feat(governance): add risk-gated paper research slice

README.md

@@ -25,6 +25,7 @@
 - `backtest` — weight-based 多日仿真（rebalance_table / compute_nav / compare_to_benchmark）
 - `portfolio_construction` — 多期因子分数 → Top-K → 等权目标权重表
 - `research_pipeline` — 因子日 → 下一真实交易日 → 显式执行价 → 日末估值 → 成本后绩效（防前视编排）
+- `governed_pipeline` — 数据快照 → 因子版本 → 策略版本 → 回测运行 → 目标组合 → 风险决策 → Paper 订单意图；全链路带确定性 ID，风险拒绝时禁止生成订单意图
 - `artifact` — 版本化、确定性、存储中立的完整 research run 事实表与 manifest
 - `attribution` — 基于实际成交后持仓的隔夜 / 日内 / 交易成本逐日收益归因与闭合审计
 - `metrics` — 绝对绩效 + 严格日期对齐的 TE / IR / alpha / beta 基准相对绩效
@@ -55,6 +56,9 @@ pytest                    # 单元测试
 pytest --cov=src          # 覆盖率
 mypy --strict src/        # 类型检查
 ruff check src/ tests/    # lint
+
+# 无网络、无数据库、无券商的架构烟测
+uv run python -m quant_engine.governed_pipeline
 ```
 
 ## 使用
@@ -191,6 +195,23 @@ print(backtest.stats())
 print(backtest.benchmark_report())
 ```
 
+## 治理垂直切片
+
+`governed_pipeline` 不复制因子、回测、组合或执行算法，只编排现有能力并补充版本与风险契约。
+调用方必须显式提供 `DatasetSnapshot`、`FactorVersion`、`StrategyVersion`、代码提交和
+`RiskPolicy`。模块只会生成 `environment="paper"` 的订单意图，不连接数据库、数据供应商或
+券商；风险决策为拒绝时，订单意图固定为空，直接调用创建函数也会失败关闭。
+
+该切片对应 ResearchHub 架构的首个可执行验收链路：
+
+```text
+DatasetSnapshot → FactorVersion → StrategyVersion → BacktestRun
+→ PortfolioTarget → RiskDecision → PaperOrderIntent
+```
+
+平台总架构、五仓职责和十二层能力映射仍以 `research_platform/docs/architecture/` 为权威；
+本仓只拥有纯计算与离线模拟合同。
+
 ## 与 research_results 的关系
 
 `research_results` 依赖 `quant_engine`（通过 re-export 保持向后兼容）：

src/quant_engine/alpha_factors.py

@@ -3271,6 +3271,363 @@ def evaluate_phase3_formula(name: str, **inputs: pd.Series) -> pd.Series:
     return function(*(inputs[field] for field in required_inputs))
 
 
+# ── Phase 4 formula contract: frozen alpha051-alpha100 surface ──────────────
+
+# Phase 4 extends the versioned formula contract without mutating the Phase 3
+# catalogue, digest, dispatch surface, or the existing formula functions.
+ALPHA158_PHASE4_FORMULA_CONTRACT_VERSION = "1.0.0"
+ALPHA158_PHASE4_FORMULA_CATALOG_SHA256 = (
+    "858daf5e2abab5063fc28fcf7c79936096e2c76dde582fc7bab78b3458f17054"
+)
+
+_PHASE4_FORMULA_FUNCTIONS: dict[str, Callable[..., pd.Series]] = {
+    "alpha_051": alpha_051,
+    "alpha_052": alpha_052,
+    "alpha_053": alpha_053,
+    "alpha_054": alpha_054,
+    "alpha_055": alpha_055,
+    "alpha_056": alpha_056,
+    "alpha_057": alpha_057,
+    "alpha_058": alpha_058,
+    "alpha_059": alpha_059,
+    "alpha_060": alpha_060,
+    "alpha_061": alpha_061,
+    "alpha_062": alpha_062,
+    "alpha_063": alpha_063,
+    "alpha_064": alpha_064,
+    "alpha_065": alpha_065,
+    "alpha_066": alpha_066,
+    "alpha_067": alpha_067,
+    "alpha_068": alpha_068,
+    "alpha_069": alpha_069,
+    "alpha_070": alpha_070,
+    "alpha_071": alpha_071,
+    "alpha_072": alpha_072,
+    "alpha_073": alpha_073,
+    "alpha_074": alpha_074,
+    "alpha_075": alpha_075,
+    "alpha_076": alpha_076,
+    "alpha_077": alpha_077,
+    "alpha_078": alpha_078,
+    "alpha_079": alpha_079,
+    "alpha_080": alpha_080,
+    "alpha_081": alpha_081,
+    "alpha_082": alpha_082,
+    "alpha_083": alpha_083,
+    "alpha_084": alpha_084,
+    "alpha_085": alpha_085,
+    "alpha_086": alpha_086,
+    "alpha_087": alpha_087,
+    "alpha_088": alpha_088,
+    "alpha_089": alpha_089,
+    "alpha_090": alpha_090,
+    "alpha_091": alpha_091,
+    "alpha_092": alpha_092,
+    "alpha_093": alpha_093,
+    "alpha_094": alpha_094,
+    "alpha_095": alpha_095,
+    "alpha_096": alpha_096,
+    "alpha_097": alpha_097,
+    "alpha_098": alpha_098,
+    "alpha_099": alpha_099,
+    "alpha_100": alpha_100,
+}
+
+_PHASE4_INPUT_CATEGORIES = {
+    1: "single",
+    2: "pair",
+    3: "triple",
+    4: "quadruple",
+    5: "quintuple",
+}
+
+
+def _build_phase4_formula_specs() -> dict[str, dict[str, Any]]:
+    specs: dict[str, dict[str, Any]] = {}
+    for alpha_id, function in _PHASE4_FORMULA_FUNCTIONS.items():
+        meta = ALPHA158_REGISTRY[alpha_id]
+        call_inputs = _phase3_call_inputs(function)
+        formula_inputs = _phase3_string_list(meta, "inputs", alpha_id)
+        input_category = _PHASE4_INPUT_CATEGORIES.get(len(call_inputs))
+        if input_category is None:
+            raise RuntimeError(f"unsupported formula input count for {alpha_id}")
+        specs[alpha_id] = {
+            "name": alpha_id,
+            "contract_version": ALPHA158_PHASE4_FORMULA_CONTRACT_VERSION,
+            "formula": meta["formula"],
+            "category": meta["category"],
+            "complexity": meta["complexity"],
+            "parameters": _phase3_string_list(meta, "params", alpha_id),
+            "description": meta["description"],
+            "references": _phase3_string_list(meta, "references", alpha_id),
+            "call_inputs": list(call_inputs),
+            "formula_inputs": formula_inputs,
+            "input_category": input_category,
+        }
+    return specs
+
+
+ALPHA158_PHASE4_FORMULA_SPECS: Mapping[str, Mapping[str, Any]] = (
+    _freeze_phase3_formula_specs(_build_phase4_formula_specs())
+)
+
+
+def list_phase4_formulas() -> tuple[str, ...]:
+    """Return the frozen alpha051-alpha100 formula IDs in stable order."""
+    return tuple(ALPHA158_PHASE4_FORMULA_SPECS)
+
+
+def evaluate_phase4_formula(name: str, **inputs: pd.Series) -> pd.Series:
+    """Evaluate a Phase 4 formula with an exact, alignment-safe input contract."""
+    if name not in ALPHA158_PHASE4_FORMULA_SPECS:
+        raise KeyError(f"formula {name!r} not registered")
+
+    spec = ALPHA158_PHASE4_FORMULA_SPECS[name]
+    required_inputs = cast(tuple[str, ...], spec["call_inputs"])
+    missing_inputs = [field for field in required_inputs if field not in inputs]
+    unexpected_inputs = sorted(field for field in inputs if field not in required_inputs)
+    if missing_inputs or unexpected_inputs:
+        details: list[str] = []
+        if missing_inputs:
+            details.append(f"missing inputs {missing_inputs}")
+        if unexpected_inputs:
+            details.append(f"unexpected inputs {unexpected_inputs}")
+        raise ValueError(f"invalid inputs for {name}: {'; '.join(details)}")
+
+    for field in required_inputs:
+        if not isinstance(inputs[field], pd.Series):
+            raise TypeError(f"{field} must be a pandas Series")
+
+    primary_field = required_inputs[0]
+    primary = inputs[primary_field]
+    for field in required_inputs[1:]:
+        if not primary.index.equals(inputs[field].index):
+            raise ValueError(f"{field} index must align with {primary_field}")
+
+    function = _PHASE4_FORMULA_FUNCTIONS[name]
+    return function(*(inputs[field] for field in required_inputs))
+
+
+# ── Phase 5 formula contract: frozen alpha101-alpha150 surface ──────────────
+
+# Phase 5 extends the versioned formula contract without mutating any earlier
+# catalogue, digest, dispatch surface, or existing formula implementation.
