EN
MACRO-02 · SPY

风险偏好开关 · 策略证据

查看公开成绩对应的数据标识、实际参数和收益计算记录。

下载复核材料 返回策略页
运行编号f14682b033d9
最后纳入的价格日期2026-11-10
证据包生成时间（UTC）2026-11-11 10:55 UTC
计算模型模拟每日目标配置 · 第 3 版
下载以下材料，可复核记录的收益与净值。重算策略决策需使用一致的原始输入，重新下载的行情可能不同。这里记录的是模拟仓位，不是券商成交。

下载复核材料

仅在点击下载时加载并校验压缩文件。负权重表示空头敞口；浏览器不会执行归档代码。

1. 下载完整复核包

下载上方 Python 工具后运行此命令。工具校验共享文件并生成 replay.zip，无需克隆代码仓库。

python e0543584b55263609cc8fd38cd032cb8432af6c2b48da3f53599e6e68c922f59.py https://quantarenas.com/data/evidence/runs/f14682b033d9f898d5c08eba4d8f7ece05b3ca7c0e12941fab8699a66069bf30/manifest.json --bundle replay.zip

2. 复核收益计算

解压 replay.zip，在解压目录用 Python 运行。仅需标准库，即可校验已记录权重、收益贡献和净值；此步骤不重算策略决策。

python scripts/replay_evidence.py data/evidence/runs/f14682b033d9f898d5c08eba4d8f7ece05b3ca7c0e12941fab8699a66069bf30/manifest.json

3. 获取并比对输入

在解压目录中使用记录的 Python 版本并安装锁定依赖。此命令将行情输入下载到本地，报告 inputHashMatches。只有哈希一致才能支持精确重算；供应商修订的数据可能不匹配。也可自行提供一致的原始快照，命名为 private-input.json。

python -m pip install -r requirements-lock.txt
python scripts/recompute_evidence.py data/evidence/runs/f14682b033d9f898d5c08eba4d8f7ece05b3ca7c0e12941fab8699a66069bf30/manifest.json --fetch-input private-input.json

4. 重算原始策略

需要完全一致的原始输入快照和记录的依赖版本，才能运行归档策略并重新滚动选参。原始行情输入不公开；重新下载的数据如有修订，将无法通过哈希匹配，不能声称精确复现。

python scripts/recompute_evidence.py data/evidence/runs/f14682b033d9f898d5c08eba4d8f7ece05b3ca7c0e12941fab8699a66069bf30/manifest.json --input private-input.json
收益计算假设

当日收益使用前一日收盘时的目标仓位计算；新的目标仓位在当日收盘时确定。这是模拟收盘价模型，不是实际测得的成交价格。

现金余额权重
现金权重等于一减去净资产权重；使用空头或杠杆时，现金可能超过百分之百或为负。
决策时间
目标权重使用截至当日收盘的行情；前一日目标权重获得本日收盘到收盘的收益。
交易费用（基点）
0
成交假设
模拟每日目标配置，不是券商订单或成交；未计融资和借券费用。
模拟调仓换手
将目标权重与当日收益后漂移的权重比较，按差的绝对值求和；表示收盘时风险资产单边调仓金额占交易前净值的比例。
收益区间
将各资产的前一日目标权重乘以本日收盘到收盘的收益，再求和；现金收益为零。
目标权重变化
各资产新旧目标权重差的绝对值之和，仅衡量策略目标变化。
滚动选参切换
候选策略保留其假设历史状态；测试段首日收益使用此前已发布的目标权重。

实际参数

{
  "active": {
    "mom_lookback": 63,
    "trend_lookback": 50,
    "vol_lookback": 16,
    "vol_threshold": 0.2
  },
  "defaults": {
    "mom_lookback": 63,
    "trend_lookback": 50,
    "vol_lookback": 20,
    "vol_threshold": 0.2
  }
}

