{
  "approach": "Grafts online retail-demand learning into the base. A small joint Bayesian posterior infers retail arrival rate and notional size from surviving partial front asks, conditioning on full-fill censoring and tracking retained orders. Its estimates adapt inventory demand targets. The selected variant adapts exposure to the inferred demand. Public edge/code-size fingerprints motivated preserving retail capture while reducing adverse selection. The independent mechanism was grafted in the final half hour and selected on tuning scores before freezing for the final holdout.",
  "blocked": null,
  "python_version": "3.12.3",
  "base_mean_edge": 102.33197908899267,
  "score": {
    "mean_edge": 107.7367374084125,
    "retail_edge": 135.91459094566451,
    "arb_edge": -28.177853537266333,
    "simulations": 600,
    "seed_start": 1000,
    "failures": 0
  },
  "holdout": {
    "mean_edge": 96.97549954237483,
    "simulations": 600,
    "seed_start": 30000,
    "failures": 0
  },
  "code": "\"\"\"Stale-quote defence with tick probes, a sub-tick belief and cheap YES stock.\n\nEvery step the arbitrageur removes any order on the wrong side of p, so tiny\ntwo-sided probes around the estimate locate p to a tick; a grid filter narrows\nthat further.  Quotes sit one tick inside the competitor and are sized by the\nmarginal rule: one more share is worth quoting while the chance a retail order\nreaches it, times its edge, exceeds the expected arbitrage loss.  Near p=0 a\nretail buy pays almost its whole notional as edge, so YES is held to cover low\nasks.  That YES is bought when it is cheap: a large bid rests at tick 1 while p\nis above it, so the arbitrageur fills it on the crossing for about 0.1 tick a\nshare instead of nearly a full tick once p is near zero, and smaller bids sit on\nwhichever tick p is within 0.15 of.  Below tick 1 the estimate of p decays with\nthe time spent there, since no probe can see it.\n\"\"\"\nimport math\nfrom orderbook_pm_challenge.strategy import BaseStrategy\nfrom orderbook_pm_challenge.types import CancelOrder, PlaceOrder, Side\n\nP = dict(lam_r=0.25, m=5.0, sig_n=1.2, lam_j=0.0019, sig_j=0.35, sig=0.02,\n         qmax=3000.0, rmin=2, kA=0.7, kL=0.15, pL=0.05,\n         ammo=1.0, ydem=0.4, W=200.0, fy=0.6, ammo_q=60.0, ammo_pmax=0.05, ammo_maxpay=0.02,\n         fb=0.9, fu=0.7, xu=0.1, sticky=1.0, min_ahead=2.0, keep_hi=1.5, keep_lo=0.7, Hu=300.0, bins=8.0, filt=1.0, uni=0.01, jumpinfo=1.0, dmin=0.0, kLb=0.08, kLf=0.08, pinf=1.0, pre=0.0, H1=100.0,\n         ydf=0.6, Wf=2000.0, g_hi=0.15, g_lo=0.15, qf=400.0, pmaxf=0.05, pre_gap=1.0,\n         ydf1=0.6, g1=5.0, f1=0.6, a1=100.0, nb1=0.0, gs=1.0, p1t=1.0, m1=1.0, zd=1.0, pfl=0.0005)\nSQ2 = math.sqrt(2.0)\nSQ2PI = math.sqrt(2.0 * math.pi)\ndef cdf(x): return 0.5 * (1.0 + math.erf(x / SQ2))\ndef pdf(x): return math.exp(-0.5 * x * x) / SQ2PI\ndef inv(p):\n    p = min(max(p, 1e-9), 1 - 1e-9)\n    t = math.sqrt(-2.0 * math.log(min(p, 1 - p)))\n    x = t - (2.515517 + 0.802853 * t + 0.010328 * t * t) / (1 + 1.432788 * t + 0.189269 * t * t + 0.001308 * t ** 3)\n    x = -x if p < 0.5 else x\n    for _ in range(2):\n        d = pdf(x)\n        if d < 1e-12:\n            break\n        x -= (cdf(x) - p) / d\n    return x\ndef lpm(k):\n    return pdf(k) + k * cdf(k)\ndef fl(q):\n    return math.floor(q * 100 + 1e-9) / 100\n\nNODES = [(-2.2, .06), (-1.3, .17), (-0.6, .27), (0.6, .27), (1.3, .17), (2.2, .06)]\nPROBE = 0.01\n\n\nclass Strategy(BaseStrategy):\n    def __init__(self):\n        self.n = 0\n        self.ahead = set()     # main orders placed strictly ahead of any competitor order at their tick\n        self.res = {}          # order id -> uncovered quantity at placement\n        self.bc = None         # belief: sub-tick grid centres (ticks)\n        self.bw = None         # belief: weights\n        self.st = 0.5          # last one-step sd in ticks\n        self.pcb = self.pca = None\n        self.pb = set()\n        self.pa = set()\n        self.tb = 0\n        self.s = None          # competitor spread parameter\n\n    def _oid(self, tag):\n        self.n += 1\n        return tag + str(self.n)\n\n    def on_step(self, state):\n        P = self._fp_parameters\n        H = max(1, state.steps_remaining)\n        cb = state.competitor_best_bid_ticks or 0\n        ca = state.competitor_best_ask_ticks or 100\n        if self.s is None:\n            self.s = max(1, (ca - cb + 1) // 2)\n        lo, hi = cb, ca\n        for o in state.own_orders:\n            if o.side is Side.BUY:\n                lo = max(lo, o.price_ticks)\n            else:\n                