{
  "item": "repos",
  "batch": "e07-public-search",
  "blocked": {
    "step": "scratch/setup.sh (uv download + amm_sim_rs wheel)",
    "error": "This machine is aarch64 (uname -m), with no python3, cargo, gcc or libc crt objects. setup.sh downloads uv-x86_64 and the x86_64-only wheel amm_sim_rs-0.1.0.post1-cp310-abi3-manylinux2014_x86_64; running uv failed with 'rosetta error: failed to open elf at /lib64/ld-linux-x86-64.so.2 / Trace/breakpoint trap'. The release sim-6b6921f8 at github.com/gtmcknight/amm-sim publishes only x86_64 wheels, and amm_sim_rs cannot be built here (no Rust toolchain, no linker). amm-match imports amm_sim_rs at module load, so `amm-match validate` could not run either. Consequence: n=0 for every paired measurement; d32, oog, top3 and code are empty. The research part (search + reading + quoting) is complete. As a weaker substitute for validate, every strategy file below was compiled with forge 1.8.3 / solc 0.8.26 under the challenge compiler settings (viaIR=true, optimizer runs=200, evmVersion=paris, from amm_competition/evm/compiler.py); see files[].forge_compiles.",
    "n": 0
  },
  "setup": {
    "checkout": "github.com/benedictbrady/amm-challenge @ 6b6921f8d463314aa7c96234e1be25a25245f704 (cloned into the workspace as amm-challenge/)",
    "rs2_sha256_verified": "f24b6ca198b744c380842c3a17964a6dc21f311bb04984b26733afa3fee8eaf2 (matches the brief; fetched from api.imd.fun artifact 19f2ec85…)",
    "rs2_forge_compiles": true,
    "rs2_deployed_bytecode_bytes": 10308
  },
  "sources": [
    {
      "url": "https://github.com/jiayaoqijia/amm-challenge-yq",
      "author": "jiayaoqijia (strategy getName 'yq')",
      "date": "2026-02-09 (last push per GitHub search API)",
      "holds": "Full strategy contracts/src/Strategy.sol (11-slot state-estimation fee rule), README with claimed official leaderboard score 523.21, RESULTS.md ablation study, research/papers.md with untried ideas (reserve-ratio inventory signal, sigma threshold regime switch, cumulative imbalance, exponential fee scaling). Not a fork of the challenge repo."
    },
    {
      "url": "https://github.com/houseofjiao/amm-challenge",
      "author": "houseofjiao (strategy getName 'category.xyz')",
      "date": "2026-02-12 (last push)",
      "holds": "Full copy of the challenge repo plus contracts/src/Strategy.sol: the yq architecture with retuned constants (SIGMA_DECAY 0.69, RET_CAP WAD/35, LAMBDA_CAP 1.82, DIR_IMPACT_MULT 4, GATE_SIGMA_MULT 7.2, TOX_CUBIC 20000 bps, TRADE_TOX_BOOST 3500 bps, STALE_ATTRACT_FRAC 1.05, TAIL_KNEE 600 bps, TAIL_SLOPE_ATTRACT 0.85, BASE_FEE 4 bps). README is the upstream README; no score claimed."
    },
    {
      "url": "https://github.com/0xniki/amm-challenge",
      "author": "0xniki (strategy getName 'L9')",
      "date": "2026-02-24 (last push)",
      "holds": "my_strategy.sol: 'Adaptive Toxicity-Based Fee Strategy', the yq architecture with slightly different constants (DIR_DECAY 0.82, TOX_DECAY 0.90, SIGMA_DECAY 0.83, PHAT_ALPHA 0.25/0.04, GATE_MIN 0.025, FLOW_COEF 4800, TOX_COEF 260, TOX_QUAD 12000, STALE_DIR 7000, STALE_ATTRACT 1.10). testing/ holds Python generators of earlier strategy families (vol-adaptive base fee from EMA of squared log-returns, reserve-skew, post-arb tightening, spike/decay). No score claimed in the repo text."