滚动前推选参记录

默认参数或首次滚动前推之前的历史不是样本外证据。各折分别记录训练和测试边界；选中的候选规则保留自身的假设状态。

1 · 2020-11-09T00:00:00 → 2021-05-07T00:00:00
{
  "config": {
    "mom_lookback": 50,
    "trend_lookback": 50,
    "vol_lookback": 16,
    "vol_threshold": 0.2
  },
  "isSharpe": 1.3351238737528563,
  "run": 0,
  "testEnd": "2021-05-07T00:00:00",
  "testStart": "2020-11-09T00:00:00",
  "trainEnd": "2020-11-06T00:00:00",
  "trainStart": "2017-11-09T00:00:00"
}
2 · 2021-05-10T00:00:00 → 2021-11-10T00:00:00
{
  "config": {
    "mom_lookback": 50,
    "trend_lookback": 50,
    "vol_lookback": 16,
    "vol_threshold": 0.2
  },
  "isSharpe": 2.2302387587446657,
  "run": 0,
  "testEnd": "2021-11-10T00:00:00",
  "testStart": "2021-05-10T00:00:00",
  "trainEnd": "2021-05-07T00:00:00",
  "trainStart": "2018-05-10T00:00:00"
}
3 · 2021-11-11T00:00:00 → 2022-05-11T00:00:00
{
  "config": {
    "mom_lookback": 50,
    "trend_lookback": 50,
    "vol_lookback": 16,
    "vol_threshold": 0.2
  },
  "isSharpe": 2.1940552859797924,
  "run": 0,
  "testEnd": "2022-05-11T00:00:00",
  "testStart": "2021-11-11T00:00:00",
  "trainEnd": "2021-11-10T00:00:00",
  "trainStart": "2018-11-12T00:00:00"
}
4 · 2022-05-12T00:00:00 → 2022-11-11T00:00:00
{
  "config": {
    "mom_lookback": 50,
    "trend_lookback": 50,
    "vol_lookback": 16,
    "vol_threshold": 0.2
  },
  "isSharpe": 1.7085846254650672,
  "run": 0,
  "testEnd": "2022-11-11T00:00:00",
  "testStart": "2022-05-12T00:00:00",
  "trainEnd": "2022-05-11T00:00:00",
  "trainStart": "2019-05-13T00:00:00"
}
5 · 2022-11-14T00:00:00 → 2023-05-12T00:00:00
{
  "config": {
    "mom_lookback": 50,
    "trend_lookback": 50,
    "vol_lookback": 16,
    "vol_threshold": 0.2
  },
  "isSharpe": 1.2777940671481374,
  "run": 0,
  "testEnd": "2023-05-12T00:00:00",
  "testStart": "2022-11-14T00:00:00",
  "trainEnd": "2022-11-11T00:00:00",
  "trainStart": "2019-11-14T00:00:00"
}
6 · 2023-05-15T00:00:00 → 2023-11-15T00:00:00
{
  "config": {
    "mom_lookback": 50,
    "trend_lookback": 50,
    "vol_lookback": 16,
    "vol_threshold": 0.2
  },
  "isSharpe": 1.1905514822496368,
  "run": 0,
  "testEnd": "2023-11-15T00:00:00",
  "testStart": "2023-05-15T00:00:00",
  "trainEnd": "2023-05-12T00:00:00",
  "trainStart": "2020-05-15T00:00:00"
}
7 · 2023-11-16T00:00:00 → 2024-05-16T00:00:00
{
  "config": {
    "mom_lookback": 76,
    "trend_lookback": 50,
    "vol_lookback": 16,
    "vol_threshold": 0.2
  },
  "isSharpe": 1.1618922240509288,
  "run": 2,
  "testEnd": "2024-05-16T00:00:00",
  "testStart": "2023-11-16T00:00:00",
  "trainEnd": "2023-11-15T00:00:00",
  "trainStart": "2020-11-16T00:00:00"
}
8 · 2024-05-17T00:00:00 → 2024-11-15T00:00:00
{
  "config": {
    "mom_lookback": 76,
    "trend_lookback": 50,
    "vol_lookback": 24,
    "vol_threshold": 0.2
  },
  "isSharpe": 0.6339514463480073,
  "run": 8,
  "testEnd": "2024-11-15T00:00:00",
  "testStart": "2024-05-17T00:00:00",
  "trainEnd": "2024-05-16T00:00:00",
  "trainStart": "2021-05-17T00:00:00"
}
9 · 2024-11-18T00:00:00 → 2025-05-16T00:00:00
{
  "config": {
    "mom_lookback": 76,
    "trend_lookback": 50,
    "vol_lookback": 24,
    "vol_threshold": 0.2
  },
  "isSharpe": 0.46518589964486023,
  "run": 8,
  "testEnd": "2025-05-16T00:00:00",
  "testStart": "2024-11-18T00:00:00",
  "trainEnd": "2024-11-15T00:00:00",
  "trainStart": "2021-11-18T00:00:00"
}
10 · 2025-05-19T00:00:00 → 2025-11-19T00:00:00
{
  "config": {
    "mom_lookback": 76,
    "trend_lookback": 50,
    "vol_lookback": 16,
    "vol_threshold": 0.2
  },
  "isSharpe": 0.8961589182137517,
  "run": 2,
  "testEnd": "2025-11-19T00:00:00",
  "testStart": "2025-05-19T00:00:00",
  "trainEnd": "2025-05-16T00:00:00",
  "trainStart": "2022-05-19T00:00:00"
}
11 · 2025-11-20T00:00:00 → 2026-05-20T00:00:00
{
  "config": {
    "mom_lookback": 50,
    "trend_lookback": 50,
    "vol_lookback": 16,
    "vol_threshold": 0.2
  },
  "isSharpe": 1.1748543757506411,
  "run": 0,
  "testEnd": "2026-05-20T00:00:00",
  "testStart": "2025-11-20T00:00:00",
  "trainEnd": "2025-11-19T00:00:00",
  "trainStart": "2022-11-21T00:00:00"
}
12 · 2026-05-21T00:00:00 → 2026-11-10T00:00:00
{
  "config": {
    "mom_lookback": 63,
    "trend_lookback": 50,
    "vol_lookback": 16,
    "vol_threshold": 0.2
  },
  "isSharpe": 1.1955412426707177,
  "run": 1,
  "testEnd": "2026-11-10T00:00:00",
  "testStart": "2026-05-21T00:00:00",
  "trainEnd": "2026-05-20T00:00:00",
  "trainStart": "2023-05-22T00:00:00"
}