hi = min(hi, o.price_ticks)\n        if P['pinf'] > 0:\n            sb = {o.price_ticks for o in state.own_orders if o.side is Side.BUY and o.remaining_quantity <= PROBE + 1e-9}\n            sa = {o.price_ticks for o in state.own_orders if o.side is Side.SELL and o.remaining_quantity <= PROBE + 1e-9}\n            ea = self.pa - sa\n            eb = self.pb - sb\n            if sa and ea:\n                lo = max(lo, min(max(ea), hi))\n            if eb and (sb or min(eb) <= P['p1t']):\n                hi = min(hi, max(min(eb), lo))\n        if P['jumpinfo'] > 0 and self.pcb is not None:\n            if ca > self.pca + 1:\n                lo = max(lo, ca - 1)\n            if cb < self.pcb - 1:\n                hi = min(hi, cb + 1)\n        self.pcb, self.pca = cb, ca\n        if lo > hi:\n            lo, hi = hi, lo\n        # sub-tick belief over p_(t-1): predict with one-step noise, truncate to [lo, hi]\n        wid = max(hi - lo, 1e-6)\n        nb = max(2, min(48, int(round(wid * P['bins']))))\n        cs = [lo + (i + 0.5) * wid / nb for i in range(nb)]\n        if self.bc is None or P['filt'] <= 0:\n            ws = [1.0 / nb] * nb\n        else:\n            sg = max(self.st, 0.6 * wid / nb)\n            k = -0.5 / (sg * sg)\n            ws = []\n            uni = P['uni'] * sg\n            for c in cs:\n                t = uni\n                for ci, wi in zip(self.bc, self.bw):\n                    d = c - ci\n                    d2 = d * d * k\n                    if d2 > -12.5:\n                        t += wi * math.exp(d2)\n                ws.append(t)\n            tot = sum(ws)\n            ws = [w / tot for w in ws]\n        self.bc, self.bw = cs, ws\n        mean_t = sum(c * w for c, w in zip(cs, ws))\n        var_t = max(sum(c * c * w for c, w in zip(cs, ws)) - mean_t * mean_t, 0.0)\n        phat = min(max(mean_t / 100.0, 0.002), 0.998)\n        var = P['sig'] ** 2 + P['lam_j'] * P['sig_j'] ** 2\n        sd = math.sqrt(H * var)\n        self.tb = self.tb + 1 if hi <= 1 else 0\n        if self.tb > 0 and P['zd'] > 0:\n            # below the last tick there is no probe: the score keeps walking away from it\n            phat = min(phat, max(P['pfl'], cdf(inv(0.01) - P['zd'] * 0.02 * math.sqrt(self.tb) / sd)))\n        z = inv(phat)\n        Z = z * sd\n        sp = pdf(z) / sd * P['sig']\n        self.st = sp * 100.0\n        s = math.sqrt(sp * sp + var_t / 1e4 + 1e-8)\n        sd1 = math.sqrt(max(1.0, H - 1.0) * var)\n        jp = [(cdf((Z + n * P['sig_j']) / sd1), w) for n, w in NODES]\n        lj = P['lam_j']\n\n        def AB(x):\n            k = (x - phat) / s\n            a = s * lpm(k)\n            b = a - (x - phat)\n            ja = sum(w * max(0.0, x - p) for p, w in jp)\n            jb = sum(w * max(0.0, p - x) for p, w in jp)\n            return (1 - lj) * a + lj * ja, (1 - lj) * b + lj * jb\n\n        if H == 1:\n            def AB(x):\n                return (1 - phat) * x, phat * (1 - x)\n        rho = P['lam_r'] * 0.5\n        mu = math.log(P['m']) - 0.5 * P['sig_n'] ** 2\n        kA = P['kL'] if phat < P['pL'] else P['kA']\n\n        def qsize(loss, gain, px):\n            loss *= kA\n            if gain <= 0:\n                return 0.0\n            tail = loss / (rho * gain)\n            if tail >= 1.0:\n                return 0.0\n            return min(P['qmax'], math.exp(mu + P['sig_n'] * inv(1 - max(tail, 1e-6))) / px)\n\n        ptk = phat * 100\n        def qbid(t):\n            if not 1 <= t <= 99:\n                return 0.0\n            a, b = AB(t / 100.0)\n            if phat * 100 - t < P['dmin']:\n                return 0.0\n            if P['nb1'] > 0 and self.s == 1 and t == 1 and ptk < 1.0:\n                return 0.0\n            if phat < P['pL']:\n                ytg = P['ydem'] * min(H, P['W']) / max(ca - 1, ptk + 1) * 100.0\n                a *= (P['kLb'] if state.yes_inventory < ytg else P['kLf']) / P['kL']\n            return qsize(a, b, max(phat, 0.05))\n\n        def qask(t):\n            if not 1 <= t <= 99:\n                return 0.0\n            a, b = AB(t / 100.0)\n            if t - phat * 100 < P['dmin']:\n                return 0.0\n            return qsize(b, a, t / 100.0)\n\n        bids = [o for o in state.own_orders if o.side is Side.BUY and o.remaining_quantity > PROBE + 1e-9]\n        asks = [o for o in state.own_orders if o.side is Side.SELL and o.remaining_quantity > PROBE + 1e-9]\n        sticky = P['sticky'] > 0\n        def top(orders, comp, step):\n            if sticky and sum(o.remaining_quantity for o in orders if o.price_ticks == comp and