+ALPHA158_PHASE5_FORMULA_CONTRACT_VERSION = "1.0.0"
+ALPHA158_PHASE5_FORMULA_CATALOG_SHA256 = (
+    "3368796169c9fbd39c4a34ea137e569964b15882fbf4de25124790d548db6533"
+)
+
+_PHASE5_FORMULA_FUNCTIONS: dict[str, Callable[..., pd.Series]] = {
+    "alpha_101": alpha_101,
+    "alpha_102": alpha_102,
+    "alpha_103": alpha_103,
+    "alpha_104": alpha_104,
+    "alpha_105": alpha_105,
+    "alpha_106": alpha_106,
+    "alpha_107": alpha_107,
+    "alpha_108": alpha_108,
+    "alpha_109": alpha_109,
+    "alpha_110": alpha_110,
+    "alpha_111": alpha_111,
+    "alpha_112": alpha_112,
+    "alpha_113": alpha_113,
+    "alpha_114": alpha_114,
+    "alpha_115": alpha_115,
+    "alpha_116": alpha_116,
+    "alpha_117": alpha_117,
+    "alpha_118": alpha_118,
+    "alpha_119": alpha_119,
+    "alpha_120": alpha_120,
+    "alpha_121": alpha_121,
+    "alpha_122": alpha_122,
+    "alpha_123": alpha_123,
+    "alpha_124": alpha_124,
+    "alpha_125": alpha_125,
+    "alpha_126": alpha_126,
+    "alpha_127": alpha_127,
+    "alpha_128": alpha_128,
+    "alpha_129": alpha_129,
+    "alpha_130": alpha_130,
+    "alpha_131": alpha_131,
+    "alpha_132": alpha_132,
+    "alpha_133": alpha_133,
+    "alpha_134": alpha_134,
+    "alpha_135": alpha_135,
+    "alpha_136": alpha_136,
+    "alpha_137": alpha_137,
+    "alpha_138": alpha_138,
+    "alpha_139": alpha_139,
+    "alpha_140": alpha_140,
+    "alpha_141": alpha_141,
+    "alpha_142": alpha_142,
+    "alpha_143": alpha_143,
+    "alpha_144": alpha_144,
+    "alpha_145": alpha_145,
+    "alpha_146": alpha_146,
+    "alpha_147": alpha_147,
+    "alpha_148": alpha_148,
+    "alpha_149": alpha_149,
+    "alpha_150": alpha_150,
+}
+
+
+def _build_phase5_formula_specs() -> dict[str, dict[str, Any]]:
+    specs: dict[str, dict[str, Any]] = {}
+    for alpha_id, function in _PHASE5_FORMULA_FUNCTIONS.items():
+        meta = ALPHA158_REGISTRY[alpha_id]
+        call_inputs = _phase3_call_inputs(function)
+        formula_inputs = _phase3_string_list(meta, "inputs", alpha_id)
+        input_category = _PHASE4_INPUT_CATEGORIES.get(len(call_inputs))
+        if input_category is None:
+            raise RuntimeError(f"unsupported formula input count for {alpha_id}")
+        specs[alpha_id] = {
+            "name": alpha_id,
+            "contract_version": ALPHA158_PHASE5_FORMULA_CONTRACT_VERSION,
+            "formula": meta["formula"],
+            "category": meta["category"],
+            "complexity": meta["complexity"],
+            "parameters": _phase3_string_list(meta, "params", alpha_id),
+            "description": meta["description"],
+            "references": _phase3_string_list(meta, "references", alpha_id),
+            "call_inputs": list(call_inputs),
+            "formula_inputs": formula_inputs,
+            "input_category": input_category,
+        }
+    return specs
+
+
+ALPHA158_PHASE5_FORMULA_SPECS: Mapping[str, Mapping[str, Any]] = (
+    _freeze_phase3_formula_specs(_build_phase5_formula_specs())
+)
+
+
+def list_phase5_formulas() -> tuple[str, ...]:
+    """Return the frozen alpha101-alpha150 formula IDs in stable order."""
+    return tuple(ALPHA158_PHASE5_FORMULA_SPECS)
+
+
+def evaluate_phase5_formula(name: str, **inputs: pd.Series) -> pd.Series:
+    """Evaluate a Phase 5 formula with an exact, alignment-safe input contract."""
+    if name not in ALPHA158_PHASE5_FORMULA_SPECS:
+        raise KeyError(f"formula {name!r} not registered")
+
+    spec = ALPHA158_PHASE5_FORMULA_SPECS[name]
+    required_inputs = cast(tuple[str, ...], spec["call_inputs"])
+    missing_inputs = [field for field in required_inputs if field not in inputs]
+    unexpected_inputs = sorted(field for field in inputs if field not in required_inputs)
+    if missing_inputs or unexpected_inputs:
+        details: list[str] = []
+        if missing_inputs:
+            details.append(f"missing inputs {missing_inputs}")
+        if unexpected_inputs:
+            details.append(f"unexpected inputs {unexpected_inputs}")
+        raise ValueError(f"invalid inputs for {name}: {'; '.join(details)}")
+
+    for field in required_inputs:
+        if not isinstance(inputs[field], pd.Series):
+            raise TypeError(f"{field} must be a pandas Series")
+
+    primary_field = required_inputs[0]
+    primary = inputs[primary_field]
+    for field in required_inputs[1:]:
+        if len(inputs[field]) != len(primary):
+            raise ValueError(f"{field} length must match {primary_field}")
+        if not primary.index.equals(inputs[field].index):
+            raise ValueError(f"{field} index must align with {primary_field}")
+
+    function = _PHASE5_FORMULA_FUNCTIONS[name]
+    return function(*(inputs[field] for field in required_inputs))
+
+
+# ── Phase 6 formula contract: frozen alpha151-alpha158 surface ──────────────
+
+# Phase 6 completes the versioned formula contract without mutating any
+# earlier catalogue, digest, dispatch surface, or formula implementation.
+ALPHA158_PHASE6_FORMULA_CONTRACT_VERSION = "1.0.0"
+ALPHA158_PHASE6_FORMULA_CATALOG_SHA256 = (
+    "70ffae16ca6cbb59a1e8644f5cdeec1d291fa75ff8b5647fb534af0083408219"
+)
+
+_PHASE6_FORMULA_FUNCTIONS: dict[str, Callable[..., pd.Series]] = {
+    "alpha_151": alpha_151,
+    "alpha_152": alpha_152,
+    "alpha_153": alpha_153,
+    "alpha_154": alpha_154,
+    "alpha_155": alpha_155,
+    "alpha_156": alpha_156,
+    "alpha_157": alpha_157,
+    "alpha_158": alpha_158,
+}
+
+
+def _build_phase6_formula_specs() -> dict[str, dict[str, Any]]:
+    specs: dict[str, dict[str, Any]] = {}
+    for alpha_id, function in _PHASE6_FORMULA_FUNCTIONS.items():
+        meta = ALPHA158_REGISTRY[alpha_id]
+        call_inputs = _phase3_call_inputs(function)
+        formula_inputs = _phase3_string_list(meta, "inputs", alpha_id)
+        input_category = _PHASE4_INPUT_CATEGORIES.get(len(call_inputs))
+        if input_category is None:
+            raise RuntimeError(f"unsupported formula input count for {alpha_id}")
+        specs[alpha_id] = {
+            "name": alpha_id,
+            "contract_version": ALPHA158_PHASE6_FORMULA_CONTRACT_VERSION,
+            "formula": meta["formula"],
+            "category": meta["category"],
+            "complexity": meta["complexity"],
+            "parameters": _phase3_string_list(meta, "params", alpha_id),
+            "description": meta["description"],
+            "references": _phase3_string_list(meta, "references", alpha_id),
+            "call_inputs": list(call_inputs),
+            "formula_inputs": formula_inputs,
+            "input_category": input_category,
+        }
+    return specs
+
+
+ALPHA158_PHASE6_FORMULA_SPECS: Mapping[str, Mapping[str, Any]] = (
+    _freeze_phase3_formula_specs(_build_phase6_formula_specs())
+)
+
+
+def list_phase6_formulas() -> tuple[str, ...]:
+    """Return the frozen alpha151-alpha158 formula IDs in stable order."""
+    return tuple(ALPHA158_PHASE6_FORMULA_SPECS)
+
+
+def evaluate_phase6_formula(name: str, **inputs: pd.Series) -> pd.Series:
+    """Evaluate a Phase 6 formula with an exact, alignment-safe input contract."""