    },
    {
      "url": "https://github.com/lumen-dev-a11y/amm-challenge-lab",
      "author": "lumen-dev-a11y (strategy getName 'night_7')",
      "date": "2026-03-06 (last push)",
      "holds": "amm-challenge/my_strategy.sol: yq architecture plus a hard sigma regime switch on the base fee (SIGMA_HIGH_THRESHOLD 0.0011, PROTECTIVE_BASE_FEE 3 bps, COMPETITIVE_BASE_FEE 1 bps) and retuned cubic/boost/quad constants; amm-challenge/contracts/src/Strategy.sol is a fixed 55 bps baseline claiming '~379 edge / 200 sims'; notes/0x94305-principles.md summarises Alex Nezlobin tweets. No leaderboard score claimed."
    },
    {
      "url": "https://github.com/ylintern/ammchallenge",
      "author": "ylintern (strategy getName 'taMole_Reverso_milli')",
      "date": "2026-04-24 (last push)",
      "holds": "Strategy.sol only (plus a PNG): yq architecture with three commented tweaks (TOX_DECAY 0.895, PHAT_ALPHA 0.265, TOX_CUBIC 13300 bps, TAIL_SLOPE_PROTECT 0.875). No README, no score."
    },
    {
      "url": "https://github.com/dipankar/amm-challenge",
      "author": "dipankar",
      "date": "2026-10-25 (last push)",
      "holds": "33 strategy contracts, README with a local score table (SigmaAdaptive 526.54, MaxEdgeFinal 410.58, WinningV5 295.57 on 500 sims), ANALYSIS.md post-mortem contrasting 'trade classification' (spike/decay) with 'state estimation' (pHat) paradigms, research/ documents, experiments/. SigmaAdaptive is an explicit re-implementation of jiayaoqijia's yq. Only compiles with viaIR (stack too deep otherwise)."
    },
    {
      "url": "https://github.com/happybigmtn/ammchallenge-lab",
      "author": "happybigmtn (strategy getName 'PHatSkewPostSpotToxOnlyV4')",
      "date": "2026-02-12 (last push)",
      "holds": "~150 strategy .sol files from an evolutionary search, AGENTS.md with confirmed score 507.27 on the official 1000-seed set, per-seed comparison against the then live #1 (josusanmartin 'v351', avgEdge 527.0346), live-top telemetry (near-symmetric ~36 bps, spread 0.04–0.26 bps, arb volume 23.5k–24.0k Y) and a list of failed families with scores. Different mechanism family: fee-boundary arb detector + inverted skew + toxic-side-only mispricing premium."
    },
    {
      "url": "https://github.com/cardmanjonez/ammchallenge",
      "author": "cardmanjonez",
      "date": "2026-02-10 (last push)",
      "holds": "Only CLAUDE.md workflow template and empty tasks/ files. No strategy, no fee rule."
    },
    {
      "url": "https://github.com/banr1/tailored-amm-challenge",
      "author": "banr1",
      "date": "2026-02-07 (last push)",
      "holds": "Copy of the upstream repo with only the starter/vanilla contracts and the generic README. No own strategy."
    },
    {
      "url": "https://github.com/wanxinwanxin/pvp_amm_challenge",
      "author": "wanxinwanxin",
      "date": "2026-02-11 (last push)",
      "holds": "A player-vs-player web app built on the simulator (Streamlit, Railway). Contracts are only the upstream starter/vanilla files. No fee rule of its own."
    },
    {
      "url": "https://github.com/gtmcknight/amm-sim",
      "author": "gtmcknight",
      "date": "2026-11-08 (last push)",
      "holds": "Prebuilt x86_64 amm_sim_rs wheels (release sim-6b6921f8), harness.py, score.py, setup-fast.sh. Tooling only, no strategy."