o.order_id in self.ahead) >= P['min_ahead']:\n                return comp\n            return comp + step\n        bid_t = top(bids, cb, 1)\n        ask_t = top(asks, ca, -1)\n        qb = qbid(bid_t)\n        qa = qask(ask_t)\n\n        Y = state.yes_inventory\n        if ask_t == 1 and H > P['H1'] and P['pre'] > 0:\n            qa = qa if Y >= P['a1'] > 0 else 0.0          # let the competitor's 2-tick ask erode instead of fronting it for nothing\n        # ammunition: hold YES so that low asks are covered\n        ptick = phat * 100\n        amt = 0\n        qam = 0.0\n        if P['ammo'] > 0 and ask_t >= 2 and phat < max(P['ammo_pmax'], P['pmaxf']):\n            wealth = state.cash + Y * phat\n            x = max(ask_t, ptick + 1) / 100.0\n            capw = P['fy'] * wealth / max(phat, 0.005)\n            target = min(P['ydem'] * min(H, P['W']) / x, capw)\n            full = min(P['ydf'] * min(H, P['Wf']) / x, capw) if phat < P['pmaxf'] else 0.0\n            t = max(1, min(99, int(math.ceil(ptick + 1e-9))))\n            g = t - ptick\n            if Y < full and g <= P['g_hi'] and bid_t <= t < ask_t:\n                amt, qam = t, min(P['qf'], full - Y)     # a tick p has just crossed: nearly free\n            elif Y < target and phat < P['ammo_pmax'] and bid_t <= t < ask_t and t / 100.0 - phat < P['ammo_maxpay']:\n                amt, qam = t, min(P['ammo_q'], target - Y)\n            elif Y < full and t >= 2 and 1.0 - g <= P['g_lo']:\n                amt, qam = t - 1, min(P['qf'], full - Y)  # rest just under p: filled only on the crossing, or by retail\n\n        if P['pre'] > 0 and ca == 2 and phat < 0.01 and H > P['H1'] and Y < P['pre'] and not qam and 1.0 - ptick < P['pre_gap']:\n            amt, qam = 1, min(P['ammo_q'], P['pre'] - Y)\n        q1 = 0.0\n        if P['ydf1'] > 0 and P['m1'] < ptick < P['g1'] + 1.0 and amt != 1:\n            q1 = max(0.0, P['ydf1'] * min(H, P['Wf']) / (self.s / 100.0) - Y)\n        # order management\n        actions = []\n        free = state.free_cash - 2.0\n        keep_ask = 0.0\n        new_b = []\n        new_a = []\n        def cancel(o):\n            nonlocal free\n            actions.append(CancelOrder(o.order_id))\n            x = o.price_ticks / 100.0\n            if o.side is Side.BUY:\n                free += x * o.remaining_quantity\n            else:\n                free += (1 - x) * min(self.res.get(o.order_id, 0.0), o.remaining_quantity)\n            self.ahead.discard(o.order_id)\n            self.res.pop(o.order_id, None)\n        for o in state.own_orders:\n            if o.remaining_quantity <= PROBE + 1e-9:\n                cancel(o)\n        kept_ids = set()\n        def manage(orders, t, q, comp):\n            kept = 0.0\n            for o in orders:\n                if o.price_ticks == t and q > 0 and kept + o.remaining_quantity <= max(q * P['keep_hi'], 1.0):\n                    kept += o.remaining_quantity\n                    kept_ids.add(o.order_id)\n                else:\n                    cancel(o)\n            return kept\n        kb = manage(bids, bid_t, qb, cb)\n        ka = manage(asks, ask_t, qa, ca)\n        add_b = (qb - kb) if (kb < qb * P['keep_lo'] and bid_t != cb) else 0.0\n        add_a = (qa - ka) if (ka < qa * P['keep_lo'] and ask_t != ca) else 0.0\n\n        budget = P['fb'] * max(free, 0.0)\n        cost = add_b * bid_t / 100.0 + qam * amt / 100.0\n        if cost > budget:\n            f = budget / cost\n            add_b *= f\n            qam *= f\n            cost = budget\n        out = []\n        q1 = min(q1, P['f1'] * max(budget - cost, 0.0) * 100.0)\n        if q1 >= 0.01:\n            out.append(PlaceOrder(side=Side.BUY, price_ticks=1, quantity=fl(q1)))\n            free -= fl(q1) / 100.0 + 1e-6\n        if amt and qam >= 0.01:\n            q = fl(qam)\n            out.append(PlaceOrder(side=Side.BUY, price_ticks=amt, quantity=q))\n            free -= q * amt / 100.0 + 1e-6\n        if bid_t >= 1 and add_b >= 0.01:\n            q = fl(add_b)\n            oid = self._oid('b')\n            out.append(PlaceOrder(side=Side.BUY, price_ticks=bid_t, quantity=q, client_order_id=oid))\n            free -= q * bid_t / 100.0 + 1e-6\n            if bid_t > cb:\n                self.ahead.add(oid)\n        if ask_t <= 99 and add_a >= 0.01:\n            x = ask_t / 100.0\n            covk = sum(o.remaining_quantity - min(self.res.get(o.order_id, 0.0), o.remaining_quantity)\n                       for o in asks if o.order_id in kept_ids)\n            cov = max(0.0, Y - covk - 0.1)\n            unc = max(0.0, add_a - cov)\n            if unc > 