运行摘要

价格观测数2259
期末净值4.81393
累计收益381.39%
目标权重变化天数152
累计目标权重变化19100.00%
累计模拟调仓换手21250.00%
目标总敞口100.00%
目标净敞口100.00%
目标现金余额权重0.00%
生效总敞口100.00%
生效净敞口100.00%
生效现金余额权重0.00%

最新资产权重

市场目标生效
BTC-USD0.00%0.00%
DBA0.00%0.00%
DBC0.00%0.00%
ETH-USD0.00%0.00%
EURUSD=X0.00%0.00%
GLD100.00%100.00%
SLV0.00%0.00%
SPY0.00%0.00%
UNG0.00%0.00%
USO0.00%0.00%

目标权重变化衡量策略目标本身的变化。模拟调仓换手还计入收益产生的权重漂移，因此目标不变也可能需要调仓。现金为扣除净敞口后的带符号余额。

相同输入下的计算方式差异

{
  "changedReturnDays": 1,
  "correctedTerminalEquity": 4.813925247680104,
  "legacyTerminalEquity": 4.775707226021295,
  "maximumDailyReturnDelta": 0.007977978653540366,
  "model": "coherent-lagged-target-v2",
  "terminalEquityDelta": 0.0382180216588095
}
完整运行标识与技术记录
{
  "inputHash": "4b4f01f22cb30b527cfc6ae07b12bf366baa28e8070df946d9f2be7d2ca083cb",
  "ruleVersion": null,
  "sameInputAccountingCorrection": {
    "changedReturnDays": 1,
    "correctedTerminalEquity": 4.813925247680104,
    "legacyTerminalEquity": 4.775707226021295,
    "maximumDailyReturnDelta": 0.007977978653540366,
    "model": "coherent-lagged-target-v2",
    "terminalEquityDelta": 0.0382180216588095
  },
  "sourceHashes": {
    "scripts/build_data.py": "e3dcc988d80a46f7f5a24396bb4ffcac240da8e1249de18a5a282eacccc52320",
    "scripts/enrich_metrics.py": "05611fa9c6de11c261f631cdd8a3314212e7c567f24878a396d057d88752051c",
    "scripts/evidence_v2.py": "c4c6523dbbb9513464dca25150f42cbaac7cbfd71bd8afdc855659b2c763d72e",
    "scripts/performance_io.py": "42282ff10697ceb56dc7ca2edf79538c2e1e8344da8e6d69ceac4d16ba5c668a",
    "scripts/recompute_evidence.py": "a7f59004cb7c356940ee3748439024cd83cac434acb90c0bead2ae19f81be127",
    "scripts/replay_evidence.py": "4e04d937c19b9040da6dc9c3c80ac22181b45ffba0131f3a63a4def95865fad6",
    "scripts/requirements.txt": "7424048072a2cf3d8c3b4e4f37a5b36c1e7ff3732d893c43295d17d85c5f97d1",
    "scripts/strategies/risk_on_off.py": "ab96481380a362512e1e36cfe9ccaa0aa66cc3b338e8092be11c54de85ce4f4b",
    "scripts/strategy_evidence.py": "a3a93d62a84298470af3eb8e7d74e9f0ca0d29cf672d1d2d5449b41904e3d58c"
  },
  "runId": "f14682b033d9f898d5c08eba4d8f7ece05b3ca7c0e12941fab8699a66069bf30",
  "generatedAt": "2026-11-11T10:55:31.449953+00:00",
  "executionModel": {
    "cash": "1 - sum(weights); residual may exceed one or be negative for shorts/leverage.",
    "decision": "Target weights use bars through close t; previous targets earn (t-1,t].",
    "feesBasisPoints": 0,
    "fills": "Synthetic daily target allocations, not broker orders or fills; financing/borrow costs omitted.",
    "id": "synthetic-daily-target-v3",
    "rebalanceTurnover": "Sum(abs(target[t] - target[t-1]*(1+assetReturn[t])/(1+grossReturn[t]))); one-way risky notional at close t after returns, divided by pre-trade equity.",
    "return": "Sum of previous target weight times each asset close-to-close return; cash return is zero.",
    "targetChange": "Sum(abs(target[t] - target[t-1])); changes in rule targets only.",
    "wfo": "Candidate retains its own hypothetical state; first test return uses previous published target."
  }
}

订阅每周 Arena 周报

每周一封邮件：排名变化、新信号和新研究。不发垃圾邮件，随时可以退订。