+    if name not in ALPHA158_PHASE6_FORMULA_SPECS:
+        raise KeyError(f"formula {name!r} not registered")
+
+    spec = ALPHA158_PHASE6_FORMULA_SPECS[name]
+    required_inputs = cast(tuple[str, ...], spec["call_inputs"])
+    missing_inputs = [field for field in required_inputs if field not in inputs]
+    unexpected_inputs = sorted(field for field in inputs if field not in required_inputs)
+    if missing_inputs or unexpected_inputs:
+        details: list[str] = []
+        if missing_inputs:
+            details.append(f"missing inputs {missing_inputs}")
+        if unexpected_inputs:
+            details.append(f"unexpected inputs {unexpected_inputs}")
+        raise ValueError(f"invalid inputs for {name}: {'; '.join(details)}")
+
+    for field in required_inputs:
+        if not isinstance(inputs[field], pd.Series):
+            raise TypeError(f"{field} must be a pandas Series")
+
+    primary_field = required_inputs[0]
+    primary = inputs[primary_field]
+    for field in required_inputs[1:]:
+        if len(inputs[field]) != len(primary):
+            raise ValueError(f"{field} length must match {primary_field}")
+        if not primary.index.equals(inputs[field].index):
+            raise ValueError(f"{field} index must align with {primary_field}")
+
+    function = _PHASE6_FORMULA_FUNCTIONS[name]
+    return function(*(inputs[field] for field in required_inputs))
+
+
 __all__ = [
     "rank",
     "delta",
@@ -3309,6 +3666,21 @@ __all__ = [
     "ALPHA158_PHASE3_FORMULA_SPECS",
     "list_phase3_formulas",
     "evaluate_phase3_formula",
+    "ALPHA158_PHASE4_FORMULA_CONTRACT_VERSION",
+    "ALPHA158_PHASE4_FORMULA_CATALOG_SHA256",
+    "ALPHA158_PHASE4_FORMULA_SPECS",
+    "list_phase4_formulas",
+    "evaluate_phase4_formula",
+    "ALPHA158_PHASE5_FORMULA_CONTRACT_VERSION",
+    "ALPHA158_PHASE5_FORMULA_CATALOG_SHA256",
+    "ALPHA158_PHASE5_FORMULA_SPECS",
+    "list_phase5_formulas",
+    "evaluate_phase5_formula",
+    "ALPHA158_PHASE6_FORMULA_CONTRACT_VERSION",
+    "ALPHA158_PHASE6_FORMULA_CATALOG_SHA256",
+    "ALPHA158_PHASE6_FORMULA_SPECS",
+    "list_phase6_formulas",
+    "evaluate_phase6_formula",
     "alpha_001",
     "alpha_002",
     "alpha_003",

src/quant_engine/governed_pipeline.py

@@ -0,0 +1,540 @@
+"""Versioned, risk-gated, paper-only quantitative research vertical slice.
+
+The module composes existing ``quant_engine`` calculations into a small set of
+storage-neutral governance contracts.  It never reads a database, calls a data
+vendor, loads credentials, or routes an order to a broker.
+"""
+
+from __future__ import annotations
+
+import hashlib
+import json
+import math
+import re
+from collections.abc import Mapping
+from dataclasses import asdict, dataclass
+from datetime import UTC, datetime
+from enum import StrEnum
+from types import MappingProxyType
+
+import pandas as pd
+
+from quant_engine.execution import ExecutionConfig
+from quant_engine.research_pipeline import FactorBacktestResult, run_factor_backtest_research
+
+_SHA256 = re.compile(r"^[0-9a-f]{64}$")
+_GIT_SHA = re.compile(r"^[0-9a-f]{40}$")
+
+__all__ = [
+    "DatasetSnapshot",
+    "FactorVersion",
+    "StrategyStage",
+    "StrategyVersion",
+    "BacktestRun",
+    "PortfolioTarget",
+    "RiskPolicy",
+    "RiskDecisionStatus",
+    "RiskDecision",
+    "PaperOrderIntent",
+    "GovernedFactorSliceResult",
+    "evaluate_portfolio_risk",
+    "create_paper_order_intent",
+    "run_governed_factor_slice",
+]
+
+
+def _required_text(value: str, name: str) -> str:
+    normalized = value.strip()
+    if not normalized:
+        raise ValueError(f"{name} must be non-empty")
+    return normalized
+
+
+def _aware_utc(value: datetime, name: str) -> datetime:
+    if not isinstance(value, datetime) or value.tzinfo is None or value.utcoffset() is None:
+        raise ValueError(f"{name} must be timezone-aware")
+    return value.astimezone(UTC)
+
+
+def _canonical_value(value: object) -> object:
+    if value is None or isinstance(value, str | bool | int):
+        return value
+    if isinstance(value, float):
+        if math.isnan(value):
+            return "NaN"
+        if math.isinf(value):
+            return "Infinity" if value > 0 else "-Infinity"
+        return value
+    if isinstance(value, datetime):
+        return _aware_utc(value, "datetime").isoformat()
+    if isinstance(value, StrEnum):
+        return value.value
+    if isinstance(value, Mapping):
+        return {
+            str(key): _canonical_value(item)
+            for key, item in sorted(value.items(), key=lambda pair: str(pair[0]))
+        }
+    if isinstance(value, list | tuple):
+        return [_canonical_value(item) for item in value]
+    raise TypeError(f"unsupported canonical value: {type(value).__name__}")
+
+
+def _stable_id(prefix: str, payload: Mapping[str, object]) -> str:
+    encoded = json.dumps(
+        _canonical_value(payload),
+        ensure_ascii=False,
+        sort_keys=True,
+        separators=(",", ":"),
+        allow_nan=False,
+    ).encode("utf-8")
+    return f"{prefix}:{hashlib.sha256(encoded).hexdigest()}"
+
+
+def _immutable_weights(values: Mapping[str, float]) -> Mapping[str, float]:
+    normalized: dict[str, float] = {}
+    for raw_asset, raw_weight in values.items():
+        asset = _required_text(str(raw_asset), "asset_id")
+        weight = float(raw_weight)
+        if not math.isfinite(weight) or weight < 0:
+            raise ValueError("portfolio weights must be finite and non-negative")
+        if asset in normalized:
+            raise ValueError(f"duplicate asset_id: {asset}")
+        normalized[asset] = weight
+    return MappingProxyType(dict(sorted(normalized.items())))
+
+
+@dataclass(frozen=True, slots=True)
+class DatasetSnapshot:
+    """Point-in-time identity for caller-supplied research data."""
+
+    snapshot_id: str
+    schema_version: str
+    content_sha256: str
+    effective_at: datetime
+    available_at: datetime
+    ingested_at: datetime
+
+    def __post_init__(self) -> None:
+        object.__setattr__(self, "snapshot_id", _required_text(self.snapshot_id, "snapshot_id"))
+        object.__setattr__(
+            self, "schema_version", _required_text(self.schema_version, "schema_version")
+        )
+        digest = self.content_sha256.strip().lower()
+        if not _SHA256.fullmatch(digest):
+            raise ValueError("content_sha256 must be a lowercase SHA-256 digest")
+        object.__setattr__(self, "content_sha256", digest)
+        effective_at = _aware_utc(self.effective_at, "effective_at")
+        available_at = _aware_utc(self.available_at, "available_at")
+        ingested_at = _aware_utc(self.ingested_at, "ingested_at")
+        if not effective_at <= available_at <= ingested_at:
+            raise ValueError("timestamps must satisfy effective_at <= available_at <= ingested_at")
+        object.__setattr__(self, "effective_at", effective_at)
+        object.__setattr__(self, "available_at", available_at)
+        object.__setattr__(self, "ingested_at", ingested_at)
+
+
+@dataclass(frozen=True, slots=True)
+class FactorVersion:
+    """Versioned factor definition identity without factor implementation duplication."""
+
+    factor_id: str
+    version: str
+    definition_sha256: str
+    dataset_schema_version: str
+
+    def __post_init__(self) -> None:
+        object.__setattr__(self, "factor_id", _required_text(self.factor_id, "factor_id"))
+        object.__setattr__(self, "version", _required_text(self.version, "version"))
+        object.__setattr__(
+            self,
+            "dataset_schema_version",
+            _required_text(self.dataset_schema_version, "dataset_schema_version"),
+        )
+        digest = self.definition_sha256.strip().lower()
+        if not _SHA256.fullmatch(digest):
+            raise ValueError("definition_sha256 must be a lowercase SHA-256 digest")
+        object.__setattr__(self, "definition_sha256", digest)
+
+    @property
+    def version_id(self) -> str:
+        return f"{self.factor_id}@{self.version}"
+
+
+class StrategyStage(StrEnum):
+    DRAFT = "Draft"
+    RESEARCH = "Research"
+    VALIDATED = "Validated"
+    APPROVED = "Approved"
+    PAPER = "Paper"
+    LIVE = "Live"
+    PAUSED = "Paused"
+    RETIRED = "Retired"
+
+
+@dataclass(frozen=True, slots=True)
+class StrategyVersion:
+    """Strategy lineage and lifecycle state used by the governed slice."""