0:\n                ub = max(0.0, free - (1 - P['fu']) * max(state.cash, 0.0)) if x >= P['xu'] else 0.0\n                if H < P['Hu'] and phat < 0.05:\n                    ub = max(0.0, free * (1 - H / P['Hu']))\n                unc = min(unc, ub / (1 - x))\n            q = fl(min(add_a, cov + unc))\n            if q >= 0.01:\n                oid = self._oid('a')\n                out.append(PlaceOrder(side=Side.SELL, price_ticks=ask_t, quantity=q, client_order_id=oid))\n                r = max(0.0, q - cov)\n                self.res[oid] = r\n                free -= r * (1 - x) + 1e-6\n                if ask_t < ca:\n                    self.ahead.add(oid)\n\n        # probes: tiny orders that the arbitrageur removes when p crosses them\n        f = int(math.floor(ptick + 1e-9))\n        r = max(int(P['rmin']), int(math.ceil(2.0 * s * 100)))\n        self.pb = set()\n        self.pa = set()\n        if free > 1.0:\n            for t in range(f, f + r + 1):\n                if cb < t < ca and 1 <= t <= 99:\n                    out.append(PlaceOrder(side=Side.BUY, price_ticks=t, quantity=PROBE))\n                    self.pb.add(t)\n            for t in range(f + 1 - r, f + 2):\n                if cb < t < ca and 1 <= t <= 99:\n                    out.append(PlaceOrder(side=Side.SELL, price_ticks=t, quantity=PROBE))\n                    self.pa.add(t)\n        return actions + out\n\nimport math\nfrom orderbook_pm_challenge.types import PlaceOrder,Side,CancelOrder,CancelAll\nclass _FPRetailLearner:\n    def __init__(self):\n        self.pending=None;self.events=0.;self.exposure=0.\n        self.nodes=[]\n        self.caps=[2.,2.5,3.,4.,5.,6.,7.]+[8.*math.exp(.2*j) for j in range(32)]\n        for i in range(9):\n            arrival=.154+(.352-.154)*(i+.5)/9.;r=arrival*.5;p0=math.exp(-r)\n            for j in range(13):\n                mean=2.64+(6.336-2.64)*(j+.5)/13.;terms=[];weight=p0\n                for k in range(1,5):\n                    weight*=r/k;variance=math.log(1.+math.expm1(1.44)/k)\n                    mu=math.log(k*mean)-.5*variance\n                    terms.append((weight,mu,math.sqrt(variance)))\n                self.nodes.append((arrival,mean,p0,terms))\n        self.weights=[1./len(self.nodes)]*len(self.nodes)\n        self.cdfs=[]\n        for cap in self.caps:\n            logcap=math.log(cap)\n            self.cdfs.append([p0+sum(w*.5*math.erfc(-(logcap-mu)/(sig*math.sqrt(2.))) for w,mu,sig in terms) for _,_,p0,terms in self.nodes])\n        self.prior_arrival=.253;self.prior_mean=4.488\n        self._arrival=self.prior_arrival;self._mean=self.prior_mean;self._flow=1.135464\n    def update(self,s):\n        if self.pending is None:return\n        step,tick,q=self.pending;self.pending=None\n        if s.step!=step+1:return\n        remaining=sum(o.remaining_quantity for o in s.own_orders if o.side==Side.SELL and o.price_ticks==tick)\n        if remaining<.005 or remaining>q+.005:return\n        cap=q*tick/100.;value=max(0.,(q-remaining)*tick/100.)\n        self.exposure+=1.\n        if cap>=8.:\n            idx=max(0.,min(30.99999,math.log(cap/8.)/.2));j=int(idx)+7;frac=idx-int(idx)\n        else:\n            j=next((i for i in range(7) if cap<=self.caps[i+1]),6)\n            frac=max(0.,min(1.,math.log(cap/self.caps[j])/math.log(self.caps[j+1]/self.caps[j])))\n        ga=self.cdfs[j];gb=self.cdfs[j+1];posterior=[]\n        positive=value>.02\n        if positive:self.events+=1.;logvalue=math.log(value)\n        for i,(arrival,mean,p0,terms) in enumerate(self.nodes):\n            g=max(1e-12,ga[i]*(1.-frac)+gb[i]*frac)\n            if positive:\n                density=sum(w*math.exp(-.5*((logvalue-mu)/sig)**2)/sig for w,mu,sig in terms)\n                likelihood=density/g\n            else:likelihood=p0/g\n            posterior.append(self.weights[i]*likelihood)\n        total=sum(posterior)\n        if total<=0:return\n        self.weights=[w/total for w in posterior]\n        self._arrival=sum(w*n[0] for w,n in zip(self.weights,self.nodes))\n        self._mean=sum(w*n[1] for w,n in zip(self.weights,self.nodes))\n        self._flow=sum(w*n[0]*n[1] for w,n in zip(self.weights,self.nodes))\n    def observe(self,s,actions):\n        resting={o.order_id:(o.price_ticks,o.remaining_quantity) for o in s.own_orders if o.side==Side.SELL}\n        for i,a in enumerate(actions):\n            if isinstance(a,CancelAll):resting.clear()\n            elif isinstance(a,CancelOrder):resting.pop(a.order_id,None)\n            elif isinstance(a,PlaceOrder) and a.side==Side.SELL:\n                resting[a.client_order_id if a.client_order_id is not None else ('new',i)]=(a.price_ticks,math.floor(a.quantity*100.