    },
    {
      "url": "https://github.com/identity-md-launches/goal-3094-take-network-1st-paradigm-s",
      "author": "identity-md-launches (this network's own goal repo)",
      "date": "2026-11-07 (created and pushed)",
      "holds": "best/Strategy.sol, best/SCOREBOARD.md and 13 candidate families (band-state, bandit, fair-centered, flow-estimate, interval-tracker, inventory, public-replica, router-optimal, size-toxicity, static-grid, vol-estimate…) with NOTES.md each. Excluded from 'entrants' because it is this network's own repository, listed here for completeness only; not read further."
    }
  ],
  "mechanisms": [
    {
      "name": "pHat fair-price estimate with adaptive sigma-gated EMA update (yq)",
      "rule": "pImplied = isBuy ? spot*(1-feeUsed) : spot/(1-feeUsed); ret = |pImplied-pHat|/pHat; gate = max(sigmaHat*GATE_SIGMA_MULT, MIN_GATE) with GATE_SIGMA_MULT=10, MIN_GATE=0.03; if ret <= gate: pHat = (1-alpha)*pHat + alpha*pImplied, alpha = PHAT_ALPHA=0.26 on the first trade of a step, PHAT_ALPHA_RETAIL=0.05 otherwise. State: slots[5]=pHat, slots[6]=sigmaHat, slots[10]=stepTradeCount.",
      "quote": "uint256 adaptiveGate = wmul(sigmaHat, GATE_SIGMA_MULT);\n            if (adaptiveGate < MIN_GATE) adaptiveGate = MIN_GATE;\n            if (ret <= adaptiveGate) {\n                pHat = wmul(pHat, WAD - alpha) + wmul(pImplied, alpha);\n            }",
      "url": "https://raw.githubusercontent.com/jiayaoqijia/amm-challenge-yq/HEAD/contracts/src/Strategy.sol",
      "board": "README: 'Official leaderboard score: **523.21**'; local cross-seed mean 523.21 vs 522.13 baseline (+1.08). RESULTS.md ablation: removing improved_phat costs -0.42, the only feature outside noise."
    },
    {
      "name": "Sigma update on first-in-step only, capped return (yq)",
      "rule": "if firstInStep: ret = min(ret, RET_CAP=0.10); sigmaHat = SIGMA_DECAY*sigmaHat + (1-SIGMA_DECAY)*ret, SIGMA_DECAY=0.824. Initial sigmaHat 0.00095.",
      "quote": "if (firstInStep) {\n                if (ret > RET_CAP) ret = RET_CAP;\n                sigmaHat = wmul(sigmaHat, SIGMA_DECAY) + wmul(ret, WAD - SIGMA_DECAY);\n            }",
      "url": "https://raw.githubusercontent.com/jiayaoqijia/amm-challenge-yq/HEAD/contracts/src/Strategy.sol",
      "board": "part of the 523.21 strategy"
    },
    {
      "name": "Three-layer fee: base (sigma+lambda+flow) + toxicity polynomial (linear+quad+cubic) + activity + sigma×tox (yq)",
      "rule": "flowSize = lambdaHat*sizeHat; fBase = 3bps + 0.20*sigmaHat + 12bps*lambdaHat + 4842bps*flowSize; fMid = fBase + 250bps*tox + 11700bps*tox^2 + 91843bps*actEma + 500bps*sigmaHat*tox + 15000bps*tox^3, where tox = toxEma = 0.051*toxEma + 0.949*min(|spot-pHat|/pHat, 0.20). lambdaHat = 0.99*lambdaHat + 0.01*min(stepTradeCount/elapsed, 5) at each new step; sizeHat/actEma are EMAs of tradeRatio=amountY/reserveY (only when tradeRatio > WAD/500) with blends 0.818 and 0.985 and per-step decays 0.70^elapsed (elapsed capped at 8).",
      "quote": "uint256 fBase = BASE_FEE + wmul(SIGMA_COEF, sigmaHat) + wmul(LAMBDA_COEF, lambdaHat) + wmul(FLOW_SIZE_COEF, flowSize);\n        uint256 fMid = fBase + wmul(TOX_COEF, toxSignal) + wmul(TOX_QUAD_COEF, wmul(toxSignal, toxSignal)) + wmul(ACT_COEF, actEma);",