+
+    strategy_id: str
+    version: str
+    factor_version_id: str
+    stage: StrategyStage
+
+    def __post_init__(self) -> None:
+        object.__setattr__(self, "strategy_id", _required_text(self.strategy_id, "strategy_id"))
+        object.__setattr__(self, "version", _required_text(self.version, "version"))
+        object.__setattr__(
+            self,
+            "factor_version_id",
+            _required_text(self.factor_version_id, "factor_version_id"),
+        )
+        if not isinstance(self.stage, StrategyStage):
+            raise TypeError("stage must be a StrategyStage")
+
+    @property
+    def version_id(self) -> str:
+        return f"{self.strategy_id}@{self.version}"
+
+
+@dataclass(frozen=True, slots=True)
+class BacktestRun:
+    run_id: str
+    dataset_snapshot_id: str
+    factor_version_id: str
+    strategy_version_id: str
+    code_revision: str
+    config_hash: str
+    created_at: datetime
+
+
+@dataclass(frozen=True, slots=True)
+class PortfolioTarget:
+    target_id: str
+    backtest_run_id: str
+    dataset_snapshot_id: str
+    weights: Mapping[str, float]
+    created_at: datetime
+
+    def __post_init__(self) -> None:
+        object.__setattr__(self, "weights", _immutable_weights(self.weights))
+
+    @property
+    def gross_exposure(self) -> float:
+        return float(sum(abs(weight) for weight in self.weights.values()))
+
+
+@dataclass(frozen=True, slots=True)
+class RiskPolicy:
+    policy_id: str
+    max_gross_exposure: float
+    max_single_asset_weight: float
+    max_positions: int
+
+    def __post_init__(self) -> None:
+        object.__setattr__(self, "policy_id", _required_text(self.policy_id, "policy_id"))
+        if not math.isfinite(self.max_gross_exposure) or self.max_gross_exposure <= 0:
+            raise ValueError("max_gross_exposure must be positive and finite")
+        if not math.isfinite(self.max_single_asset_weight) or not (
+            0 < self.max_single_asset_weight <= 1
+        ):
+            raise ValueError("max_single_asset_weight must be in (0, 1]")
+        if (
+            isinstance(self.max_positions, bool)
+            or not isinstance(self.max_positions, int)
+            or self.max_positions <= 0
+        ):
+            raise ValueError("max_positions must be a positive integer")
+
+
+class RiskDecisionStatus(StrEnum):
+    APPROVED = "approved"
+    REJECTED = "rejected"
+
+
+@dataclass(frozen=True, slots=True)
+class RiskDecision:
+    decision_id: str
+    portfolio_target_id: str
+    policy_id: str
+    status: RiskDecisionStatus
+    reasons: tuple[str, ...]
+    created_at: datetime
+
+
+@dataclass(frozen=True, slots=True, init=False)
+class PaperOrderIntent:
+    intent_id: str
+    portfolio_target_id: str
+    risk_decision_id: str
+    environment: str
+    target_weights: Mapping[str, float]
+    created_at: datetime
+
+    def __init__(self, target: PortfolioTarget, decision: RiskDecision) -> None:
+        if decision.status is not RiskDecisionStatus.APPROVED:
+            raise ValueError("an approved risk decision is required")
+        if decision.portfolio_target_id != target.target_id:
+            raise ValueError("risk decision does not bind the supplied portfolio target")
+        target_weights = _immutable_weights(target.weights)
+        payload = {
+            "portfolio_target_id": target.target_id,
+            "risk_decision_id": decision.decision_id,
+            "environment": "paper",
+            "target_weights": target_weights,
+            "created_at": decision.created_at,
+        }
+        object.__setattr__(self, "intent_id", _stable_id("order-intent", payload))
+        object.__setattr__(self, "portfolio_target_id", target.target_id)
+        object.__setattr__(self, "risk_decision_id", decision.decision_id)
+        object.__setattr__(self, "environment", "paper")
+        object.__setattr__(self, "target_weights", target_weights)
+        object.__setattr__(self, "created_at", decision.created_at)
+
+
+@dataclass(frozen=True, slots=True)
+class GovernedFactorSliceResult:
+    dataset_snapshot: DatasetSnapshot
+    factor_version: FactorVersion
+    strategy_version: StrategyVersion
+    research_result: FactorBacktestResult
+    backtest_run: BacktestRun
+    portfolio_target: PortfolioTarget
+    risk_decision: RiskDecision
+    order_intent: PaperOrderIntent | None
+
+
+def evaluate_portfolio_risk(
+    target: PortfolioTarget,
+    policy: RiskPolicy,
+    *,
+    created_at: datetime | None = None,
+) -> RiskDecision:
+    """Apply fail-closed paper risk limits to a versioned target portfolio."""
+    decision_time = target.created_at if created_at is None else _aware_utc(created_at, "created_at")
+    reasons: list[str] = []
+    if target.gross_exposure > policy.max_gross_exposure + 1e-12:
+        reasons.append(
+            f"gross exposure {target.gross_exposure:.12g} exceeds "
+            f"limit {policy.max_gross_exposure:.12g}"
+        )
+    positive_weights = [weight for weight in target.weights.values() if weight > 0]
+    largest_weight = max(positive_weights, default=0.0)
+    if largest_weight > policy.max_single_asset_weight + 1e-12:
+        reasons.append(
+            f"single-asset weight {largest_weight:.12g} exceeds "
+            f"limit {policy.max_single_asset_weight:.12g}"
+        )
+    if len(positive_weights) > policy.max_positions:
+        reasons.append(
+            f"position count {len(positive_weights)} exceeds limit {policy.max_positions}"
+        )
+    status = RiskDecisionStatus.REJECTED if reasons else RiskDecisionStatus.APPROVED
+    payload = {
+        "portfolio_target_id": target.target_id,
+        "policy_id": policy.policy_id,
+        "status": status,
+        "reasons": reasons,
+        "created_at": decision_time,
+    }
+    return RiskDecision(
+        decision_id=_stable_id("risk-decision", payload),
+        portfolio_target_id=target.target_id,
+        policy_id=policy.policy_id,
+        status=status,
+        reasons=tuple(reasons),
+        created_at=decision_time,
+    )
+
+
+def create_paper_order_intent(
+    target: PortfolioTarget,
+    decision: RiskDecision,
+) -> PaperOrderIntent:
+    """Create a paper-only intent after an exact, approved risk decision."""
+    return PaperOrderIntent(target, decision)
+
+
+def run_governed_factor_slice(
+    *,
+    factor_scores: pd.DataFrame,
+    execution_prices: pd.DataFrame,
+    valuation_prices: pd.DataFrame,
+    dataset_snapshot: DatasetSnapshot,
+    factor_version: FactorVersion,
+    strategy_version: StrategyVersion,
+    risk_policy: RiskPolicy,
+    code_revision: str,
+    created_at: datetime,
+    top_k: int,
+    execution_price_field: str,
+    valuation_price_field: str,
+    lag_sessions: int = 1,
+    gross_exposure: float = 1.0,
+    largest: bool = True,
+    initial_cash: float = 1_000_000.0,
+    execution_config: ExecutionConfig | None = None,
+) -> GovernedFactorSliceResult:
+    """Run snapshot → factor → strategy → backtest → target → risk → paper intent."""
+    if strategy_version.factor_version_id != factor_version.version_id:
+        raise ValueError("strategy factor lineage does not match factor_version")
+    if dataset_snapshot.schema_version != factor_version.dataset_schema_version:
+        raise ValueError("factor dataset schema does not match dataset snapshot")
+    if strategy_version.stage not in {StrategyStage.APPROVED, StrategyStage.PAPER}:
+        raise ValueError("strategy must be in Approved or Paper stage")
+    normalized_revision = code_revision.strip().lower()
+    if not _GIT_SHA.fullmatch(normalized_revision):
+        raise ValueError("code_revision must be a lowercase 40-character Git SHA")
+    run_time = _aware_utc(created_at, "created_at")
+    if dataset_snapshot.available_at > run_time or dataset_snapshot.ingested_at > run_time:
+        raise ValueError("dataset snapshot must be available before the research run")
+    if not factor_scores.empty:
+        latest_decision = pd.Timestamp(factor_scores.index.max())
+        latest_decision_date = latest_decision.date()
+        if latest_decision_date > run_time.date():