+1e-7)/100.)\n        normal=[v for v in resting.values() if v[1]>=1.]\n        self.pending=None\n        if not normal:return\n        tick=min(v[0] for v in normal)\n        if s.competitor_best_ask_ticks is not None and tick>=s.competitor_best_ask_ticks:return\n        total=sum(q for p,q in resting.values() if p==tick)\n        if total*tick/100.+1e-9<2.:return\n        self.pending=(s.step,tick,total)\n    def arrival_rate(self):return self._arrival\n    def mean_notional(self):return self._mean\n    def flow_ratio(self):return self._flow/(self.prior_arrival*self.prior_mean)\n\n_FPBaseStrategy=Strategy\nclass Strategy(_FPBaseStrategy):\n    def __init__(self):\n        super().__init__()\n        self._fp_retail=_FPRetailLearner()\n        self._fp_original=dict(P)\n        self._fp_parameters=dict(P)\n    def on_step(self,state):\n        self._fp_retail.update(state)\n        params=dict(self._fp_original)\n        ratio=self._fp_retail.flow_ratio()**1.0\n        for key in ['ydem','ydf','ydf1']:params[key]*=ratio\n        self._fp_parameters=params\n        actions=super().on_step(state)\n        self._fp_retail.observe(state,actions)\n        return actions\n",
  "code_sha256": "aafa7ec9047fe9332d376e2daf445f9963d2c3d3fbf135f51a99ba35736a35d6",
  "tried": [
    "Public fingerprints: KING retail 130.771634 / arb -21.17949 / codeLength 134648; Littoral v16 retail 127.734891 / arb -18.307675 / codeLength 59197. Similar retail capture but smaller arb losses suggests selective exposure; code cannot be inferred uniquely from these aggregate numbers.",
    "Oracle isolated-flow restriction all: 159.68544079506526 mean potential edge on 600 regimes, seeds 1000..1599. Next p and retail side/size revealed; unlimited collateral and fixed no-participant competitor path. These overlapping relaxations are not feasible global strategy bounds.",
    "Oracle isolated-flow restriction buy_side: 30.946752622636794 mean potential edge on 600 regimes, seeds 1000..1599. Next p and retail side/size revealed; unlimited collateral and fixed no-participant competitor path. These overlapping relaxations are not feasible global strategy bounds.",
    "Oracle isolated-flow restriction sell_side: 128.73868817242845 mean potential edge on 600 regimes, seeds 1000..1599. Next p and retail side/size revealed; unlimited collateral and fixed no-participant competitor path. These overlapping relaxations are not feasible global strategy bounds.",
    "Oracle isolated-flow restriction late_500: 83.77756868764912 mean potential edge on 600 regimes, seeds 1000..1599. Next p and retail side/size revealed; unlimited collateral and fixed no-participant competitor path. These overlapping relaxations are not feasible global strategy bounds.",
    "Oracle isolated-flow restriction extreme_p: 117.01970729220618 mean potential edge on 600 regimes, seeds 1000..1599. Next p and retail side/size revealed; unlimited collateral and fixed no-participant competitor path. These overlapping relaxations are not feasible global strategy bounds.",
    "Oracle isolated-flow restriction middle_p: 42.66573350285909 mean potential edge on 600 regimes, seeds 1000..1599. Next p and retail side/size revealed; unlimited collateral and fixed no-participant competitor path. These overlapping relaxations are not feasible global strategy bounds.",
    "Oracle isolated-flow restriction large_retail_20: 45.3794263394823 mean potential edge on 600 regimes, seeds 1000..1599. Next p and retail side/size revealed; unlimited collateral and fixed no-participant competitor path. These overlapping relaxations are not feasible global strategy bounds.",
    "Oracle isolated-flow restriction outside_competitor: 25.340030564676113 mean potential edge on 600 regimes, seeds 1000..1599. Next p and retail side/size revealed; unlimited collateral and fixed no-participant competitor path. These overlapping relaxations are not feasible global strategy bounds.",
    "Oracle isolated-flow restriction wide_edge_5c: 5.857029006498238 mean potential edge on 600 regimes, seeds 1000..1599. Next p and retail side/size revealed; unlimited collateral and fixed no-participant competitor path. These overlapping relaxations are not feasible global strategy bounds.",
    "Oracle isolated-flow restriction depletion_no_competitor: 0.7213729237863137 mean potential edge on 600 regimes, seeds 1000..1599. Next p and retail side/size revealed; unlimited collateral and fixed no-participant competitor path. These overlapping relaxations are not feasible global strategy bounds.",