      "url": "https://raw.githubusercontent.com/jiayaoqijia/amm-challenge-yq/HEAD/contracts/src/Strategy.sol",
      "board": "523.21 (README)"
    },
    {
      "name": "Directional skew from decayed buy/sell pressure, weighted by toxicity (yq)",
      "rule": "dirState centered at WAD; push = min(tradeRatio*2, 0.25) added on buys, subtracted on sells (only if tradeRatio > WAD/500); decays toward WAD by 0.80^elapsed per new step. dirDev=|dirState-WAD|; skew = 20bps*dirDev + 100bps*dirDev*tox; the pressured side gets fMid+skew, the other fMid-skew.",
      "quote": "uint256 skew = wmul(DIR_COEF, dirDev) + wmul(DIR_TOX_COEF, wmul(dirDev, toxSignal));",
      "url": "https://raw.githubusercontent.com/jiayaoqijia/amm-challenge-yq/HEAD/contracts/src/Strategy.sol",
      "board": "523.21 (README)"
    },
    {
      "name": "Stale-price protection with asymmetric attract discount (yq)",
      "rule": "staleShift = 6850bps*tox; attractShift = staleShift*1.124; if spot >= pHat: bidFee += staleShift, askFee -= attractShift; else askFee += staleShift, bidFee -= attractShift.",
      "quote": "uint256 staleShift = wmul(STALE_DIR_COEF, toxSignal);\n            uint256 attractShift = wmul(staleShift, STALE_ATTRACT_FRAC);\n            if (spot >= pHat) {\n                bidFee = bidFee + staleShift;\n                askFee = askFee > attractShift ? askFee - attractShift : 0;",
      "url": "https://raw.githubusercontent.com/jiayaoqijia/amm-challenge-yq/HEAD/contracts/src/Strategy.sol",
      "board": "523.21 (README); RESULTS.md: 'Without asymmetric_tail | 522.48 | -0.01 | neutral'"
    },
    {
      "name": "Trade-aligned toxicity boost (yq)",
      "rule": "tradeAligned = (isBuy && spot >= pHat) || (!isBuy && spot < pHat); if aligned, add 2500bps*tradeRatio to the fee of the side just traded.",
      "quote": "bool tradeAligned = (trade.isBuy && spot >= pHat) || (!trade.isBuy && spot < pHat);\n            if (tradeAligned) {\n                uint256 tradeBoost = wmul(TRADE_TOX_BOOST, tradeRatio);",
      "url": "https://raw.githubusercontent.com/jiayaoqijia/amm-challenge-yq/HEAD/contracts/src/Strategy.sol",
      "board": "523.21 (README)"
    },
    {
      "name": "Asymmetric tail compression above a knee (yq)",
      "rule": "if fee > TAIL_KNEE=500bps: fee = 500bps + slope*(fee-500bps), slope 0.93 on the protect side, 0.955 on the attract side; then clampFee.",
      "quote": "function _compressTailWithSlope(uint256 fee, uint256 slope) internal pure returns (uint256) {\n        if (fee <= TAIL_KNEE) return fee;\n        return TAIL_KNEE + wmul(fee - TAIL_KNEE, slope);\n    }",
      "url": "https://raw.githubusercontent.com/jiayaoqijia/amm-challenge-yq/HEAD/contracts/src/Strategy.sol",
      "board": "523.21 (README)"
    },
    {
      "name": "Hard sigma regime switch on the base fee (lumen-dev-a11y 'night_7'; idea listed as untried in yq research/papers.md)",
      "rule": "baseFee = sigmaHat > 0.0011 ? 3bps : 1bps, substituted for BASE_FEE inside the yq fBase formula; other yq constants retuned (TOX_CUBIC 13662bps, TRADE_TOX_BOOST 2530bps, FLOW_SIZE 4779bps, TOX 315bps, TOX_QUAD 12228bps, ACT 94855bps, STALE_DIR 6666bps).",