+            raise ValueError("factor_scores contain future decision dates")
+    config = ExecutionConfig() if execution_config is None else execution_config
+    if not isinstance(config, ExecutionConfig):
+        raise TypeError("execution_config must be an ExecutionConfig")
+
+    parameters: dict[str, object] = {
+        "top_k": top_k,
+        "execution_price_field": execution_price_field,
+        "valuation_price_field": valuation_price_field,
+        "lag_sessions": lag_sessions,
+        "gross_exposure": gross_exposure,
+        "largest": largest,
+        "initial_cash": initial_cash,
+        "execution_config": asdict(config),
+    }
+    config_hash = _stable_id("config", parameters).split(":", maxsplit=1)[1]
+    run_payload = {
+        "dataset_snapshot_id": dataset_snapshot.snapshot_id,
+        "dataset_content_sha256": dataset_snapshot.content_sha256,
+        "factor_version_id": factor_version.version_id,
+        "factor_definition_sha256": factor_version.definition_sha256,
+        "strategy_version_id": strategy_version.version_id,
+        "code_revision": normalized_revision,
+        "config_hash": config_hash,
+        "created_at": run_time,
+    }
+    backtest_run = BacktestRun(
+        run_id=_stable_id("backtest-run", run_payload),
+        dataset_snapshot_id=dataset_snapshot.snapshot_id,
+        factor_version_id=factor_version.version_id,
+        strategy_version_id=strategy_version.version_id,
+        code_revision=normalized_revision,
+        config_hash=config_hash,
+        created_at=run_time,
+    )
+
+    research_result = run_factor_backtest_research(
+        factor_scores,
+        execution_prices,
+        valuation_prices,
+        top_k=top_k,
+        execution_price_field=execution_price_field,
+        valuation_price_field=valuation_price_field,
+        lag_sessions=lag_sessions,
+        gross_exposure=gross_exposure,
+        largest=largest,
+        initial_cash=initial_cash,
+        config=config,
+    )
+    if research_result.schedule.decision_weights.empty:
+        raise ValueError("governed slice requires at least one target portfolio")
+    final_weights = {
+        str(asset): float(weight)
+        for asset, weight in research_result.schedule.decision_weights.iloc[-1].items()
+    }
+    target_payload = {
+        "backtest_run_id": backtest_run.run_id,
+        "dataset_snapshot_id": dataset_snapshot.snapshot_id,
+        "weights": final_weights,
+        "created_at": run_time,
+    }
+    portfolio_target = PortfolioTarget(
+        target_id=_stable_id("portfolio-target", target_payload),
+        backtest_run_id=backtest_run.run_id,
+        dataset_snapshot_id=dataset_snapshot.snapshot_id,
+        weights=final_weights,
+        created_at=run_time,
+    )
+    risk_decision = evaluate_portfolio_risk(
+        portfolio_target,
+        risk_policy,
+        created_at=run_time,
+    )
+    order_intent = (
+        create_paper_order_intent(portfolio_target, risk_decision)
+        if risk_decision.status is RiskDecisionStatus.APPROVED
+        else None
+    )
+    return GovernedFactorSliceResult(
+        dataset_snapshot=dataset_snapshot,
+        factor_version=factor_version,
+        strategy_version=strategy_version,
+        research_result=research_result,
+        backtest_run=backtest_run,
+        portfolio_target=portfolio_target,
+        risk_decision=risk_decision,
+        order_intent=order_intent,
+    )
+
+
+def _demo() -> Mapping[str, object]:
+    dates = pd.date_range("2026-01-05", periods=4, freq="B")
+    scores = pd.DataFrame({"A": [2.0, 1.0], "B": [1.0, 2.0]}, index=dates[:2])
+    opens = pd.DataFrame({"A": [10.0, 10.0, 10.2, 10.3], "B": [20.0, 20.0, 20.5, 21.0]}, index=dates)
+    result = run_governed_factor_slice(
+        factor_scores=scores,
+        execution_prices=opens,
+        valuation_prices=opens * 1.01,
+        dataset_snapshot=DatasetSnapshot(
+            snapshot_id="dataset:architecture-smoke-v1",
+            schema_version="1.0.0",
+            content_sha256="a" * 64,
+            effective_at=datetime(2026, 1, 8, 7, tzinfo=UTC),
+            available_at=datetime(2026, 1, 8, 8, tzinfo=UTC),
+            ingested_at=datetime(2026, 1, 8, 8, 5, tzinfo=UTC),
+        ),
+        factor_version=FactorVersion(
+            factor_id="factor:architecture-smoke",
+            version="1.0.0",
+            definition_sha256="b" * 64,
+            dataset_schema_version="1.0.0",
+        ),
+        strategy_version=StrategyVersion(
+            strategy_id="strategy:architecture-smoke",
+            version="1.0.0",
+            factor_version_id="factor:architecture-smoke@1.0.0",
+            stage=StrategyStage.APPROVED,
+        ),
+        risk_policy=RiskPolicy(
+            policy_id="risk:architecture-smoke@1.0.0",
+            max_gross_exposure=1.0,
+            max_single_asset_weight=0.6,
+            max_positions=10,
+        ),
+        code_revision="c" * 40,
+        created_at=datetime(2026, 1, 9, 1, tzinfo=UTC),
+        top_k=2,
+        execution_price_field="open",
+        valuation_price_field="close",
+        execution_config=ExecutionConfig(
+            commission_bps=0,
+            stamp_tax_bps=0,
+            slippage_bps=0,
+            min_trade_amount=0,
+        ),
+    )
+    return {
+        "backtest_run_id": result.backtest_run.run_id,
+        "portfolio_target_id": result.portfolio_target.target_id,
+        "risk_decision": result.risk_decision.status.value,
+        "order_intent_id": result.order_intent.intent_id if result.order_intent else None,
+        "environment": result.order_intent.environment if result.order_intent else None,
+    }
+
+
+if __name__ == "__main__":
+    print(json.dumps(_demo(), ensure_ascii=False, sort_keys=True))

tests/test_alpha_factors.py

@@ -14,6 +14,15 @@ from quant_engine.alpha_factors import (
     ALPHA158_PHASE3_FORMULA_CATALOG_SHA256,
     ALPHA158_PHASE3_FORMULA_CONTRACT_VERSION,
     ALPHA158_PHASE3_FORMULA_SPECS,
+    ALPHA158_PHASE4_FORMULA_CATALOG_SHA256,
+    ALPHA158_PHASE4_FORMULA_CONTRACT_VERSION,
+    ALPHA158_PHASE4_FORMULA_SPECS,
+    ALPHA158_PHASE5_FORMULA_CATALOG_SHA256,
+    ALPHA158_PHASE5_FORMULA_CONTRACT_VERSION,
+    ALPHA158_PHASE5_FORMULA_SPECS,
+    ALPHA158_PHASE6_FORMULA_CATALOG_SHA256,
+    ALPHA158_PHASE6_FORMULA_CONTRACT_VERSION,
+    ALPHA158_PHASE6_FORMULA_SPECS,
     alpha_001,
     alpha_002,
     alpha_003,
@@ -175,9 +184,15 @@ from quant_engine.alpha_factors import (
     evaluate_phase1_operator,
     evaluate_phase2_operator,
     evaluate_phase3_formula,
+    evaluate_phase4_formula,
+    evaluate_phase5_formula,
+    evaluate_phase6_formula,
     list_phase1_operators,
     list_phase2_operators,
     list_phase3_formulas,
+    list_phase4_formulas,
+    list_phase5_formulas,
+    list_phase6_formulas,
     correlation,
     covariance,
     decay_linear,
@@ -1641,3 +1656,504 @@ def test_phase3_dispatch_rejects_implicit_series_alignment():
             close=inputs["close"],
             volume=misaligned_volume,
         )
+
+
+# ── Alpha158 Phase 4: versioned alpha051-alpha100 formula contract ──────────
+
+
+def test_phase4_formula_catalog_is_versioned_exact_and_content_addressed():
+    import hashlib
+    import json
+    from collections import Counter
+
+    expected_ids = tuple(f"alpha_{number:03d}" for number in range(51, 101))
+    expected_fields = {
+        "name",
+        "contract_version",
+        "formula",
+        "category",
+        "complexity",
+        "parameters",
+        "description",
+        "references",
+        "call_inputs",
+        "formula_inputs",
+        "input_category",
+    }
+
+    assert ALPHA158_PHASE4_FORMULA_CONTRACT_VERSION == "1.0.0"
+    assert list_phase4_formulas() == expected_ids
+    assert tuple(ALPHA158_PHASE4_FORMULA_SPECS) == expected_ids
+    assert Counter(
+        spec["input_category"] for spec in ALPHA158_PHASE4_FORMULA_SPECS.values()
+    ) == {"single": 1, "pair": 27, "triple": 11, "quadruple": 10, "quintuple": 1}
+
+    for alpha_id, spec in ALPHA158_PHASE4_FORMULA_SPECS.items():
+        assert set(spec) == expected_fields
+        assert spec["name"] == alpha_id
