Rank213 long-history review with funding（frozen spec）

目标：在完全冻结当前 spec 下，把历史资金费率与结算时间点计入回测。

spec 锚点：f64_h12_floor150_mult2p0 / 30币 universe / 15m / 64 / 12 / top3-bottom3 / max(1.5%,2.0×median) / 4bps×turnover_x。

不含 funding 的长历史页 · 短样本页 · honesty_audit · ffill_impact_audit · readiness_note · asof_universe_long_history_review · regime_review

1) 回测区间与数据可用性（含本地优先策略）

common start

2025-10-24T08:30:00Z

common end

2026-04-10T23:45:00Z

bars

16190

calendar days

168.64d

rebalances

1343

价格数据：local cache 优先，缺失部分再从 data.binance.vision 月/日 zip 补齐； funding 数据：local cache 优先，月包补齐后再对当前月用 API 增量拉取；API 遇 418/429 按指数退避重试（最多 8 次）。

2) plain baseline vs veto（资金费率已计入）

plain net mean

1.90 bps

plain net cumulative

-15.97%

plain win rate

49.96%

plain max drawdown

-78.73%

plain funding mean

0.57 bps

veto net mean

10.49 bps

veto net cumulative

227.23%

veto win rate

49.89%

veto max drawdown

-37.16%

veto funding mean

1.53 bps

3) 1Y / 2Y / 3Y 冻结窗口验证（资金费率已计入）

window	available	required_start_utc	available_start_utc	note
1Y	no	2025-04-10T23:45:00Z	2025-10-24T08:30:00Z	common-history shorter than requested window under frozen universe
2Y	no	2024-04-10T23:45:00Z	2025-10-24T08:30:00Z	common-history shorter than requested window under frozen universe
3Y	no	2023-04-11T23:45:00Z	2025-10-24T08:30:00Z	common-history shorter than requested window under frozen universe

窗口不可用就直接标 no，不改 universe，不调参数，不做“扩样本后重调”。

4) baseline vs veto 的增量是否稳定

Δ net mean

8.60 bps

Δ net cumulative

243.21%

Δ funding mean

0.96 bps

Δ avg turnover

0.282x

MDD reduction

41.57%

half-split stability

half	start_utc	end_utc	rebalances	plain_net_mean_bps	plain_net_cum_pct	veto_net_mean_bps	veto_net_cum_pct	delta_net_mean_bps	delta_net_cum_pct
1	2025-10-25T00:30:00Z	2026-01-16T21:30:00Z	672	-1.99 bps	-22.87%	5.30 bps	30.41%	7.29 bps	53.28%
2	2026-01-17T00:30:00Z	2026-04-10T18:30:00Z	671	5.79 bps	8.94%	15.69 bps	150.92%	9.90 bps	141.98%

5) 按月 / 按季度 / 按年表现（资金费率已计入）

按月

month	rebalances	plain_net_mean_bps	plain_net_cum_pct	plain_win_rate	plain_max_dd_pct	plain_avg_turnover_x	plain_funding_mean_bps	veto_net_mean_bps	veto_net_cum_pct	veto_win_rate	veto_max_dd_pct	veto_avg_turnover_x	veto_funding_mean_bps
2025-10	56	5.84 bps	2.16%	55.36%	-15.25%	1.000x	0.27 bps	3.38 bps	1.03%	51.79%	-14.10%	1.310x	0.12 bps
2025-11	240	-3.40 bps	-13.29%	48.33%	-34.05%	1.000x	0.68 bps	4.43 bps	6.32%	49.17%	-21.94%	1.212x	0.65 bps
2025-12	248	7.08 bps	15.18%	49.19%	-25.71%	1.000x	0.36 bps	11.26 bps	28.82%	47.18%	-19.87%	1.276x	2.24 bps
2026-01	248	-8.52 bps	-23.94%	51.21%	-29.00%	1.000x	2.24 bps	8.85 bps	21.14%	51.61%	-15.00%	1.272x	3.93 bps
2026-02	224	-35.76 bps	-60.91%	47.77%	-64.96%	1.000x	-0.75 bps	-8.36 bps	-20.16%	47.32%	-26.64%	1.214x	-0.31 bps
2026-03	248	30.23 bps	95.47%	51.21%	-26.01%	1.000x	0.19 bps	23.11 bps	67.87%	52.42%	-22.34%	1.352x	1.20 bps
2026-04	79	49.45 bps	41.70%	51.90%	-18.17%	1.000x	0.77 bps	50.55 bps	45.66%	53.16%	-7.38%	1.498x	1.67 bps

按季度

quarter	rebalances	plain_net_mean_bps	plain_net_cum_pct	plain_win_rate	plain_max_dd_pct	plain_avg_turnover_x	plain_funding_mean_bps	veto_net_mean_bps	veto_net_cum_pct	veto_win_rate	veto_max_dd_pct	veto_avg_turnover_x	veto_funding_mean_bps
2025Q4	544	2.33 bps	2.03%	49.45%	-36.49%	1.000x	0.49 bps	7.44 bps	38.37%	48.53%	-21.94%	1.251x	1.32 bps
2026Q1	720	-3.65 bps	-41.88%	50.14%	-74.36%	1.000x	0.60 bps	8.41 bps	62.35%	50.56%	-37.16%	1.281x	1.67 bps
2026Q2	79	49.45 bps	41.70%	51.90%	-18.17%	1.000x	0.77 bps	50.55 bps	45.66%	53.16%	-7.38%	1.498x	1.67 bps

按年

year	rebalances	plain_net_mean_bps	plain_net_cum_pct	plain_win_rate	plain_max_dd_pct	plain_avg_turnover_x	plain_funding_mean_bps	veto_net_mean_bps	veto_net_cum_pct	veto_win_rate	veto_max_dd_pct	veto_avg_turnover_x	veto_funding_mean_bps
2025	544	2.33 bps	2.03%	49.45%	-36.49%	1.000x	0.49 bps	7.44 bps	38.37%	48.53%	-21.94%	1.251x	1.32 bps
2026	799	1.60 bps	-17.64%	50.31%	-74.36%	1.000x	0.62 bps	12.57 bps	136.49%	50.81%	-37.16%	1.303x	1.67 bps

6) 不同年份里 veto 是改善 short leg，还是只是降频

year	rebalances	plain_short_price_mean_bps	veto_short_price_mean_bps	delta_short_price_bps	plain_short_funding_mean_bps	veto_short_funding_mean_bps	delta_short_funding_bps	plain_avg_turnover_x	veto_avg_turnover_x	delta_turnover_x
2025	544	-3.30 bps	1.98 bps	5.28 bps	-0.99 bps	-0.16 bps	0.83 bps	1.000x	1.251x	0.251x
2026	799	-2.98 bps	8.16 bps	11.13 bps	-1.54 bps	-0.49 bps	1.05 bps	1.000x	1.303x	0.303x

同时展示 short leg 的价格贡献与 funding 贡献；如果 delta_short_price 为正且并非靠显著降频（delta_turnover 不为负），则更接近真实 short-leg 改善。

7) veto 命中率 / 误杀率（定义同旧页）

pct rebalances with any veto：60.01%
avg veto count / rebalance：0.846
total vetoed names：1136
name-level hit rate：47.36%
name-level false-kill rate：52.02%

8) 最终 verdict

在冻结当前 universe/spec 下，公共共同历史只到 2025-10-24T08:30:00Z，仍不足 1Y。因此本轮（含 funding 结算计提）不能宣称已通过 1Y/2Y/3Y。可确认的是：在最大可用共同历史上，veto 在 funding-adjusted 口径下仍明显优于 plain。