_20c', '20_to_80c', '80_to_95c', 'above_95c']\nSTEPS = ['0-199', '200-999', '1000-1799', '1800-1999']\n\ndef probability_bucket(p):\n # Boundary convention: [0,.05), [.05,.20), [.20,.80], (.80,.95], (.95,1].\n return 0 if p < .05 else 1 if p < .20 else 2 if p <= .80 else 3 if p <= .95 else 4\n\ndef step_bucket(s):\n return 0 if s < 200 else 1 if s < 1000 else 2 if s < 1800 else 3\n\ndef main():\n simulator, strategy, mode, output = sys.argv[1:5]\n assert hashlib.sha256(Path(strategy).read_bytes()).hexdigest() == PIN\n sys.path.insert(0, str(Path(simulator).resolve()))\n from orderbook_pm_challenge.config import ChallengeConfig, ParameterVariance\n from orderbook_pm_challenge.loader import load_strategy_factory\n from orderbook_pm_challenge.runner import run_batch\n from orderbook_pm_challenge.market import PredictionMarket, RETAIL, ARB\n from orderbook_pm_challenge.types import Side\n from orderbook_pm_challenge.utils import tick_to_price\n factory = load_strategy_factory(strategy)\n original = PredictionMarket.record_participant_fills\n collected = None\n step = 0\n def observe(market, fills, *, probability):\n nonlocal step\n original(market, fills, probability=probability)\n b = collected[4 * probability_bucket(probability) + step_bucket(step)]\n b['steps'] += 1\n for f in fills:\n assert f.step == step\n price = tick_to_price(f.price_ticks)\n edge = f.quantity * (probability-price) if f.side is Side.BUY else f.quantity * (price-probability)\n assert f.aggressor in (RETAIL, ARB)\n b['count'] += 1\n key = 'retail' if f.aggressor == RETAIL else 'arb'\n b[key] += edge\n b[key+'_count'] += 1\n step += 1\n def run(seed, instrument):\n nonlocal collected, step\n assert 233000 <= seed <= 233999\n collected = [dict(retail=0., arb=0., count=0, retail_count=0, arb_count=0, steps=0) for _ in range(20)]\n step = 0\n PredictionMarket.record_participant_fills = observe if instrument else original\n start = time.perf_counter()\n try:\n r = run_batch(factory, base_config=ChallengeConfig(), variance=ParameterVariance(), n_simulations=1, seed_start=seed, workers=1).simulation_results[0]\n finally:\n PredictionMarket.record_participant_fills = original\n elapsed = time.perf_counter()-start\n if instrument and not r.failed:\n assert step == 2000\n assert sum(b['count'] for b in collected) == r.fill_count\n assert abs(sum(b['retail'] for b in collected)-r.retail_edge) < 1e-6\n assert abs(sum(b['arb'] for b in collected)-r.arb_edge) < 1e-6\n return dict(seed=seed, result=asdict(r), buckets=collected if instrument else None, seconds=elapsed)\n records=[]\n if mode == 'pilot':\n for s in range(233000,233020):\n baseline=run(s, False)\n instrumented=run(s, True)\n assert baseline['result'] == instrumented['result'], ('changed outcome',s)\n instrumented['baseline_seconds']=baseline['seconds']\n instrumented['all_result_fields_identical']=True\n records.append(instrumented)\n Path(output).write_text(json.dumps(records))\n print(s, 'identical', round(baseline['seconds'],3), round(instrumented['seconds'],3),flush=True)\n elif mode == 'block':\n start,end=map(int,sys.argv[5:7])\n assert 233020 <= start < end <= 234000\n for s in range(start,end):\n records.append(run(s, True))\n Path(output).write_text(json.dumps(records))\n if len(records)%10 == 0:\n print(s, 'n',len(records),'seconds/sim',round(statistics.mean(r['seconds'] for r in records),3),flush=True)\n else:\n raise ValueError(mode)\n\n\n\"\"\"Summarize pilot.json plus block.json. No market simulations are run here.