      "quote": "uint256 baseFee = sigmaHat > SIGMA_HIGH_THRESHOLD ? PROTECTIVE_BASE_FEE : COMPETITIVE_BASE_FEE;",
      "url": "https://raw.githubusercontent.com/lumen-dev-a11y/amm-challenge-lab/HEAD/amm-challenge/my_strategy.sol",
      "board": "none claimed"
    },
    {
      "name": "Retuned yq constants (houseofjiao 'category.xyz')",
      "rule": "Same rule as yq with: SIGNAL_THRESHOLD WAD/650, DIR_DECAY 0.75, ACT/SIZE_DECAY 0.53, SIGMA_DECAY 0.69, SIZE_BLEND 0.828, PHAT_ALPHA 0.27/0.06, DIR_IMPACT_MULT 4, GATE_SIGMA_MULT 7.2, TOX_CUBIC 20000bps, TRADE_TOX_BOOST 3500bps, STALE_ATTRACT_FRAC 1.05, RET_CAP WAD/35, LAMBDA_CAP 1.82, BASE_FEE 4bps, TOX 300bps, TOX_QUAD 15000bps, TAIL_KNEE 600bps, TAIL_SLOPE_ATTRACT 0.85.",
      "quote": "uint256 constant SIGMA_DECAY = 690000000000000000;\n    uint256 constant LAMBDA_DECAY = 990000000000000000;\n    uint256 constant SIZE_BLEND_DECAY = 828000000000000000;",
      "url": "https://raw.githubusercontent.com/houseofjiao/amm-challenge/HEAD/contracts/src/Strategy.sol",
      "board": "none claimed"
    },
    {
      "name": "Fee-boundary arb detector + fee-boundary pHat inversion (happybigmtn 'PHatSkewPostSpotToxOnlyV4')",
      "rule": "Once per timestamp: gamma = 1-feeUsed; targetSpot = isBuy ? pHat/gamma : pHat*gamma; isArb = |spot-targetSpot|/targetSpot <= 55 bps. If arb: pHat = isBuy ? spot*gamma : spot/gamma; else pHat = (6*pHat + spot)/7. emaVolBps = (414*ema + relMoveBps)/415 (unused: VOL_MULT_X100=0). State: slots[0] lastTs, [1] pHat, [2] emaVolBps, [3] bidFee, [4] askFee.",
      "quote": "uint256 targetSpot = trade.isBuy ? wdiv(pHat, gamma) : wmul(pHat, gamma);\n            if (targetSpot == 0) targetSpot = spot;\n            uint256 errBps = wadToBps(wdiv(absDiff(spot, targetSpot), targetSpot));\n            bool isArb = (errBps <= ARB_SPOT_TOL_BPS);",
      "url": "https://raw.githubusercontent.com/happybigmtn/ammchallenge-lab/HEAD/amm_submissions/Strategy.sol",
      "board": "AGENTS.md: 'Confirmed score (official fixed seedset, 1000 seeds): avgEdge = 507.27', W/L/D 1000/0/0, deficit 19.77 vs live #1 527.0346"
    },
    {
      "name": "Inverted skew + toxic-side-only mispricing premium + toxic-side floor (happybigmtn)",
      "rule": "misBps = |spot-pHat|/pHat in bps; misAdj = max(misBps-18, 0); feeMid = 23bps + 0.18*tradeRatioBps; toxAdd = 0.19*misAdj; adj = min(600*misAdj/1000, 500bps). If spot >= pHat: bid = feeMid+adj+toxAdd, ask = feeMid-adj; else mirrored. Toxic side floored at TOX_MIN_BPS=32; safe-side cap disabled.",
      "quote": "if (spot >= pHat) {\n            // Spot high: deter sells-to-us (raise bid fee), attract buys-from-us (lower ask fee).\n            bidBps = feeMid + adj + toxAdd;\n            askBps = (feeMid > adj) ? (feeMid - adj) : MIN_BPS;",
      "url": "https://raw.githubusercontent.com/happybigmtn/ammchallenge-lab/HEAD/amm_submissions/Strategy.sol",
      "board": "507.27 (AGENTS.md); 'Inventory-skew orientation flip (non-inverted): severe degradation (~340 edge), so toxic-side inverted skew is critical.'"
