关于「实证闭环:订货量计算器、季后偏差复盘与预注册检验方案」,《季前订货临界分位母典》给出的定量结论是什么?
结论:第八章给出订货量计算器、季后偏差复盘与预注册检验方案,若拉动强度与实验室相当,33 个款式季即可检出偏差。书中以标注框区分已发表的证据、可运行的 Python 计算(21 段程序)、原典与作者的假说。
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。
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: 33
| pull a sell-out p1 SKU-seasons needed (p0 | 0.1667) |
| 0.200 | 0.2195 418 |
| 0.400 | 0.2808 95 |
| 0.644 | 0.3653 33 |
| 0.800 | 0.4233 20 |
| 1.000 | 0.5000 12 |
| paired profit test (p22 luxury): diff 817.9, sd 2099.9 | |
| effect at 1.00 of model gap | 52 SKU-seasons |
| effect at 0.50 of model gap | 207 SKU-seasons |
| effect at 0.25 of model gap | 828 SKU-seasons |
查看完整验算程序
# Chapter 8. p83: Sample sizes for the pre-registered test.
# H1 (book): buyers who order near the mean sell out of high-CF SKUs
# more often than 1 - CF. Test 1: one-sample binomial test of the
# sell-out rate of CF = 0.8333 SKUs against p0 = 1 - CF, when the
# true pull is a (sell-out p1 = 1 - Phi((1 - a) z*)). Two-sided 5 %,
# power 80 %: n = [z_a sqrt(p0 q0) + z_b sqrt(p1 q1)]^2 / (p1 - p0)^2.
# Test 2: paired SKU-season profit gain of the fractile rule over the
# mean rule, using the mean and s.d. of the difference from p22.
import math
from statistics import NormalDist
Z = NormalDist()
ZA, ZB = Z.inv_cdf(0.975), Z.inv_cdf(0.80)
CF = 75.0 / 90.0
ZS = Z.inv_cdf(CF)
P0 = 1.0 - CF
print("pull a sell-out p1 SKU-seasons needed (p0 = {:.4f})".format(P0))
for a in (0.2, 0.4, 0.644, 0.8, 1.0):
p1 = 1.0 - Z.cdf((1.0 - a) * ZS)
num = ZA * math.sqrt(P0 * (1.0 - P0)) + ZB * math.sqrt(p1 * (1.0 - p1))
n = (num / (p1 - P0)) ** 2
print("{:6.3f} {:11.4f} {:8d}".format(a, p1, math.ceil(n)))
print("paired profit test (p22 luxury): diff 817.9, sd 2099.9")
for share in (1.0, 0.5, 0.25):
md = 817.9 * share
n = ((ZA + ZB) * 2099.9 / md) ** 2
print(" effect at {:.2f} of model gap: {:5d} SKU-seasons"
.format(share, math.ceil(n)))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 8 章「实证闭环:订货量计算器、季后偏差复盘与预注册检验方案」(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}
}