    "Oracle isolated-flow restriction wide_competitor_gap: 35.42837898653695 mean potential edge on 600 regimes, seeds 1000..1599. Next p and retail side/size revealed; unlimited collateral and fixed no-participant competitor path. These overlapping relaxations are not feasible global strategy bounds.",
    "Oracle isolated-flow restriction small_retail_5: 46.36070601749343 mean potential edge on 600 regimes, seeds 1000..1599. Next p and retail side/size revealed; unlimited collateral and fixed no-participant competitor path. These overlapping relaxations are not feasible global strategy bounds.",
    "Standalone same-price FIFO queue-tail policy: mean -28.127411968373167, retail 4.360839133688167, arb -32.4882511020625 on all 600 tuning simulations, zero failures; discarded.",
    "Oracle current-probability version of the fair-value quote optimizer: -8.030296320276; oracle next-probability version: 24.289285273587 with zero arbitrage loss. Both 100 simulations starting 1000, zero failures, and still exclude the three reserved moves. Knowledge of current value alone does not remove next-step adverse selection.",
    "Own-order ID counter reveals delayed competitor births exactly; checked against 39980 step observations over 20 training simulations. However adding this feature did not help the standalone policy: all-600 mean 2.8810563650361667 with it versus 2.9827307612880003 without it; both zero failures. Dropped from the leading mechanism.",
    "Online retail-demand calibration: partial, surviving front asks identify retail; fully removed quotes are censored. With real latent paths and informed flow but an oracle current-p front ask, no competitor and relaxed cash, all 600 training regimes yielded demand-ratio MSE 0.011345173542956504 after 2000 steps versus prior 0.11500179685725608. This is an estimator test, not a challenge score. Priors match the published parameter bands.",
    "Applying learned demand inside the restricted standalone quote optimizer: mean 3.3940071470626667, retail 4.027929128698667, arb -0.633921981636, all 600 training simulations, zero failures; below the fixed-demand version.",
    "Forward endpoint duration model: fit seeds 1000..1399, validate 1400..1599, sampled every four steps. On 13098 low-price observations from 115 validation regimes, long-history duration-fraction MSE 0.05508959634845679 versus 0.0601740007435476 with short history; regime-bootstrap 95% difference [-0.007515231506387663,-0.002661354378974306]. Jump overshoot prediction failed to beat its prior and was discarded. Runtime long-history features match 300000 training rows within float32 rounding.",
    "Starter sandbox check: 20/20 infrastructure failures on Python 3.12.3: ImportError: Import of ntpath is not allowed in sandbox mode.",
    "Final independent retail learner: joint discrete posterior for arrival rate and mean notional, conditioned on partial front-ask survival. Corrected-scale MSE 0.010133596405820738 at $30 cap and 0.017277942106544028 at $8.5 cap after 2000 steps on 600 training regimes; lower than moment-update errors. Calibration uses oracle current-p asks, relaxed cash, and no competitor; not a challenge score.",
    "Independent planning benchmark, 600 training retail regimes, 1000-period immediate-replenishment model: jump-aware inventory DP 134.35817104141876 versus remaining-horizon quantile 130.5574893502864 at exit hazard .001 / loss 1; at hazard .003 / loss 5, DP 26.71577543376294 versus -92.76391759055174. Planning-model edge units, not challenge scores; actual covered asks have a one-step reservation delay omitted here.",
    "Best fresh standalone checkpoint: stop-loss-moment boosted forecast and utility optimizer, selected risk 1.4 and size 2.0 on seeds 1000..1099; all-600 mean 3.418838039699, retail 4.047874850646167, arb -0.6290368109471667, zero failures.",
    "Selected mechanism diagnostic: dyn_stock improves the paired tuning mean by 0.4925281846609999; paired standard error 0.4342523383708021. Gains are 0.10042114632425013 on the first 400 and 1.2767422613344994 on the last 200. The measured improvement is small and statistically uncertain; no private-holdout outcome was used in selection.",
    "Oracle true retail parameters in downward base inventory-target scaling: mean 133.913797038642, paired gain 2.062657746196 on 100 simulations starting 1000, zero failures. This is unavailable-information research, not a candidate or a global optimal ceiling.",