+        assert spec["contract_version"] == ALPHA158_PHASE4_FORMULA_CONTRACT_VERSION
+        assert spec["formula"] == ALPHA158_REGISTRY[alpha_id]["formula"]
+
+    serializable_specs = {
+        alpha_id: {
+            field: list(value) if isinstance(value, tuple) else value
+            for field, value in spec.items()
+        }
+        for alpha_id, spec in ALPHA158_PHASE4_FORMULA_SPECS.items()
+    }
+    encoded = json.dumps(
+        serializable_specs,
+        sort_keys=True,
+        separators=(",", ":"),
+        ensure_ascii=False,
+    ).encode()
+    assert hashlib.sha256(encoded).hexdigest() == ALPHA158_PHASE4_FORMULA_CATALOG_SHA256
+    assert ALPHA158_PHASE4_FORMULA_CATALOG_SHA256 == (
+        "858daf5e2abab5063fc28fcf7c79936096e2c76dde582fc7bab78b3458f17054"
+    )
+
+
+def test_phase4_formula_catalog_is_recursively_immutable():
+    import operator
+
+    with pytest.raises(TypeError):
+        operator.setitem(ALPHA158_PHASE4_FORMULA_SPECS, "alpha_051", {})
+    with pytest.raises(TypeError):
+        operator.setitem(
+            ALPHA158_PHASE4_FORMULA_SPECS["alpha_051"],
+            "formula",
+            "changed",
+        )
+    with pytest.raises(TypeError):
+        operator.setitem(
+            ALPHA158_PHASE4_FORMULA_SPECS["alpha_051"]["call_inputs"],
+            0,
+            "volume",
+        )
+
+
+def test_phase4_catalog_freezes_callable_and_formula_inputs():
+    import inspect
+
+    for alpha_id, spec in ALPHA158_PHASE4_FORMULA_SPECS.items():
+        function = getattr(alpha_factors_module, alpha_id)
+        signature_inputs = tuple(
+            "open" if name == "open_" else name
+            for name in inspect.signature(function).parameters
+        )
+        assert spec["call_inputs"] == signature_inputs
+        assert spec["formula_inputs"] == tuple(ALPHA158_REGISTRY[alpha_id]["inputs"])
+
+    assert ALPHA158_PHASE4_FORMULA_SPECS["alpha_055"]["call_inputs"] == (
+        "open",
+        "high",
+        "low",
+        "volume",
+        "close",
+    )
+
+
+def test_phase4_dispatch_matches_all_existing_alpha051_alpha100_functions():
+    inputs = _phase3_market_inputs()
+
+    for alpha_id, spec in ALPHA158_PHASE4_FORMULA_SPECS.items():
+        call_inputs = spec["call_inputs"]
+        function = getattr(alpha_factors_module, alpha_id)
+        expected = function(*(inputs[name] for name in call_inputs))
+        actual = evaluate_phase4_formula(
+            alpha_id,
+            **{name: inputs[name] for name in reversed(call_inputs)},
+        )
+        pd.testing.assert_series_equal(actual, expected)
+
+
+def test_phase4_dispatch_rejects_unknown_missing_extra_and_non_series_inputs():
+    inputs = _phase3_market_inputs()
+
+    with pytest.raises(KeyError, match="not registered"):
+        evaluate_phase4_formula("alpha_050", close=inputs["close"])
+    with pytest.raises(ValueError, match=r"missing inputs.*low"):
+        evaluate_phase4_formula("alpha_051", high=inputs["high"])
+    with pytest.raises(ValueError, match=r"unexpected inputs.*vwap"):
+        evaluate_phase4_formula(
+            "alpha_051",
+            high=inputs["high"],
+            low=inputs["low"],
+            vwap=inputs["vwap"],
+        )
+    with pytest.raises(TypeError, match="high must be a pandas Series"):
+        evaluate_phase4_formula(  # type: ignore[arg-type]
+            "alpha_051",
+            high=[1.0, 2.0],
+            low=inputs["low"],
+        )
+
+
+def test_phase4_dispatch_rejects_implicit_series_alignment():
+    inputs = _phase3_market_inputs()
+    misaligned_low = inputs["low"].rename(index={79: 80})
+
+    with pytest.raises(ValueError, match="low index must align with high"):
+        evaluate_phase4_formula(
+            "alpha_051",
+            high=inputs["high"],
+            low=misaligned_low,
+        )
+
+
+# ── Alpha158 Phase 5: versioned alpha101-alpha150 formula contract ──────────
+
+
+def test_phase5_formula_catalog_is_versioned_exact_and_content_addressed():
+    import hashlib
+    import json
+    from collections import Counter
+
+    expected_ids = tuple(f"alpha_{number:03d}" for number in range(101, 151))
+    expected_fields = {
+        "name",
+        "contract_version",
+        "formula",
+        "category",
+        "complexity",
+        "parameters",
+        "description",
+        "references",
+        "call_inputs",
+        "formula_inputs",
+        "input_category",
+    }
+
+    assert ALPHA158_PHASE5_FORMULA_CONTRACT_VERSION == "1.0.0"
+    assert list_phase5_formulas() == expected_ids
+    assert tuple(ALPHA158_PHASE5_FORMULA_SPECS) == expected_ids
+    assert Counter(
+        spec["input_category"] for spec in ALPHA158_PHASE5_FORMULA_SPECS.values()
+    ) == {"pair": 33, "triple": 14, "quadruple": 3}
+
+    for alpha_id, spec in ALPHA158_PHASE5_FORMULA_SPECS.items():
+        assert set(spec) == expected_fields
+        assert spec["name"] == alpha_id
+        assert spec["contract_version"] == ALPHA158_PHASE5_FORMULA_CONTRACT_VERSION
+        assert spec["formula"] == ALPHA158_REGISTRY[alpha_id]["formula"]
+
+    serializable_specs = {
+        alpha_id: {
+            field: list(value) if isinstance(value, tuple) else value
+            for field, value in spec.items()
+        }
+        for alpha_id, spec in ALPHA158_PHASE5_FORMULA_SPECS.items()
+    }
+    encoded = json.dumps(
+        serializable_specs,
+        sort_keys=True,
+        separators=(",", ":"),
+        ensure_ascii=False,
+    ).encode()
+    assert hashlib.sha256(encoded).hexdigest() == ALPHA158_PHASE5_FORMULA_CATALOG_SHA256
+    assert ALPHA158_PHASE5_FORMULA_CATALOG_SHA256 == (
+        "3368796169c9fbd39c4a34ea137e569964b15882fbf4de25124790d548db6533"
+    )
+
+
+def test_phase5_formula_catalog_is_recursively_immutable():
+    import operator
+
+    with pytest.raises(TypeError):
+        operator.setitem(ALPHA158_PHASE5_FORMULA_SPECS, "alpha_101", {})
+    with pytest.raises(TypeError):
+        operator.setitem(
+            ALPHA158_PHASE5_FORMULA_SPECS["alpha_101"],
+            "formula",
+            "changed",
+        )
+    with pytest.raises(TypeError):
+        operator.setitem(
+            ALPHA158_PHASE5_FORMULA_SPECS["alpha_101"]["call_inputs"],
+            0,
+            "volume",
+        )
+
+
+def test_phase5_catalog_freezes_callable_and_formula_inputs():
+    import inspect
+
+    for alpha_id, spec in ALPHA158_PHASE5_FORMULA_SPECS.items():
+        function = getattr(alpha_factors_module, alpha_id)
+        signature_inputs = tuple(
+            "open" if name == "open_" else name
+            for name in inspect.signature(function).parameters
+        )
+        assert spec["call_inputs"] == signature_inputs
+        assert spec["formula_inputs"] == tuple(ALPHA158_REGISTRY[alpha_id]["inputs"])
+
+    assert ALPHA158_PHASE5_FORMULA_SPECS["alpha_101"]["call_inputs"] == (
+        "close",
+        "high",
+        "low",
+    )
+
+
+def test_phase5_dispatch_matches_all_existing_alpha101_alpha150_functions():
+    inputs = _phase3_market_inputs()
+
+    for alpha_id, spec in ALPHA158_PHASE5_FORMULA_SPECS.items():
+        call_inputs = spec["call_inputs"]
+        function = getattr(alpha_factors_module, alpha_id)
+        expected = function(*(inputs[name] for name in call_inputs))
+        actual = evaluate_phase5_formula(
+            alpha_id,
+            **{name: inputs[name] for name in reversed(call_inputs)},
+        )
+        pd.testing.assert_series_equal(actual, expected)
+
+
+def test_phase5_dispatch_rejects_unknown_missing_extra_and_non_series_inputs():
+    inputs = _phase3_market_inputs()
+
+    with pytest.raises(KeyError, match="not registered"):
+        evaluate_phase5_formula("alpha_100", close=inputs["close"])
+    with pytest.raises(KeyError, match="not registered"):
+        evaluate_phase5_formula("alpha_151", close=inputs["close"])
+    with pytest.raises(ValueError, match=r"missing inputs.*low"):
+        evaluate_phase5_formula(
+            "alpha_101",
+            close=inputs["close"],
+            high=inputs["high"],
+        )
+    with pytest.raises(ValueError, match=r"unexpected inputs.*vwap"):
+        evaluate_phase5_formula(
+            "alpha_101",
