关于「实证闭环:收单风控台账、实战工具与检查器规格」,《收单风控阶跃与准备金生存》给出的定量结论是什么?

结论:第八章给出收单风控台账、蒸馏小工具「收单风控红线与滚动准备金生存试算器」、按四件套分组的检查器规格表与预注册检验,退款台检验每组需要 63,900 笔交易。书中以标注框区分已发表的证据、可运行的 Python 计算(22 段程序)、原典与作者的假说。

数据口径:书中参数为示意,全部计算由书中程序生成。 本书是非个人化的经营与风险分析,不构成法律、融资或投资建议,不向任何人承诺收入。

English version

对照数据表

下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 63900

T1 payments per arm 63,900; months at 6,000 a month 10.7
T2 payments per arm 269,726; months at 6,000 a month 45.0
T3 days of monitoring 950; reject ARL0 365 at 6 alarms
T4 zero-breach seasons for share < 0.05: 59
T4 zero-breach seasons for share < 0.10: 29
查看完整验算程序
# Chapter 8. p82: sample sizes for the pre-registered tests.
# T1 refund desk: dispute rate 0.55 % without the desk against 0.44 %
#    with it (x = 0.20 in p42); two-sided alpha 0.05, power 0.80,
#    normal approximation for two proportions, payments per arm.
# T2 the same with a smaller effect, 0.55 % against 0.495 % (x = 0.10).
# T3 CUSUM calibration: are false alarms as rare as ARL0 = 365 days?
#    Days of monitoring with no true change needed for a one-sided
#    Poisson test to tell a rate of 1/365 from 1/120 a day at alpha
#    0.05 and power 0.80 (exact Poisson tails, searched day by day).
# T4 buffer adequacy: campaign seasons with zero breaches needed to
#    claim, at 95 % confidence, a breach share below 0.05 or 0.10.
import math

Z_A, Z_B = 1.959964, 0.841621


def n_two_prop(p1, p2):
    pb = 0.5 * (p1 + p2)
    a = Z_A * math.sqrt(2.0 * pb * (1.0 - pb))
    b = Z_B * math.sqrt(p1 * (1.0 - p1) + p2 * (1.0 - p2))
    return math.ceil(((a + b) / (p1 - p2)) ** 2)


def pois_sf(k, lam):
    """P(X >= k) for X ~ Poisson(lam)"""
    terms = [math.exp(-lam + j * math.log(lam) - math.lgamma(j + 1))
             for j in range(0, k)]
    return 1.0 - math.fsum(terms)


def days_needed(r0, r1):
    for days in range(30, 20000, 10):
        lam0, lam1 = r0 * days, r1 * days
        k = 0
        while pois_sf(k, lam0) > 0.05:
            k += 1
        if pois_sf(k, lam1) >= 0.80:
            return days, k
    return None, None


def main():
    for name, p1, p2 in (("T1", 0.0055, 0.0044), ("T2", 0.0055, 0.00495)):
        n = n_two_prop(p1, p2)
        print("{} payments per arm {:,}; months at 6,000 a month {:.1f}"
              .format(name, n, n / 6000.0))
    d, k = days_needed(1 / 365.0, 1 / 120.0)
    print("T3 days of monitoring {:,}; reject ARL0 365 at {} alarms".format(
        d, k))
    for cap in (0.05, 0.10):
        n = math.ceil(math.log(0.05) / math.log(1.0 - cap))
        print("T4 zero-breach seasons for share < {:.2f}: {}".format(cap, n))


if __name__ == "__main__":
    main()

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《数字商品收单风控阶跃模型、多网关热备切换与防冻卡滚动准备金母典》第 8 章「实证闭环:收单风控台账、实战工具与检查器规格」(版本 1.0.0). 5D Library. DOI 待发布

DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-397 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026397,
  author    = {Fan, Yafu},
  title     = {Payment-Acquirer Risk Step-Functions, Multi-Gateway Hot Failover and Rolling-Reserve Survival for Digital Fleets},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论