    },
    {
      "name": "Live-top telemetry (happybigmtn observation of leaderboard entries, not a rule)",
      "rule": "Top entries near v351 average ~36 bps, bid/ask spread 0.04–0.26 bps, arb volume 23.5k–24.0k Y; their 507 strategy had spread ~2.93 bps and arb volume ~27.5k Y, deficit correlated +0.609 with own arb volume and growing with lambda.",
      "quote": "Top submissions around `v351` are near-symmetric around ~**36 bps** average with tiny average spread (~**0.04–0.26 bps** in top entries) and lower arb volume (~**23.5k–24.0k Y**).",
      "url": "https://raw.githubusercontent.com/happybigmtn/ammchallenge-lab/HEAD/AGENTS.md",
      "board": "live #1 on 2026-02-11: josusanmartin 'v351' avgEdge 527.0346"
    },
    {
      "name": "Trade-size spike/decay with EWMA arb threshold (dipankar WinningV5; the 'classification' paradigm)",
      "rule": "tradeRatio = amountY/reserveY; ewma = (5/6)*ewma + (1/6)*tradeRatio; isArb = tradeRatio > 2% || tradeRatio > 5*ewma -> both fees 65bps (+15bps on the traded side); tiny trades (<0.4%) -> 20bps; otherwise decay 7bps/step toward 24bps base; inventory protection +15bps when reserveX*100/reserveY leaves [0.7,1.3].",
      "quote": "bool isArb = tradeRatio > ARB_THRESHOLD || tradeRatio > ewma * 5;",
      "url": "https://raw.githubusercontent.com/dipankar/amm-challenge/HEAD/contracts/src/WinningV5Strategy.sol",
      "board": "README: WinningV5 295.57, MaxEdgeFinal 410.58, SigmaAdaptive (yq re-implementation) 526.54 on 500 local sims; the author notes local seeds differ from the server"
    },
    {
      "name": "Untried ideas listed by yq author (research/papers.md)",
      "rule": "(1) reserve-ratio inventory signal: deviation = |reserveY/reserveX - initialRatio|/initialRatio, fee += RESERVE_DEV_COEF*deviation; (2) sigma threshold regime switch; (3) cumulative signed-flow imbalance with decay, fee += IMBALANCE_COEF*|imbalance|; (4) adaptive shock gate (later adopted); (5) exponential fee scaling approximated by adding a cubic term (later adopted).",
      "quote": "- **Implementable formula**: fee* ≈ a(t) + b(t)*(y - y0)",
      "url": "https://raw.githubusercontent.com/jiayaoqijia/amm-challenge-yq/HEAD/research/papers.md",
      "board": "none; ideas only"
    }
  ],
  "files": [
    {"url": "https://raw.githubusercontent.com/jiayaoqijia/amm-challenge-yq/HEAD/contracts/src/Strategy.sol", "name": "yq", "sha256": "00fae418c49cd681ad0879c6f96ebed0163fab5a660b38030f11e6947a39afd4", "valid": null, "forge_compiles": true, "deployed_bytecode_bytes": 4379, "d32": null, "oog": null},
    {"url": "https://raw.githubusercontent.com/houseofjiao/amm-challenge/HEAD/contracts/src/Strategy.sol", "name": "category.xyz", "sha256": "07d5510da1972193d7cf20623de5693d54c123ce035ed158e2a7c1dee529bde0", "valid": null, "forge_compiles": true, "deployed_bytecode_bytes": 4381, "d32": null, "oog": null},
    {"url": "https://raw.githubusercontent.com/0xniki/amm-challenge/HEAD/my_strategy.sol", "name": "L9", "sha256": "74229c6535524db0bccc1116edd36e0952eb1ddb2acb4a25ec3da3f15e4806d9", "valid": null, "forge_compiles": true, "deployed_bytecode_bytes": 4379, "d32": null, "oog": null},
