Ranked by target-spread mean_diff/SE (descending); inputs linked below. Screens only; no promotion.

|Rank/input|Target d ± SE|d/SE|
|---|---:|---:|
|1. [s1 flow](../../../../../../.imd/reads/artifacts/item-remove-s1-censored-flow)|-0.1188 ± 0.5185|-0.23|
|2. [s4 late](../../../../../../.imd/reads/artifacts/item-remove-late-uncovered)|-0.1650 ± 0.0957|-1.72|
|3. [s2 joint](../../../../../../.imd/reads/artifacts/item-remove-s2-neutral-joint)|-2.0945 ± 0.3523|-5.94|
|4. [s2 queue](../../../../../../.imd/reads/artifacts/item-remove-s2-old-price-queue)|-0.5251 ± 0.0683|-7.69|
|5. [s3 joint](../../../../../../.imd/reads/artifacts/item-remove-s3-neutral-joint)|-170.5416 ± 17.1639|-9.94|
|— [hysteresis](../../../../../../.imd/reads/artifacts/item-remove-hysteresis)|Not applicable|—|

```json
{
  "rows": [
    {
      "key": "remove-s1-censored-flow",
      "idea": "Replace only the spread-1 censored retail-flow posterior outputs with their fixed initialization priors; no tuning.",
      "what_was_deleted": "Spread-1 posterior output assignments replaced with fixed initialization priors: fl_rate=0.568734, fl_r=0.1265, fl_m=4.488; posterior bookkeeping remains.",
      "not_applicable": false,
      "code_sha256": "55a9172ab633d809f20dfd8c27a1d52f1ae334164a831f3324acfaa82eea0f15",
      "target_spreads": [
        1
      ],
      "complete": true,
      "h_n": 500,
      "incumbent_mean": 95.54201511346659,
      "mean_diff": -0.031125698969399783,
      "se": 0.13548025282531734,
      "retail_diff": 0.12172185658580033,
      "arb_diff": -0.15284755554980006,
      "by_spread": {
        "1": {
          "n": 131,
          "mean_diff": -0.1188003777458007,
          "se": 0.5184886348058438,
          "sum_d": -15.56284948469989,
          "sum_d2": 4580.031681518338
        },
        "2": {
          "n": 120,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        },
        "3": {
          "n": 133,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        },
        "4": {
          "n": 116,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        }
      },
      "failures": [],
      "check_values_ok": true,
      "python_version": "3.12.3",
      "flag": null
    },
    {
      "key": "remove-late-uncovered",
      "idea": "Delete the late-run uncovered-ask allowance (H<Hu and phat<0.05 -> ub=free*(1-H/Hu)) for spread 4 only.",
      "what_was_deleted": "Spread-4 guard added to: `if H < P['Hu'] and phat < 0.05:\n    ub = max(0.0, free * (1 - H / P['Hu']))` -> `if H < P['Hu'] and phat < 0.05 and self.s != 4:`. For s==4 the ordinary ub (free - (1-fu)*cash if x>=xu else 0) now stands. Off-target spreads gave d exactly 0 on all 384 seeds.",
      "not_applicable": false,
      "code_sha256": "b03f8009b0ff4721502af44abc4dfe271a32bc923e7ceb6d858e9c91128603d0",
      "target_spreads": [
        4
      ],
      "complete": true,
      "h_n": 500,
      "incumbent_mean": 95.54201511346659,
      "mean_diff": -0.03827361750640007,
      "se": 0.022355122818631192,
      "retail_diff": -0.021398069077008586,
      "arb_diff": -0.01687554842920136,
      "by_spread": {
        "1": {
          "n": 131,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        },
        "2": {
          "n": 120,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        },
        "3": {
          "n": 133,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        },
        "4": {
          "n": 116,
          "mean_diff": -0.16497248925172445,
          "se": 0.09573490354020356,
          "sum_d": -19.136808753200036,
          "sum_d2": 125.42043819941631
        }
      },
      "failures": [],
      "check_values_ok": true,
      "python_version": "3.12.3",
      "flag": null
    },
    {
      "key": "remove-s2-neutral-joint",
      "idea": "Joint deletion of three individually neutral spread-2 mechanisms, without tuning.",
      "what_was_deleted": "For spread 2: removed manage() old-price queue-retention block; disabled H==1 branch and stock rule at ct=floor(ptick) for 0.02<phat<0.1.",
      "not_applicable": false,
      "code_sha256": "fafdaa67dff2106329dc6b05ebdc621d69360061a4f809113bdfc73240091ab6",
      "target_spreads": [
        2
      ],
      "complete": true,
      "h_n": 500,
      "incumbent_mean": 95.54201511346659,
      "mean_diff": -0.5026720585804001,
      "se": 0.09331458161729382,
      "retail_diff": -0.1769463155517999,
      "arb_diff": -0.3257257430282001,
      "by_spread": {
        "1": {
          "n": 131,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        },
        "2": {
          "n": 120,
          "mean_diff": -2.094466910751667,
          "se": 0.3523106582136081,
          "sum_d": -251.33602929020006,
          "sum_d2": 2298.8885792701703
        },
        "3": {
          "n": 133,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        },
        "4": {
          "n": 116,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        }
      },
      "failures": [],
      "check_values_ok": true,
      "python_version": "3.12.3",
      "flag": null
    },
    {
      "key": "remove-s2-old-price-queue",