+            close=inputs["close"],
+            high=inputs["high"],
+            low=inputs["low"],
+            vwap=inputs["vwap"],
+        )
+    with pytest.raises(TypeError, match="high must be a pandas Series"):
+        evaluate_phase5_formula(  # type: ignore[arg-type]
+            "alpha_101",
+            close=inputs["close"],
+            high=[1.0, 2.0],
+            low=inputs["low"],
+        )
+
+
+def test_phase5_dispatch_rejects_length_and_index_alignment_errors():
+    inputs = _phase3_market_inputs()
+    shorter_low = inputs["low"].iloc[:-1]
+    misaligned_high = inputs["high"].rename(index={79: 80})
+
+    with pytest.raises(ValueError, match="low length must match close"):
+        evaluate_phase5_formula(
+            "alpha_101",
+            close=inputs["close"],
+            high=inputs["high"],
+            low=shorter_low,
+        )
+    with pytest.raises(ValueError, match="high index must align with close"):
+        evaluate_phase5_formula(
+            "alpha_101",
+            close=inputs["close"],
+            high=misaligned_high,
+            low=inputs["low"],
+        )
+
+
+def test_phase5_contract_is_publicly_exported():
+    assert {
+        "ALPHA158_PHASE5_FORMULA_CONTRACT_VERSION",
+        "ALPHA158_PHASE5_FORMULA_CATALOG_SHA256",
+        "ALPHA158_PHASE5_FORMULA_SPECS",
+        "list_phase5_formulas",
+        "evaluate_phase5_formula",
+    } <= set(alpha_factors_module.__all__)
+
+
+# ── Alpha158 Phase 6: versioned alpha151-alpha158 formula contract ──────────
+
+
+def test_phase6_formula_catalog_is_versioned_exact_and_content_addressed():
+    import hashlib
+    import json
+    from collections import Counter
+
+    expected_ids = tuple(f"alpha_{number:03d}" for number in range(151, 159))
+    expected_fields = {
+        "name",
+        "contract_version",
+        "formula",
+        "category",
+        "complexity",
+        "parameters",
+        "description",
+        "references",
+        "call_inputs",
+        "formula_inputs",
+        "input_category",
+    }
+
+    assert ALPHA158_PHASE6_FORMULA_CONTRACT_VERSION == "1.0.0"
+    assert list_phase6_formulas() == expected_ids
+    assert tuple(ALPHA158_PHASE6_FORMULA_SPECS) == expected_ids
+    assert Counter(
+        spec["input_category"] for spec in ALPHA158_PHASE6_FORMULA_SPECS.values()
+    ) == {"pair": 6, "triple": 2}
+
+    for alpha_id, spec in ALPHA158_PHASE6_FORMULA_SPECS.items():
+        assert set(spec) == expected_fields
+        assert spec["name"] == alpha_id
+        assert spec["contract_version"] == ALPHA158_PHASE6_FORMULA_CONTRACT_VERSION
+        assert spec["formula"] == ALPHA158_REGISTRY[alpha_id]["formula"]
+
+    serializable_specs = {
+        alpha_id: {
+            field: list(value) if isinstance(value, tuple) else value
+            for field, value in spec.items()
+        }
+        for alpha_id, spec in ALPHA158_PHASE6_FORMULA_SPECS.items()
+    }
+    encoded = json.dumps(
+        serializable_specs,
+        sort_keys=True,
+        separators=(",", ":"),
+        ensure_ascii=False,
+    ).encode()
+    assert hashlib.sha256(encoded).hexdigest() == ALPHA158_PHASE6_FORMULA_CATALOG_SHA256
+    assert ALPHA158_PHASE6_FORMULA_CATALOG_SHA256 == (
+        "70ffae16ca6cbb59a1e8644f5cdeec1d291fa75ff8b5647fb534af0083408219"
+    )
+
+
+def test_phase6_formula_catalog_is_recursively_immutable():
+    import operator
+
+    with pytest.raises(TypeError):
+        operator.setitem(ALPHA158_PHASE6_FORMULA_SPECS, "alpha_151", {})
+    with pytest.raises(TypeError):
+        operator.setitem(
+            ALPHA158_PHASE6_FORMULA_SPECS["alpha_151"],
+            "formula",
+            "changed",
+        )
+    with pytest.raises(TypeError):
+        operator.setitem(
+            ALPHA158_PHASE6_FORMULA_SPECS["alpha_151"]["call_inputs"],
+            0,
+            "volume",
+        )
+
+
+def test_phase6_catalog_freezes_callable_and_formula_inputs():
+    import inspect
+
+    for alpha_id, spec in ALPHA158_PHASE6_FORMULA_SPECS.items():
+        function = getattr(alpha_factors_module, alpha_id)
+        signature_inputs = tuple(
+            "open" if name == "open_" else name
+            for name in inspect.signature(function).parameters
+        )
+        assert spec["call_inputs"] == signature_inputs
+        assert spec["formula_inputs"] == tuple(ALPHA158_REGISTRY[alpha_id]["inputs"])
+
+    assert ALPHA158_PHASE6_FORMULA_SPECS["alpha_158"]["call_inputs"] == (
+        "high",
+        "low",
+        "volume",
+    )
+
+
+def test_phase6_dispatch_matches_all_existing_alpha151_alpha158_functions():
+    inputs = _phase3_market_inputs()
+
+    for alpha_id, spec in ALPHA158_PHASE6_FORMULA_SPECS.items():
+        call_inputs = spec["call_inputs"]
+        function = getattr(alpha_factors_module, alpha_id)
+        expected = function(*(inputs[name] for name in call_inputs))
+        actual = evaluate_phase6_formula(
+            alpha_id,
+            **{name: inputs[name] for name in reversed(call_inputs)},
+        )
+        pd.testing.assert_series_equal(actual, expected)
+
+
+def test_phase6_dispatch_rejects_unknown_missing_extra_and_non_series_inputs():
+    inputs = _phase3_market_inputs()
+
+    with pytest.raises(KeyError, match="not registered"):
+        evaluate_phase6_formula("alpha_150", close=inputs["close"])
+    with pytest.raises(KeyError, match="not registered"):
+        evaluate_phase6_formula("alpha_159", close=inputs["close"])
+    with pytest.raises(ValueError, match=r"missing inputs.*volume"):
+        evaluate_phase6_formula("alpha_151", close=inputs["close"])
+    with pytest.raises(ValueError, match=r"unexpected inputs.*vwap"):
+        evaluate_phase6_formula(
+            "alpha_151",
+            close=inputs["close"],
+            volume=inputs["volume"],
+            vwap=inputs["vwap"],
+        )
+    with pytest.raises(TypeError, match="volume must be a pandas Series"):
+        evaluate_phase6_formula(  # type: ignore[arg-type]
+            "alpha_151",
+            close=inputs["close"],
+            volume=[1.0, 2.0],
+        )
+
+
+def test_phase6_dispatch_rejects_length_and_index_alignment_errors():
+    inputs = _phase3_market_inputs()
+    shorter_volume = inputs["volume"].iloc[:-1]
+    misaligned_low = inputs["low"].rename(index={79: 80})
+
+    with pytest.raises(ValueError, match="volume length must match close"):
+        evaluate_phase6_formula(
+            "alpha_151",
+            close=inputs["close"],
+            volume=shorter_volume,
+        )
+    with pytest.raises(ValueError, match="low index must align with high"):
+        evaluate_phase6_formula(
+            "alpha_158",
+            high=inputs["high"],
+            low=misaligned_low,
+            volume=inputs["volume"],
+        )
+
+
+def test_phase6_preserves_complete_alpha001_alpha158_formula_body_fingerprint():
+    import ast
+    import hashlib
+    import inspect
+    import json
+    import textwrap
+
+    fingerprints = {}
+    for number in range(1, 159):
+        alpha_id = f"alpha_{number:03d}"
+        source = textwrap.dedent(inspect.getsource(getattr(alpha_factors_module, alpha_id)))
+        node = ast.parse(source).body[0]
+        assert isinstance(node, (ast.FunctionDef, ast.AsyncFunctionDef))
+        body = ast.dump(
+            ast.Module(body=node.body, type_ignores=[]),
+            include_attributes=False,
+        )
+        fingerprints[alpha_id] = hashlib.sha256(body.encode()).hexdigest()
+
+    encoded = json.dumps(
+        fingerprints,
+        sort_keys=True,
+        separators=(",", ":"),
+    ).encode()
+    assert hashlib.sha256(encoded).hexdigest() == (
+        "f3ae807983cf8ca083d0b924c3807ffd84a62a5bd354f9efb1a1a16ca40c7da6"
+    )
+
+
+def test_phase6_contract_is_publicly_exported():
+    assert {
+        "ALPHA158_PHASE6_FORMULA_CONTRACT_VERSION",
+        "ALPHA158_PHASE6_FORMULA_CATALOG_SHA256",
+        "ALPHA158_PHASE6_FORMULA_SPECS",
+        "list_phase6_formulas",
+        "evaluate_phase6_formula",
+    } <= set(alpha_factors_module.__all__)

tests/test_governed_pipeline.py

@@ -0,0 +1,320 @@
+"""Governed Personal Quant OS vertical-slice contracts."""
