关于「从总量到尺码:分尺码报童、碎片溢价与断码」,《季前订货临界分位母典》给出的定量结论是什么?

结论:第四章说明高毛利款分尺码订量之和比整款订量多出 17.9 件。

数据口径:书中的商品、门店与需求参数均为示意;全部计算由书中程序生成;本书是非个人化经营建议。 第一章用边际算术推出临界分位,示意参数下只有毛利率恰为 0.3750 时最优订量才等于平均数,毛利八成的商品按平均数订会少赚 6.96 %;据施韦策与卡雄实验的已发表数据重建,受试者把订量向平均数拉了 0.644(高利润)与 0.787(低利润)的距离。

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: 17.9

luxury CF 0.8333: aggregate q* = 258.0
sizemeansdq_kq_k(A=inf)
XS13.37.620.718.5
S55.620.375.272.8
M62.222.283.781.4
L48.918.466.764.2
XL20.09.829.527.1
sum of size orders 275.9 (+17.9); A=inf 264.0 (+6.0)
basic CF 0.4286: aggregate q* = 189.2
sizemeansdq_kq_k(A=inf)
XS13.37.612.012.4
S55.620.351.952.4
M62.222.258.258.7
L48.918.445.646.0
XL20.09.818.218.7
sum of size orders 185.9 (-3.3); A=inf 188.1 (-1.1)

核心公式

\[\mathrm{Var}(D_k) = \mu\,(p_k - E\theta_k^2) + E[D^2]\,E\theta_k^2 - \mu^2 p_k^2,\qquad E\theta_k^2 = \frac{p_k(1-p_k)}{A+1} + p_k^2.\]
查看完整验算程序
# Chapter 4. p41: Per-size newsvendor orders do not add up to the total.
# Total demand D ~ Normal(200, 60). Size share theta ~ Dirichlet with
# mean p (posterior of p31) and total concentration A = 90. Given D
# and theta, size demand D_k ~ Binomial(D, theta_k). Moments:
#   E D_k = mu p_k
#   Var D_k = mu (p_k - E th^2) + E[D^2] E th^2 - mu^2 p_k^2,
#   E th^2 = p_k (1 - p_k) / (A + 1) + p_k^2.
# Normal approximation per size, q_k = E D_k + z(CF) sd_k. Compared
# with the aggregate order mu + z sigma. Parameters illustrative.
import math
from statistics import NormalDist

Z = NormalDist()
MU, SD, A = 200.0, 60.0, 90.0
P = (0.0667, 0.2778, 0.3111, 0.2444, 0.1000)
SIZES = ("XS", "S", "M", "L", "XL")
ED2 = SD * SD + MU * MU


def size_sd(p, a):
    eth2 = p * (1.0 - p) / (a + 1.0) + p * p
    var = math.fsum([MU * (p - eth2), ED2 * eth2, -MU * MU * p * p])
    return math.sqrt(var)


for name, cf in (("luxury", 75.0 / 90.0), ("basic", 30.0 / 70.0)):
    z = Z.inv_cdf(cf)
    agg = MU + z * SD
    qk = [MU * p + z * size_sd(p, A) for p in P]
    qinf = [MU * p + z * size_sd(p, 1e12) for p in P]
    print("{} CF {:.4f}: aggregate q* = {:.1f}".format(name, cf, agg))
    print("  size  mean    sd     q_k    q_k(A=inf)")
    for k, s in enumerate(SIZES):
        print("  {:4}  {:5.1f}  {:5.1f}  {:5.1f}  {:5.1f}".format(
            s, MU * P[k], size_sd(P[k], A), qk[k], qinf[k]))
    tot, tinf = math.fsum(qk), math.fsum(qinf)
    print("  sum of size orders {:.1f} ({:+.1f}); A=inf {:.1f} ({:+.1f})"
          .format(tot, tot - agg, tinf, tinf - agg))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《买手季前订货报童临界分位、尺码曲线狄利克雷先验与季末降价动态规划母典》第 4 章「从总量到尺码:分尺码报童、碎片溢价与断码」(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-211 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026211,
  author    = {Fan, Yafu},
  title     = {Newsvendor Critical Fractiles for Pre-Season Buying, Dirichlet Size-Curve Priors and Dynamic-Programming End-of-Season Markdowns},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论