      "idea": "Remove spread-2 retention of orders away from the current target price for their queue position.",
      "what_was_deleted": "For spread 2: removed manage() retention at x==comp and x!=t when ahead and gain>=1; removed special bid retention at cb when bid_t==cb+1.",
      "not_applicable": false,
      "code_sha256": "2fc634eac3ba780ff4e3c17a608c56351b3bf4e9e1461632f1d5d44b0bb44895",
      "target_spreads": [
        2
      ],
      "complete": true,
      "h_n": 500,
      "incumbent_mean": 95.5420151134666,
      "mean_diff": -0.1260320644096,
      "se": 0.01917932125593179,
      "retail_diff": -0.15522850337519994,
      "arb_diff": 0.029196438963999902,
      "by_spread": {
        "1": {
          "n": 131,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        },
        "2": {
          "n": 120,
          "mean_diff": -0.5251336017066667,
          "se": 0.06830678881694997,
          "sum_d": -63.0160322048,
          "sum_d2": 99.71970840731291
        },
        "3": {
          "n": 133,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        },
        "4": {
          "n": 116,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        }
      },
      "failures": [],
      "check_values_ok": true,
      "python_version": "3.12.3",
      "flag": null
    },
    {
      "key": "remove-s3-neutral-joint",
      "idea": "Jointly delete spread-3 inventory-risk switch, q1 tick-1 stock rule, and probe renewal; q1 interpretation is uncertain and the switch is parametrically a no-op in N.",
      "what_was_deleted": "-                a *= (P['kLb'] if state.yes_inventory < ytg else P['kLf']) / P['kL']\n+                a *= (P['kLf'] if self.s == 3 else P['kLb'] if state.yes_inventory < ytg else P['kLf']) / P['kL']\n+            if self.s == 3:\n+                q1 = 0.0\n-        if free > 1.0:\n+        if free > 1.0 and self.s != 3:\n",
      "not_applicable": false,
      "code_sha256": "98dde8a07c73109e082ebeebe488c27a4ea3710ac703cd5064a85e0f15515353",
      "target_spreads": [
        3
      ],
      "complete": true,
      "h_n": 500,
      "incumbent_mean": 95.54201511346659,
      "mean_diff": -45.364058484933395,
      "se": 5.666497751364028,
      "retail_diff": -10.2052885591618,
      "arb_diff": -35.1587699257868,
      "by_spread": {
        "1": {
          "n": 131,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        },
        "2": {
          "n": 120,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        },
        "3": {
          "n": 133,
          "mean_diff": -170.5415732516293,
          "se": 17.163860784764537,
          "sum_d": -22682.0292424667,
          "sum_d2": 9040193.494282518
        },
        "4": {
          "n": 116,
          "mean_diff": 0.0,
          "se": 0.0,
          "sum_d": 0.0,
          "sum_d2": 0.0
        }
      },
      "failures": [],
      "check_values_ok": true,
      "python_version": "3.14.4",
      "flag": null
    },
    {
      "key": "remove-hysteresis",
      "idea": "Delete the spread-3 price hysteresis (quote held at previous tick until target moves beyond a band), guarded by self.s==3.",
      "what_was_deleted": "Nothing. Item reports no spread-3 price-band hysteresis in N. Exact-target size retention and queue-priority stickiness are different mechanisms; no simulations run.",
      "not_applicable": true,
      "code_sha256": null,
      "target_spreads": [
        3
      ],
      "complete": null,
      "h_n": null,
      "incumbent_mean": null,
      "mean_diff": null,
      "se": null,
      "retail_diff": null,
      "arb_diff": null,
      "by_spread": {},
      "failures": [],
      "check_values_ok": null,
      "python_version": "3.12.3",
      "flag": "Not applicable: no candidate code/hash, h_n, completion or check values; no simulations were required."
    }
  ],
  "top": null,
  "notes": "Sources are the six linked input JSONs; statistics are copied, descriptions shortened to fit. h_n is overall paired n (500), not target-spread n (116–133). All five screens report seeds 430000–430499, identical incumbent hash bf0d179f3e47d28819f6fa7e0d25548c2ced43374c216d36460b00a26a26ffae, matching candidate hashes, and zero off-target differences. check_values_ok means agreement with the common four-value vector in the inputs, not an independently verified simulator run. Keep spread-2 old-price queue retention and both joint bundles: their deletions are harmful. Spread-3 joint result cannot identify the harmful member; its stock-rule interpretation is uncertain and its Python differs. Spread-1 censored-flow deletion is within noise: a simplification candidate only, not proven equivalent. The requester flags one unverified external source for that item; no source URL/passage is supplied in its JSON, so its external attribution cannot be checked and is not relied on. Spread-4 late-uncovered deletion is mildly negative (<2 SE on target): retain provisionally, or replicate before simplification. Hysteresis is not applicable, not measured neutral. No positive target ratio, so top=null; no promotion or finalist evidence. Finalist seeds were not run here; no simulations run."
}
```