+
+from __future__ import annotations
+
+from datetime import UTC, datetime
+
+import pandas as pd
+import pytest
+
+from quant_engine.execution import ExecutionConfig
+from quant_engine.governed_pipeline import (
+    DatasetSnapshot,
+    FactorVersion,
+    PaperOrderIntent,
+    RiskDecisionStatus,
+    RiskPolicy,
+    StrategyStage,
+    StrategyVersion,
+    create_paper_order_intent,
+    run_governed_factor_slice,
+)
+
+
+def _calendar() -> pd.DatetimeIndex:
+    return pd.date_range("2026-01-05", periods=4, freq="B")
+
+
+def _scores() -> pd.DataFrame:
+    dates = _calendar()
+    return pd.DataFrame(
+        {"A": [3.0, 1.0], "B": [2.0, 3.0], "C": [1.0, 2.0]},
+        index=dates[:2],
+    )
+
+
+def _prices() -> tuple[pd.DataFrame, pd.DataFrame]:
+    dates = _calendar()
+    opens = pd.DataFrame(
+        {"A": [10.0, 10.0, 10.2, 10.4], "B": [20.0, 20.0, 20.5, 21.0], "C": [30.0, 30.0, 30.0, 30.0]},
+        index=dates,
+    )
+    closes = opens * 1.01
+    return opens, closes
+
+
+def _snapshot() -> DatasetSnapshot:
+    return DatasetSnapshot(
+        snapshot_id="dataset:cn-a-daily-20260108-v1",
+        schema_version="1.0.0",
+        content_sha256="a" * 64,
+        effective_at=datetime(2026, 1, 8, 7, tzinfo=UTC),
+        available_at=datetime(2026, 1, 8, 8, tzinfo=UTC),
+        ingested_at=datetime(2026, 1, 8, 8, 5, tzinfo=UTC),
+    )
+
+
+def _factor() -> FactorVersion:
+    return FactorVersion(
+        factor_id="factor:demo-momentum",
+        version="1.0.0",
+        definition_sha256="b" * 64,
+        dataset_schema_version="1.0.0",
+    )
+
+
+def _strategy() -> StrategyVersion:
+    return StrategyVersion(
+        strategy_id="strategy:demo-top2",
+        version="1.0.0",
+        factor_version_id="factor:demo-momentum@1.0.0",
+        stage=StrategyStage.APPROVED,
+    )
+
+
+def _execution_config() -> ExecutionConfig:
+    return ExecutionConfig(
+        commission_bps=0,
+        stamp_tax_bps=0,
+        slippage_bps=0,
+        min_trade_amount=0,
+    )
+
+
+def test_governed_slice_is_reproducible_and_creates_only_paper_intent() -> None:
+    opens, closes = _prices()
+    created_at = datetime(2026, 1, 9, 1, tzinfo=UTC)
+    policy = RiskPolicy(
+        policy_id="risk:paper-default@1.0.0",
+        max_gross_exposure=1.0,
+        max_single_asset_weight=0.6,
+        max_positions=10,
+    )
+
+    result = run_governed_factor_slice(
+        factor_scores=_scores(),
+        execution_prices=opens,
+        valuation_prices=closes,
+        dataset_snapshot=_snapshot(),
+        factor_version=_factor(),
+        strategy_version=_strategy(),
+        risk_policy=policy,
+        code_revision="c" * 40,
+        created_at=created_at,
+        top_k=2,
+        execution_price_field="open",
+        valuation_price_field="close",
+        execution_config=_execution_config(),
+    )
+
+    assert result.backtest_run.dataset_snapshot_id == _snapshot().snapshot_id
+    assert result.backtest_run.factor_version_id == _factor().version_id
+    assert result.backtest_run.strategy_version_id == _strategy().version_id
+    assert result.backtest_run.code_revision == "c" * 40
+    assert len(result.backtest_run.config_hash) == 64
+    assert result.portfolio_target.backtest_run_id == result.backtest_run.run_id
+    assert result.risk_decision.status is RiskDecisionStatus.APPROVED
+    assert result.risk_decision.portfolio_target_id == result.portfolio_target.target_id
+    assert result.order_intent is not None
+    assert result.order_intent.environment == "paper"
+    assert result.order_intent.risk_decision_id == result.risk_decision.decision_id
+    assert result.order_intent.portfolio_target_id == result.portfolio_target.target_id
+
+    repeated = run_governed_factor_slice(
+        factor_scores=_scores(),
+        execution_prices=opens,
+        valuation_prices=closes,
+        dataset_snapshot=_snapshot(),
+        factor_version=_factor(),
+        strategy_version=_strategy(),
+        risk_policy=policy,
+        code_revision="c" * 40,
+        created_at=created_at,
+        top_k=2,
+        execution_price_field="open",
+        valuation_price_field="close",
+        execution_config=_execution_config(),
+    )
+    assert repeated.backtest_run.run_id == result.backtest_run.run_id
+    assert repeated.portfolio_target.target_id == result.portfolio_target.target_id
+    assert repeated.risk_decision.decision_id == result.risk_decision.decision_id
+    assert repeated.order_intent == result.order_intent
+
+
+def test_risk_rejection_blocks_order_intent() -> None:
+    opens, closes = _prices()
+    result = run_governed_factor_slice(
+        factor_scores=_scores(),
+        execution_prices=opens,
+        valuation_prices=closes,
+        dataset_snapshot=_snapshot(),
+        factor_version=_factor(),
+        strategy_version=_strategy(),
+        risk_policy=RiskPolicy(
+            policy_id="risk:no-concentration@1.0.0",
+            max_gross_exposure=1.0,
+            max_single_asset_weight=0.4,
+            max_positions=10,
+        ),
+        code_revision="c" * 40,
+        created_at=datetime(2026, 1, 9, 1, tzinfo=UTC),
+        top_k=2,
+        execution_price_field="open",
+        valuation_price_field="close",
+        execution_config=_execution_config(),
+    )
+
+    assert result.risk_decision.status is RiskDecisionStatus.REJECTED
+    assert any("single-asset weight" in reason for reason in result.risk_decision.reasons)
+    assert result.order_intent is None
+    with pytest.raises(ValueError, match="approved risk decision"):
+        create_paper_order_intent(result.portfolio_target, result.risk_decision)
+    with pytest.raises(ValueError, match="approved risk decision"):
+        PaperOrderIntent(result.portfolio_target, result.risk_decision)
+
+
+def test_dataset_snapshot_requires_point_in_time_ordering_and_aware_times() -> None:
+    with pytest.raises(ValueError, match="timezone-aware"):
+        DatasetSnapshot(
+            snapshot_id="dataset:invalid",
+            schema_version="1.0.0",
+            content_sha256="a" * 64,
+            effective_at=datetime(2026, 1, 8, 7),
+            available_at=datetime(2026, 1, 8, 8, tzinfo=UTC),
+            ingested_at=datetime(2026, 1, 8, 9, tzinfo=UTC),
+        )
+
+    with pytest.raises(ValueError, match="effective_at <= available_at <= ingested_at"):
+        DatasetSnapshot(
+            snapshot_id="dataset:invalid",
+            schema_version="1.0.0",
+            content_sha256="a" * 64,
+            effective_at=datetime(2026, 1, 8, 9, tzinfo=UTC),
+            available_at=datetime(2026, 1, 8, 8, tzinfo=UTC),
+            ingested_at=datetime(2026, 1, 8, 10, tzinfo=UTC),
+        )
+
+
+def test_strategy_factor_lineage_must_match() -> None:
+    opens, closes = _prices()
+    mismatched = StrategyVersion(
+        strategy_id="strategy:demo-top2",
+        version="1.0.0",
+        factor_version_id="factor:other@1.0.0",
+        stage=StrategyStage.APPROVED,
+    )
+
+    with pytest.raises(ValueError, match="factor lineage"):
+        run_governed_factor_slice(
+            factor_scores=_scores(),
+            execution_prices=opens,
+            valuation_prices=closes,
+            dataset_snapshot=_snapshot(),
+            factor_version=_factor(),
+            strategy_version=mismatched,
+            risk_policy=RiskPolicy(
+                policy_id="risk:paper-default@1.0.0",
+                max_gross_exposure=1.0,
+                max_single_asset_weight=0.6,
+                max_positions=10,
+            ),
+            code_revision="c" * 40,
+            created_at=datetime(2026, 1, 9, 1, tzinfo=UTC),
+            top_k=2,
+            execution_price_field="open",
+            valuation_price_field="close",
+            execution_config=_execution_config(),
+        )
+
+
+def test_governed_slice_requires_matching_schema_and_snapshot_available_by_run_time() -> None:
+    opens, closes = _prices()
+    common = {
+        "factor_scores": _scores(),
+        "execution_prices": opens,
+        "valuation_prices": closes,
+        "strategy_version": _strategy(),
+        "risk_policy": RiskPolicy(
+            policy_id="risk:paper-default@1.0.0",
+            max_gross_exposure=1.0,
+            max_single_asset_weight=0.6,
+            max_positions=10,
+        ),
+        "code_revision": "c" * 40,
+        "top_k": 2,
+        "execution_price_field": "open",
+        "valuation_price_field": "close",
+        "execution_config": _execution_config(),
+    }
+
+    with pytest.raises(ValueError, match="dataset schema"):
+        run_governed_factor_slice(
+            dataset_snapshot=_snapshot(),
+            factor_version=FactorVersion(
+                factor_id="factor:demo-momentum",
+                version="1.0.0",
+                definition_sha256="b" * 64,
+                dataset_schema_version="2.0.0",
+            ),
+            created_at=datetime(2026, 1, 9, 1, tzinfo=UTC),
+            **common,
+        )
+
+    with pytest.raises(ValueError, match="available before the research run"):
+        run_governed_factor_slice(
+            dataset_snapshot=_snapshot(),
+            factor_version=_factor(),
+            created_at=datetime(2026, 1, 8, 7, 30, tzinfo=UTC),
+            **common,
+        )
+
+    future_scores = _scores()
+    future_scores.index = pd.date_range("2026-01-12", periods=2, freq="B")
+    with pytest.raises(ValueError, match="future decision dates"):
+        run_governed_factor_slice(
+            dataset_snapshot=_snapshot(),
+            factor_version=_factor(),
+            factor_scores=future_scores,
+            execution_prices=opens,
+            valuation_prices=closes,
+            strategy_version=common["strategy_version"],
+            risk_policy=common["risk_policy"],
+            code_revision="c" * 40,
+            created_at=datetime(2026, 1, 9, 1, tzinfo=UTC),
+            top_k=2,
+            execution_price_field="open",
+            valuation_price_field="close",
+            execution_config=_execution_config(),
+        )
+
+
+def test_paper_intent_requires_approved_strategy_stage() -> None:
+    opens, closes = _prices()
+    validated = StrategyVersion(
+        strategy_id="strategy:demo-top2",
+        version="1.0.0",
+        factor_version_id=_factor().version_id,
+        stage=StrategyStage.VALIDATED,
+    )
+
+    with pytest.raises(ValueError, match="Approved or Paper"):
+        run_governed_factor_slice(
+            factor_scores=_scores(),
+            execution_prices=opens,
+            valuation_prices=closes,
+            dataset_snapshot=_snapshot(),
+            factor_version=_factor(),
+            strategy_version=validated,
+            risk_policy=RiskPolicy(
+                policy_id="risk:paper-default@1.0.0",
+                max_gross_exposure=1.0,
+                max_single_asset_weight=0.6,
+                max_positions=10,
+            ),
+            code_revision="c" * 40,
+            created_at=datetime(2026, 1, 9, 1, tzinfo=UTC),
+            top_k=2,
+            execution_price_field="open",
+            valuation_price_field="close",
+            execution_config=_execution_config(),
+        )