    "Executable oracle current_value: mean 5.947295229041667, retail 8.380433894151667, arb -2.43313866512; 60 simulations starting 1000, zero failures. Scratch simulator reveals unavailable information; fresh-lane price restrictions retained. This is an oracle research policy, not a candidate or a global optimal bound.",
    "Executable oracle next_value: mean 8.814029771145, retail 8.814029771145, arb 0.0; 60 simulations starting 1000, zero failures. Scratch simulator reveals unavailable information; fresh-lane price restrictions retained. This is an oracle research policy, not a candidate or a global optimal bound.",
    "Executable oracle retail_timing: mean 21.21442297759, retail 21.21442297759, arb 0.0; 60 simulations starting 1000, zero failures. Scratch simulator reveals unavailable information; fresh-lane price restrictions retained. This is an oracle research policy, not a candidate or a global optimal bound.",
    "Executable oracle retail_size: mean 34.04427242990833, retail 34.04427242990833, arb 0.0; 60 simulations starting 1000, zero failures. Scratch simulator reveals unavailable information; fresh-lane price restrictions retained. This is an oracle research policy, not a candidate or a global optimal bound.",
    "Executable oracle buy_only: mean 15.444066904041668, retail 15.444066904041668, arb 0.0; 60 simulations starting 1000, zero failures. Scratch simulator reveals unavailable information; fresh-lane price restrictions retained. This is an oracle research policy, not a candidate or a global optimal bound.",
    "Executable oracle sell_only: mean 13.306047192643332, retail 13.306047192643332, arb 0.0; 60 simulations starting 1000, zero failures. Scratch simulator reveals unavailable information; fresh-lane price restrictions retained. This is an oracle research policy, not a candidate or a global optimal bound.",
    "Executable oracle late_only: mean 12.880957932434999, retail 12.880957932434999, arb 0.0; 60 simulations starting 1000, zero failures. Scratch simulator reveals unavailable information; fresh-lane price restrictions retained. This is an oracle research policy, not a candidate or a global optimal bound.",
    "Executable oracle covered_recycling: mean 30.979491247816668, retail 30.979491247816668, arb 0.0; 60 simulations starting 1000, zero failures. Scratch simulator reveals unavailable information; fresh-lane price restrictions retained. This is an oracle research policy, not a candidate or a global optimal bound.",
    "Executable oracle queue_tail: mean 18.091799823943333, retail 18.091799823943333, arb 0.0; 60 simulations starting 1000, zero failures. Scratch simulator reveals unavailable information; fresh-lane price restrictions retained. This is an oracle research policy, not a candidate or a global optimal bound.",
    "Executable oracle wide_margin: mean 4.4357666400983335, retail 4.4357666400983335, arb 0.0; 60 simulations starting 1000, zero failures. Scratch simulator reveals unavailable information; fresh-lane price restrictions retained. This is an oracle research policy, not a candidate or a global optimal bound.",
    "Executable oracle empty_book: mean 0.042790022988333336, retail 0.042790022988333336, arb 0.0; 60 simulations starting 1000, zero failures. Scratch simulator reveals unavailable information; fresh-lane price restrictions retained. This is an oracle research policy, not a candidate or a global optimal bound.",
    "Executable oracle best_side: mean 33.430798601056665, retail 33.430798601056665, arb 0.0; 60 simulations starting 1000, zero failures. Scratch simulator reveals unavailable information; fresh-lane price restrictions retained. This is an oracle research policy, not a candidate or a global optimal bound.",
    "Joint discrete Bayesian retail learner uses conditional likelihood given a surviving ask, including compound arrivals and censored full fills. At $8.5 cap, real-price/informed-flow calibration MSE 0.017277942106544028 versus moment update 0.02121605112132361 after 2000 steps; these are calibration metrics, not edge scores.",
    "Duration-model covariate-shift diagnostic under an oracle front-ask policy on seeds 1400..1599: MSE 0.05480647631081781 over 13127 low-price observations; relaxed collateral, not a challenge score.",