\"\"\"\nimport json, math, statistics, hashlib\nfrom pathlib import Path\n\nPROB = ['below_5c','5_to_20c','20_to_80c','80_to_95c','above_95c']\nSTEPS=['0-199','200-999','1000-1799','1800-1999']\n\ndef se(xs):\n return statistics.stdev(xs)/math.sqrt(len(xs)) if len(xs)>1 else None\n\ndef summarize(records):\n assert records\n assert len({r['seed'] for r in records}) == len(records)\n assert all(233000<=r['seed']<=233999 for r in records)\n assert all(not r['result']['failed'] for r in records), 'inspect failures before summarizing'\n rows=[]\n def add(name, indices):\n per=[]\n for record in records:\n bs=[record['buckets'][i] for i in indices]\n per.append({k:sum(b[k] for b in bs) for k in ['retail','arb','count','retail_count','arb_count','steps']})\n total=[p['retail']+p['arb'] for p in per]\n retail=[p['retail'] for p in per]\n arb=[p['arb'] for p in per]\n rows.append(dict(name=name,count=sum(p['count'] for p in per),retail_count=sum(p['retail_count'] for p in per),arb_count=sum(p['arb_count'] for p in per),step_count=sum(p['steps'] for p in per),simulations_with_steps=sum(p['steps']>0 for p in per),mean=statistics.mean(total),retail_mean=statistics.mean(retail),arb_mean=statistics.mean(arb),se=se(total),retail_se=se(retail),arb_se=se(arb)))\n for p in range(5):\n for t in range(4):\n add(PROB[p]+' x '+STEPS[t],[4*p+t])\n for p in range(5): add(PROB[p]+' x all_steps',list(range(4*p,4*p+4)))\n for t in range(4): add('all_probabilities x '+STEPS[t],[4*p+t for p in range(5)])\n edges=[r['result']['total_edge'] for r in records]\n retail=[r['result']['retail_edge'] for r in records]\n arb=[r['result']['arb_edge'] for r in records]\n incumbent=dict(mean=statistics.mean(edges),retail_mean=statistics.mean(retail),arb_mean=statistics.mean(arb),se=se(edges),retail_se=se(retail),arb_se=se(arb))\n assert abs(sum(r['mean'] for r in rows[:20])-incumbent['mean'])<1e-6\n assert sum(r['step_count'] for r in rows[:20])==2000*len(records)\n assert sum(r['count'] for r in rows[:20])==sum(r['result']['fill_count'] for r in records)\n return incumbent,rows\n\n\nif __name__ == '__main__':\n if len(sys.argv)>1 and sys.argv[1]=='summarize':\n records=[]\n for filename in sys.argv[2:]:\n records.extend(json.loads(Path(filename).read_text()))\n incumbent,findings=summarize(records)\n print(json.dumps(dict(n=len(records),incumbent=incumbent,findings=findings),indent=2))\n else:\n main()\n", "code_sha256": "62f34d767036a13d14dfbbed5fe72a0f6faf5fbb9d11597032549b58b2e4a262", "incumbent_code_sha256": "baef955ec10728839f7c3357016ec9eb0550e7ec689bbd5171b1cdd5c9e37273", "tuning": null, "seeds": { "start": 233000, "end_inclusive": 233999, "allowed_end_inclusive": 233999, "finalist_seeds_used": false }, "n": 1000, "complete": true, "incumbent": { "mean": 100.4765965316917, "retail_mean": 127.1401405735393, "arb_mean": -26.6635440418441, "se": 5.23909038504197, "retail_se": 5.616823111259487, "arb_se": 1.5700488304699465 }, "findings": [ { "name": "below_5c x 0-199", "count": 222, "retail_count": 41, "arb_count": 181, "step_count": 73, "simulations_with_steps": 2, "mean": 0.006855577491567738, "retail_mean": 0.014365423415548063, "arb_mean": -0.007509845923980326, "se": 0.004848916653813814, "retail_se": 0.010216996324606402, "arb_se": 0.005472523369017816 }, { "name": "below_5c x 200-999", "count": 49204, "retail_count": 6919, "arb_count": 42285, "step_count": 19378, "simulations_with_steps": 106, "mean": 3.882741386220933, "retail_mean": 5.536056792308257, "arb_mean": -1.6533154060873239, "se": 0.8494905121452835, "retail_se": 0.9304849091162757, "arb_se": 0.39275297163000294 }, { "name": "below_5c x 1000-1799", "count": 307956, "retail_count": 39769, "arb_count": 268187, "step_count": 136328, "simulations_with_steps": 368, "mean": 40.4855636546564, "retail_mean": 47.648967734137074, "arb_mean": -7.163404079480674, "se": 3.176032168928636, "retail_se": 3.3940775570264665, "arb_se": 0.686703534367906 }, { "name": "below_5c x 1800-1999", "count": 166906, "retail_count": 22915, "arb_count": 143991, "step_count": 76836, "simulations_with_steps": 523, "mean": 29.815404316220164, "retail_mean": 32.481671248980156, "arb_mean": -2.666266932759995, "se": 