关于「原典考辨:《孙子兵法·作战篇》与霍普金斯的《科学的广告》」,《收费站买量精算》给出的定量结论是什么?

结论:第七章考辨《孙子兵法·作战篇》与霍普金斯 1923 年的《科学的广告》,并写出给四件套的蒸馏规格。

数据口径:按示意参数,tarotmalaysia 一个破冰买家的期望净值是破冰价的 2.68 倍,按首单回本线止损,在四十条合成素材上只拿到可得利润的 0.222。 书中参数为示意,全部计算由书中程序生成。

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

CPArulespend (T)loss (T)days at RM 300loss RM
1 TW50.010.014.40
1 TZ47.69-0.074.2-2
1 TS11.35-0.041.0-1
2 TW100.0750.078.81,327
2 TZ78.0539.126.91,037
2 TS16.198.031.4213
3 TW149.5499.5413.22,639
3 TZ96.1964.248.51,703
3 TS10.617.070.9188
5 TW250.07200.0722.15,304
5 TZ113.7590.9010.12,410
5 TS8.016.400.7170
查看完整验算程序
# Chapter 7. p71: "a thousand pieces of gold a day" - the price of a
# long campaign, priced for creatives. Rule W ("wait"): judge a creative
# only after 50 conversions (a folk reading of the platform learning
# phase). Rule S: the sequential test of p32 (alpha 0.05, beta 0.10,
# H1 = 2 T, cap 60 T). Rule Z: stop at 2.996 T with zero conversions
# (p21), otherwise wait. For each true CPA: mean spend before the call,
# the loss above break-even (spend minus conversions valued at T), and
# days at RM 300 a day, T = 26.51 RM. Monte Carlo, fixed seed.
import math
import random

T, DAILY, N_WAIT = 26.51, 300.0, 50
ALPHA, BETA = 0.05, 0.10
A_UP = math.log((1 - BETA) / ALPHA)
B_LO = math.log(BETA / (1 - ALPHA))
JUMP, DRIFT, CAP = math.log(0.5), 0.5, 60.0


def sprt(rate, rng):
    s, l, n = 0.0, 0.0, 0
    while True:
        gap = rng.expovariate(rate)
        t_hit = (A_UP - l) / DRIFT
        if t_hit <= gap or s + gap >= CAP:
            return min(s + t_hit, CAP), n
        s += gap
        n += 1
        l += DRIFT * gap + JUMP
        if l <= B_LO:
            return s, n


def wait(rate, rng, zero_stop):
    s, n = 0.0, 0
    while n < N_WAIT:
        gap = rng.expovariate(rate)
        if zero_stop and n == 0 and gap > 2.996:
            return 2.996, 0
        s += gap
        n += 1
    return s, n


if __name__ == "__main__":
    rng = random.Random(20261120)
    print("  CPA   rule  spend (T)   loss (T)  days at RM 300  loss RM")
    for k in (1.0, 2.0, 3.0, 5.0):
        for name in ("W", "Z", "S"):
            sp, ls = [], []
            for _ in range(4000):
                if name == "S":
                    s, n = sprt(1.0 / k, rng)
                else:
                    s, n = wait(1.0 / k, rng, name == "Z")
                sp.append(s)
                ls.append(s - n)
            ms, ml = math.fsum(sp) / 4000, math.fsum(ls) / 4000
            print("  {:>3.0f} T   {}  {:>9.2f}  {:>9.2f}  {:>14.1f}"
                  "  {:>7,.0f}".format(k, name, ms, ml, ms * T / DAILY,
                                       ml * T))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《测评与命理收费站买量精算、服务端转化回传校准与日入百单扩量母典》第 7 章「原典考辨:《孙子兵法·作战篇》与霍普金斯的《科学的广告》」(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-345 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026345,
  author    = {Fan, Yafu},
  title     = {Paid-Ad Scaling for Diagnostic Paywalls: Meta and TikTok Stopping Rules, Server-Side CAPI Calibration and Hundred-Order Daily Unit Economics},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论