    "Graft screen identity: mean 104.24257718317126, paired gain 0.0 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen retail_half: mean 104.257136946455, paired gain 0.014559763283750731 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen retail_one: mean 104.30144952718373, paired gain 0.058872344012498634 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen retail_two: mean 103.9156732833025, paired gain -0.32690389986875124 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen quantile_life_double: mean 101.49588418875875, paired gain -2.746692994412501 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dp_0003: mean 86.28580781472125, paired gain -17.95676936845 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dp_life: mean 101.52908499063125, paired gain -2.71349219254 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dp_life_exit5: mean 101.5769205933825, paired gain -2.66565658978875 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dyn_size: mean 102.15916591310625, paired gain -2.0834112700650005 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dyn_stock: mean 104.18452680707875, paired gain -0.058050376092500876 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dyn_both: mean 101.429065384335, paired gain -2.813511798836251 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dyn_half: mean 103.92167225303001, paired gain -0.32090493014125065 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dyn_down: mean 104.45337081346625, paired gain 0.2107936302949999 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dyn_size_down: mean 104.24252185679624, paired gain -5.532637500122589e-05 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dyn_both_down: mean 104.44062355354625, paired gain 0.1980463703749991 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dyn_flow_down: mean 104.34246677606875, paired gain 0.0998895928975003 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dyn_flow_half: mean 104.441105861485, paired gain 0.1985286783137505 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dyn_mean_down: mean 103.94295428378125, paired gain -0.2996228993900008 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen refill_full: mean 104.36495497609, paired gain 0.12237779291874937 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen refill_life: mean 104.36756861718375, paired gain 0.12499143401249913 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen refill_prior: mean 104.36495497609, paired gain 0.12237779291874937 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen refill_dp: mean 104.33605795291251, paired gain 0.09348076974124955 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dyn_down_front: mean 104.41014823041375, paired gain 0.16757104724249966 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Graft screen dyn_flow_front: mean 104.29108786139125, paired gain 0.048510678219999916 on 80 tuning simulations (1000..1039 and 1400..1439), zero failures.",
    "Full tuning comparison base: {\"mean_edge\":107.2442092237515,\"retail_edge\":135.80165652288784,\"arb_edge\":-28.557447299150834,\"simulations\":600,\"seed_start\":1000,\"failures\":0}",
    "Full tuning comparison dyn_down: {\"mean_edge\":107.42664272657133,\"retail_edge\":135.4395488645935,\"arb_edge\":-28.0129061380385,\"simulations\":600,\"seed_start\":1000,\"failures\":0}",
    "Full tuning comparison retail_one: {\"mean_edge\":106.87894523519216,\"retail_edge\":135.1021852375155,\"arb_edge\":-28.223240002340003,\"simulations\":600,\"seed_start\":1000,\"failures\":0}",
    "Full tuning comparison refill_life: {\"mean_edge\":107.15547425094734,\"retail_edge\":135.62113502768668,\"arb_edge\":-28.465660776754667,\"simulations\":600,\"seed_start\":1000,\"failures\":0}",
    "Full tuning comparison refill_combo: {\"mean_edge\":107.35585322918233,\"retail_edge\":135.2642079550175,\"arb_edge\":-27.908354725851666,\"simulations\":600,\"seed_start\":1000,\"failures\":0}",
    "Full tuning comparison dyn_upper: {\"mean_edge\":107.41035937797967,\"retail_edge\":135.79364551167367,\"arb_edge\":-28.383286133708335,\"simulations\":600,\"seed_start\":1000,\"failures\":0}",
    "Final sandbox: 20/20 fail before strategy loading with ImportError: Import of ntpath is not allowed in sandbox mode. Starter also fails 20/20 with the same Python 3.12.3 harness error; unsandboxed final tuning and holdout each have zero failures.",
    "SELECTED dyn_stock: {\"mean_edge\":107.7367374084125,\"retail_edge\":135.91459094566451,\"arb_edge\":-28.177853537266333,\"simulations\":600,\"seed_start\":1000,\"failures\":0}. Code frozen with SHA256 aafa7ec9047fe9332d376e2daf445f9963d2c3d3fbf135f51a99ba35736a35d6 before viewing final holdout. Checkpoint holdout scores were kept blind during search."
  ],
  "next": "Optimize posterior demand inventory targets using different costs for cheap crossing acquisitions and later refills, including the delay of one step before newly bought YES can cover asks. Validate on fresh seeds rather than reusing 30000..30599."
}