1.469334965164831, "retail_se": 1.5440976580633496, "arb_se": 0.2557929556899085 }, { "name": "5_to_20c x 0-199", "count": 11753, "retail_count": 1485, "arb_count": 10268, "step_count": 4453, "simulations_with_steps": 73, "mean": 0.13921729854555367, "retail_mean": 0.24442375845072317, "arb_mean": -0.1052064599051695, "se": 0.040870694316552235, "retail_se": 0.05133544658938006, "arb_se": 0.031745214691970766 }, { "name": "5_to_20c x 200-999", "count": 269021, "retail_count": 35702, "arb_count": 233319, "step_count": 100364, "simulations_with_steps": 347, "mean": 5.996527396593474, "retail_mean": 7.1791759583416885, "arb_mean": -1.1826485617482148, "se": 0.5302070145001169, "retail_se": 0.575311936572953, "arb_se": 0.13401721745243214 }, { "name": "5_to_20c x 1000-1799", "count": 302239, "retail_count": 38063, "arb_count": 264176, "step_count": 116730, "simulations_with_steps": 474, "mean": 5.151183700570551, "retail_mean": 8.953995686258637, "arb_mean": -3.802811985688084, "se": 0.8778408227291201, "retail_se": 0.6366330660631273, "arb_se": 0.7222358587466211 }, { "name": "5_to_20c x 1800-1999", "count": 33558, "retail_count": 3530, "arb_count": 30028, "step_count": 13078, "simulations_with_steps": 262, "mean": -0.9809580456786521, "retail_mean": 0.6356100387621051, "arb_mean": -1.6165680844407573, "se": 0.4932155356702947, "retail_se": 0.0760270785231271, "arb_se": 0.4988924957996158 }, { "name": "20_to_80c x 0-199", "count": 490155, "retail_count": 59633, "arb_count": 430522, "step_count": 191300, "simulations_with_steps": 1000, "mean": 3.122603220506682, "retail_mean": 3.8627701802108834, "arb_mean": -0.7401669597042011, "se": 0.1345833041606478, "retail_se": 0.1304334454556669, "arb_se": 0.06378472579475183 }, { "name": "20_to_80c x 200-999", "count": 1394465, "retail_count": 166073, "arb_count": 1228392, "step_count": 561874, "simulations_with_steps": 974, "mean": 8.318745732500096, "retail_mean": 10.648816105014264, "arb_mean": -2.3300703725141676, "se": 0.35284762704785433, "retail_se": 0.3791831722328163, "arb_se": 0.149746877554575 }, { "name": "20_to_80c x 1000-1799", "count": 790580, "retail_count": 83540, "arb_count": 707040, "step_count": 295852, "simulations_with_steps": 692, "mean": 1.6903016218124411, "retail_mean": 4.8560682227818734, "arb_mean": -3.1657666009694325, "se": 0.5592883477516292, "retail_se": 0.25444298096687473, "arb_se": 0.5280895921371291 }, { "name": "20_to_80c x 1800-1999", "count": 58300, "retail_count": 5674, "arb_count": 52626, "step_count": 23249, "simulations_with_steps": 279, "mean": -0.5909176593015093, "retail_mean": 0.16559661959494287, "arb_mean": -0.7565142788964522, "se": 0.34468442275014927, "retail_se": 0.019829892645979043, "arb_se": 0.3449006740375586 }, { "name": "80_to_95c x 0-199", "count": 10973, "retail_count": 1356, "arb_count": 9617, "step_count": 4154, "simulations_with_steps": 68, "mean": 0.027444311835372628, "retail_mean": 0.04425664736787143, "arb_mean": -0.016812335532498803, "se": 0.012070147514704083, "retail_se": 0.009282794267889792, "arb_se": 0.008874779290477001 }, { "name": "80_to_95c x 200-999", "count": 247963, "retail_count": 32829, "arb_count": 215134, "step_count": 96017, "simulations_with_steps": 353, "mean": 0.736919806080671, "retail_mean": 1.0009317616485864, "arb_mean": -0.2640119555679154, "se": 0.07122294694688334, "retail_se": 0.07596142161873903, "arb_se": 0.03135048970853827 }, { "name": "80_to_95c x 1000-1799", "count": 269716, "retail_count": 33920, "arb_count": 235796, "step_count": 105987, "simulations_with_steps": 456, "mean": 0.7888986323464348, "retail_mean": 1.0219062135471966, "arb_mean": -0.23300758120076182, "se": 0.06754600682965527, "retail_se": 0.07359981513151664, "arb_se": 0.030984100537623276 }, { "name": "80_to_95c x 1800-1999", "count": 34873, "retail_count": 