    {"url": "https://raw.githubusercontent.com/lumen-dev-a11y/amm-challenge-lab/HEAD/amm-challenge/my_strategy.sol", "name": "night_7", "sha256": "cc30bc7bceaf5afff5371f81bb4bcac1a8e50706a19eccd21ffff94642f1fcde", "valid": null, "forge_compiles": true, "deployed_bytecode_bytes": 4414, "d32": null, "oog": null},
    {"url": "https://raw.githubusercontent.com/lumen-dev-a11y/amm-challenge-lab/HEAD/amm-challenge/contracts/src/Strategy.sol", "name": "SimpleFixed55", "sha256": "9af0bc4976773e86c0b5e2970d3fe32c98a2a30075535bf5c7857555605c266a", "valid": null, "forge_compiles": true, "deployed_bytecode_bytes": 598, "d32": null, "oog": null},
    {"url": "https://raw.githubusercontent.com/ylintern/ammchallenge/HEAD/Strategy.sol", "name": "taMole_Reverso_milli", "sha256": "d6e7ab50ede8f5b8aadbfd69a8f966fb6a7bc50dbb57194c081a3233488d1ae7", "valid": null, "forge_compiles": true, "deployed_bytecode_bytes": 4420, "d32": null, "oog": null},
    {"url": "https://raw.githubusercontent.com/dipankar/amm-challenge/HEAD/contracts/src/SigmaAdaptiveStrategy.sol", "name": "SigmaAdaptive", "sha256": "84da373d4ecc47b5ecaa96adfcc432ad62825a0c608e7529404fa8a0184cdf90", "valid": null, "forge_compiles": true, "forge_note": "stack too deep without viaIR; compiles with viaIR=true as the challenge compiler uses", "deployed_bytecode_bytes": 3735, "d32": null, "oog": null},
    {"url": "https://raw.githubusercontent.com/dipankar/amm-challenge/HEAD/contracts/src/WinningV5Strategy.sol", "name": "WinningV5", "sha256": "dc8dff6827ecc6aadecfe9936edd4fa8b48fc402dc2fa5257503f0feb822bd05", "valid": null, "forge_compiles": true, "deployed_bytecode_bytes": 1479, "d32": null, "oog": null},
    {"url": "https://raw.githubusercontent.com/dipankar/amm-challenge/HEAD/contracts/src/MaxEdgeFinalStrategy.sol", "name": "MaxEdgeFinal", "sha256": "0a54f608fdcbd980077c0b552bb850db0ac2cbee5e730fd96b8e15ef0f3b674b", "valid": null, "forge_compiles": true, "deployed_bytecode_bytes": 1388, "d32": null, "oog": null},
    {"url": "https://raw.githubusercontent.com/happybigmtn/ammchallenge-lab/HEAD/amm_submissions/Strategy.sol", "name": "PHatSkewPostSpotToxOnlyV4", "sha256": "536d17d72770abfc256708571d34c44919f19215916c0f4a21865c1df7e7011b", "valid": null, "forge_compiles": true, "deployed_bytecode_bytes": 2446, "d32": null, "oog": null},
    {"url": "https://raw.githubusercontent.com/happybigmtn/ammchallenge-lab/HEAD/amm_submissions/Strategy_best_494.sol", "name": "happybigmtn best_494", "sha256": "3336170920c2b452df304e76fb945421c77bd72e4c406fb14cbb4d590f61da17", "valid": null, "forge_compiles": true, "forge_note": "stack too deep without viaIR; compiles with viaIR=true", "deployed_bytecode_bytes": 2272, "d32": null, "oog": null}
  ],
  "files_downloaded_not_compiled": [
    "dipankar/amm-challenge: 13 further experimental contracts (AsymmetricV2, Best78bps, GreedyEdge, GridSearch, Hybrid, MaxEdgeV2, Optimal, Optimized, RegimeV2, SmartAsymmetric, Static80bps, WinningV3, WinningV4), sha256 recorded locally; all are the author's own 'classification paradigm' which the README scores 295–410 locally.",
    "happybigmtn/ammchallenge-lab: PHatSkewBandInv_manual.sol, Strategy_best_450.sol, Strategy_best_494_60.sol, Strategy_oracle_direct_skew_best418.sol (earlier generations of the same family; AGENTS.md scores them below 507.27)."
