关于「准备金:滚动扣留、结算冻结与现金不断血的最低缓冲」,《收单风控阶跃与准备金生存》给出的定量结论是什么?

结论:第六章算出一成、九十天的滚动准备金只需 2,900 的缓冲,而冻结三十天需要 93,000,缓冲留足 50,000 时仍有 0.0175 的年份会用光。

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: 2900, 93000

scenariobufferworst dayheld peak
no measure0-0
reserve 10% x 90d from day 12,9003228,500
reserve 10% x 180d from day 12,9003242,000
reserve 25% x 180d from day 113,50090105,000
reserve 10% x 90d from day 450-25,500
reserve 25% x 180d from day 454,6009097,500
freeze 14d from day 4543,400580
freeze 30d from day 4593,000740
freeze 30d, spend cut day 4759,400740

核心公式

\[Z = \left(\frac{3\,B\,\text{SIG}^2}{4\,I}\right)^{1/3},\]
查看完整验算程序
# Chapter 6. p61: the cash a shop must hold before a surge, by scenario.
# Daily takings: 1,000 on days 1-30, 4,000 on days 31-90 (a campaign),
# 1,500 on days 91-365. Settlement two days after the sale. Costs paid
# the same day: variable 0.65 of takings during the campaign (paid
# advertising, delivery) and 0.35 otherwise, plus a fixed 500 a day.
# Scenarios (the acquirer's measures are imposed, not chosen):
#   rolling reserve r of each settlement, released T days later,
#   either from day 1 or imposed on day 45; a payout freeze of 14 or 30
#   days from day 45 (held payouts paid at the end of the freeze),
#   with the campaign spend kept, or cut to the base share on day 47.
# Buffer = the largest fall of cash below its level on the eve of the
# measure (day 0 when there is no measure): the cash that must already
# sit aside, whatever profit came before, so that no bill goes unpaid.
# All amounts are illustrative units without currency.
import math

DAYS, LAG, FIXED = 365, 2, 500.0


def takings(t):
    if t <= 30:
        return 1000.0
    if t <= 90:
        return 4000.0
    return 1500.0


def var_share(t, cut_day=None):
    if 31 <= t <= 90 and (cut_day is None or t < cut_day):
        return 0.65
    return 0.35


def path(r=0.0, hold=0, r_from=1, frz_from=None, frz_len=0, cut=None):
    cash, low, low_day = 0.0, 0.0, 0
    event = frz_from or (r_from if r > 0 else 1)
    eve = 0.0
    release = {}
    frozen = 0.0
    held_peak, held = 0.0, 0.0
    for t in range(1, DAYS + 1):
        flows = [-FIXED, -var_share(t, cut) * takings(t)]
        gross = takings(t - LAG)      # days before 1 sold 1,000 a day
        keep = r * gross if t >= r_from else 0.0
        if keep:
            release[t + hold] = release.get(t + hold, 0.0) + keep
            held = held + keep
        pay = gross - keep
        if frz_from and frz_from <= t < frz_from + frz_len:
            frozen = frozen + pay
        else:
            flows.append(pay)
        if frz_from and t == frz_from + frz_len:
            flows.append(frozen)
            frozen = 0.0
        if t in release:
            flows.append(release[t])
            held = held - release[t]
        held_peak = max(held_peak, held)
        if t == event:
            eve = cash
        cash = cash + math.fsum(flows)
        if t >= event and cash - eve < low:
            low, low_day = cash - eve, t
    return 0.0 - low, low_day, held_peak


def main():
    rows = [("no measure", {}),
            ("reserve 10% x 90d from day 1", dict(r=0.10, hold=90)),
            ("reserve 10% x 180d from day 1", dict(r=0.10, hold=180)),
            ("reserve 25% x 180d from day 1", dict(r=0.25, hold=180)),
            ("reserve 10% x 90d from day 45",
             dict(r=0.10, hold=90, r_from=45)),
            ("reserve 25% x 180d from day 45",
             dict(r=0.25, hold=180, r_from=45)),
            ("freeze 14d from day 45", dict(frz_from=45, frz_len=14)),
            ("freeze 30d from day 45", dict(frz_from=45, frz_len=30)),
            ("freeze 30d, spend cut day 47",
             dict(frz_from=45, frz_len=30, cut=47))]
    print("scenario                          buffer    worst day  held peak")
    for name, kw in rows:
        need, day, peak = path(**kw)
        print("{:<32} {:>8,.0f}    {:>5}    {:>9,.0f}".format(
            name, need, day if day else "-", peak))


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 6 章「准备金:滚动扣留、结算冻结与现金不断血的最低缓冲」(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