3849, "arb_count": 31024, "step_count": 12870, "simulations_with_steps": 244, "mean": -0.264275705250895, "retail_mean": 0.10293615986633822, "arb_mean": -0.36721186511723325, "se": 0.2934956583450346, "retail_se": 0.0130019311196289, "arb_se": 0.2933417657021454 }, { "name": "above_95c x 0-199", "count": 61, "retail_count": 4, "arb_count": 57, "step_count": 20, "simulations_with_steps": 1, "mean": -0.000824646798327617, "retail_mean": 0.00013132595524224325, "arb_mean": -0.0009559727535698602, "se": 0.0008246467983276171, "retail_se": 0.00013132595524224325, "arb_se": 0.0009559727535698602 }, { "name": "above_95c x 200-999", "count": 57876, "retail_count": 7362, "arb_count": 50514, "step_count": 22367, "simulations_with_steps": 116, "mean": 0.24734530689915807, "retail_mean": 0.29135694631229625, "arb_mean": -0.04401163941313821, "se": 0.042778965947857475, "retail_se": 0.04484498306184281, "arb_se": 0.010965178428460638 }, { "name": "above_95c x 1000-1799", "count": 326449, "retail_count": 42021, "arb_count": 284428, "step_count": 145103, "simulations_with_steps": 357, "mean": 1.5564867158284512, "retail_mean": 1.6653567759287848, "arb_mean": -0.10887006010033366, "se": 0.11751613091819464, "retail_se": 0.12073084632245701, "arb_se": 0.011742860167248053 }, { "name": "above_95c x 1800-1999", "count": 157658, "retail_count": 20264, "arb_count": 137394, "step_count": 73967, "simulations_with_steps": 497, "mean": 0.3473339106433988, "retail_mean": 0.7857469746501069, "arb_mean": -0.4384130640067081, "se": 0.4140755897890107, "retail_se": 0.04211665142807186, "arb_se": 0.4115569197765718 }, { "name": "below_5c x all_steps", "count": 524288, "retail_count": 69644, "arb_count": 454644, "step_count": 232615, "simulations_with_steps": 535, "mean": 74.19056493458906, "retail_mean": 85.68106119884104, "arb_mean": -11.490496264251973, "se": 4.695645918293742, "retail_se": 5.12902105583206, "arb_se": 0.8946531678662225 }, { "name": "5_to_20c x all_steps", "count": 616571, "retail_count": 78780, "arb_count": 537791, "step_count": 234625, "simulations_with_steps": 614, "mean": 10.305970350030927, "retail_mean": 17.013205441813152, "arb_mean": -6.707235091782225, "se": 1.2339703956918473, "retail_se": 1.0836392651412328, "arb_se": 0.9355188259346796 }, { "name": "20_to_80c x all_steps", "count": 2733500, "retail_count": 314920, "arb_count": 2418580, "step_count": 1072275, "simulations_with_steps": 1000, "mean": 12.54073291551771, "retail_mean": 19.533251127601964, "arb_mean": -6.992518212084254, "se": 0.8341306879247998, "retail_se": 0.6819668783057375, "arb_se": 0.6591560915441962 }, { "name": "80_to_95c x all_steps", "count": 563525, "retail_count": 71954, "arb_count": 491571, "step_count": 219028, "simulations_with_steps": 596, "mean": 1.2889870450115835, "retail_mean": 2.1700307824299925, "arb_mean": -0.8810437374184092, "se": 0.3160144097433705, "retail_se": 0.12739069562066005, "arb_se": 0.29661409832581587 }, { "name": "above_95c x all_steps", "count": 542044, "retail_count": 69651, "arb_count": 472393, "step_count": 241457, "simulations_with_steps": 505, "mean": 2.1503412865726803, "retail_mean": 2.74259202284643, "arb_mean": -0.5922507362737498, "se": 0.4481939554356783, "retail_se": 0.1805712654131943, "arb_se": 0.4117763193205693 }, { "name": "all_probabilities x 0-199", "count": 513164, "retail_count": 62519, "arb_count": 450645, "step_count": 200000, "simulations_with_steps": 1000, "mean": 3.2952957615808485, "retail_mean": 4.1659473354002685, "arb_mean": -0.8706515738194195, "se": 0.14397027005599156, "retail_se": 0.14440427204981307, "arb_se": 0.07226172278399492 }, { "name": "all_probabilities x 200-999", "count": 2018529, "retail_count": 248885, "arb_count": 1769644, "step_count": 800000, "simulations_with_steps": 1000, "mean": 19.18227962829433, "retail_mean": 