  ],
  "top3": [],
  "code": null,
  "unmeasured_ranking_by_source_claims": [
    "jiayaoqijia yq (sha 00fae418…): the only file with a claimed official board score (523.21); every other state-estimation file found is a constant retune of it (houseofjiao, 0xniki, ylintern, lumen-dev-a11y, dipankar SigmaAdaptive).",
    "lumen-dev-a11y night_7 (sha cc30bc7b…): yq plus a hard sigma regime switch, the one structural change among the yq variants; no score claimed.",
    "happybigmtn PHatSkewPostSpotToxOnlyV4 (sha 536d17d7…): a different mechanism family (fee-boundary arb detector, inverted skew, toxic-side-only premium) but self-reported 507.27, i.e. about 15 below the 30 bps baseline's ~522, so unlikely to beat RS2."
  ],
  "notes": {
    "searched": "GitHub search/repositories API via the queries 'amm-challenge' (49 hits, 3 pages read), 'amm fee challenge' (5), 'optimization arena amm' (1), 'paradigm amm strategy' (1), 'amm challenge paradigm' (2), 'afterSwap fee strategy' (0), 'amm fee strategy afterSwap' (0), 'amm strategy afterSwap solidity' (0), '\"amm challenge\" fee' (1), 'ammchallenge' (5), 'AMMStrategyBase' (1), '\"amm\" \"fee strategy\" solidity' (0), 'dynamic fee amm challenge strategy' (0). Direct unauthenticated API calls from the shell were rate-limited (403, x-ratelimit-limit 10, shared egress IP), so searches were read through the fetch tool. The code-search endpoint (search/code) needs authentication and returned 401, so no code search was possible. Every non-fork hit whose name or description related to this challenge had its git tree listed; repos with no .sol beyond the upstream starter files were recorded as 'holds nothing'. Hits unrelated to the challenge (AMMO retail, 'ammissione' coding tests, Scaffold-ETH DEX challenges, Blueshift Solana AMM, Molecule challenge) were dropped. prop-amm-challenge repos (benedictbrady/prop-amm-challenge and 4 user copies) are the different quoting-mechanics variant and were not read.",
    "found": "Seven repositories with strategy code. Six of the seven state-estimation files are the same yq architecture (11 slots: bid, ask, lastTs, dirState, actEma, pHat, sigmaHat, lambdaHat, sizeHat, toxEma, stepTradeCount) with constants retuned; only lumen-dev-a11y adds a structural element (sigma regime switch), and only happybigmtn uses a different family. dipankar's ANALYSIS.md and README are secondary commentary on yq and name a Medium 'Algebra Finance write-up' that was not located on GitHub. The highest board score any GitHub source claims for itself is yq's 523.21; happybigmtn records the live #1 at 527.0346 (josusanmartin 'v351') on 2026-02-11 but holds none of that entrant's code.",
    "not_found": "No GitHub repository holding code or a written rule from any top-10 leaderboard entrant (e.g. josusanmartin v351) was found. No repository explains how the near-symmetric ~36 bps, sub-0.3 bps spread top entries work. No non-fork repository was found via the terms 'optimization arena' or 'paradigm amm strategy' other than this network's own goal repo and the wheel repo. Forks of benedictbrady/amm-challenge were out of this item's angle and were not enumerated.",
    "measurement": "Not performed: see blocked. No seed was run; no seed >= 5000000 was touched. forge_compiles is a compile-only check with solc 0.8.26 (the only aarch64 solc on this machine) under the challenge's viaIR/optimizer/paris settings and is not a substitute for `amm-match validate` (which also checks imports, contract declaration, reserved redeclarations and storage usage) or for gas_stats OOG counting.",
    "caution": "All downloaded sources were treated as data; no script from any repository was executed."
  }
}