24.65633756362509, "arb_mean": -5.47405793533076, "se": 1.1371525739641328, "retail_se": 1.2536964449647459, "arb_se": 0.44504731645941575 }, { "name": "all_probabilities x 1000-1799", "count": 1996940, "retail_count": 237313, "arb_count": 1759627, "step_count": 800000, "simulations_with_steps": 1000, "mean": 49.67243432521428, "retail_mean": 64.14629463265356, "arb_mean": -14.473860307439287, "se": 3.329887026489231, "retail_se": 3.466765475461729, "arb_se": 1.148413480749737 }, { "name": "all_probabilities x 1800-1999", "count": 451295, "retail_count": 56232, "arb_count": 395063, "step_count": 200000, "simulations_with_steps": 1000, "mean": 28.326586816632506, "retail_mean": 34.17156104185365, "arb_mean": -5.844974225221145, "se": 1.658710153974373, "retail_se": 1.5238586762505621, "arb_se": 0.8366348309269026 } ], "paired": null, "failures": { "n": 0, "seeds": [] }, "seconds_per_simulation": 5.817736739528598, "methods": { "simulator_commit": "b244530ee2a6247b5f930830eeca9dd950da0e7e", "probability_boundaries": "[0,0.05), [0.05,0.20), [0.20,0.80], (0.80,0.95], (0.95,1]; exactly 20c and 80c go in the central bucket; exactly 5c and 95c go in their adjacent inner buckets.", "step_boundaries": "0-199, 200-999, 1000-1799, 1800-1999 inclusive, using fill.step (0 based).", "count": "Number of participant fills, not orders, shares, or seeds. retail_count and arb_count split this count; step_count counts observed time steps including steps without fills.", "mean": "Every bucket mean is sum of its edge over all n simulations divided by n, including zero contribution from simulations without fills in that bucket. Marginal rows overlap the first 20 cross rows and must not be added to them.", "se": "Sample standard deviation (ddof=1) of per-simulation bucket edge divided by sqrt(n). retail_se and arb_se use the respective per-simulation components. Fills are not independent statistical units. Marginal SEs are recomputed from per-seed marginal sums.", "edge": "q*(p-price) on participant buys; q*(price-p) on sells. p is the post-advance true probability used by original scoring. Edge units are dollars per simulation, not cents per fill.", "rounding": "Incumbent metrics retain simulator stepwise rounding to 10 decimal places. Bucket sums use unrounded per-fill edge. Each seed total is checked against each recorded simulator component within 1e-6, and fill counts reconcile exactly.", "timing": "seconds_per_simulation is mean elapsed wall time of instrumented run_batch calls, including OS scheduling while four independent processes share this machine; the pilot timing gate ran alone. Excludes baseline repeats and download/setup.", "reproduction": "Provide pinned simulator checkout and exact B.py. Save code as harness.py. Run: python3 harness.py SIMULATOR_ROOT B.py pilot pilot.json ; then python3 harness.py SIMULATOR_ROOT B.py block block.json 233020 234000 ; then python3 harness.py summarize pilot.json block.json . Measured block actually used four disjoint serial OS processes: [233020,233265), [233265,233510), [233510,233755), [233755,234000). No CLI --workers or --sandbox; internal API workers=1 only. Harness uses Python standard library and the provided simulator, with no installed third-party dependency.", "scoreboard": "B remains incumbent at its required code hash. Diagnostic only; no candidate, tuning, paired candidate difference, finalist validation, or promotion.", "data_sha256_excluding_timing": "5eb886255f6483711c1659da09b9ccaf3a92f6ff85bb57f193e0d095b319868c", "timing_details": { "main_block_n": 980, "main_block_wall_seconds": 1483.3115148530342, "pilot_instrumented_n": 20, "pilot_instrumented_total_call_seconds": 29.04803106212057, "effective_wall_seconds_per_unique_measured_seed": 1.5123595459151546, "baseline_repeat_n": 20, "baseline_repeat_seconds": 27.44007290387526 } } }