关于「实证闭环:每日转化体检、蒸馏小工具与预注册证伪」,《录屏钩子赛马与收费站导流》给出的定量结论是什么?
结论:第八章给出蒸馏小工具「钩子赛马计分、桥页漏损审计与日排程器」与预注册证伪方案,其中一项检验需要 85 条视频。全部参数为示意,计算由书中 22 段只用标准库的程序生成;本书不承诺任何收入,不教任何刷量、伪装或操纵手法。
数据口径:两种打法各跑一年的合成实验里,对口的稳定案例录屏日均付款人数是赌爆款的 16.51 倍。 这是目标,不是承诺:在示意基线下,它需要每天 280,112 次播放。
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: 85
| H1. videos needed to detect r | 0.3: 85 |
| H2. taps per arm to detect 0.70 | 0.80: 294 |
| H3. if the true ratio is 0.5, upper 95 % bound of the measured ratio | |
| payers on viral days se of log ratio upper bound | |
| 10 | 0.317 0.930 |
| 20 | 0.224 0.776 |
| 40 | 0.159 0.683 |
| 80 | 0.113 0.624 |
| 160 | 0.081 0.586 |
| Refutation lines: H1 fails if the 95 % interval of r lies below 0.1 after 85 videos; | |
| H2 fails if the no-hop arm is not above 0.75 after 294 taps per arm; | |
| H3 (the book's claim) fails if the ratio's lower 95 % bound exceeds 0.8. |
查看完整验算程序
# Chapter 8. p82: sample sizes and decision lines of the pre-registered refutation plan.
# H1 (chapters 3-4): across videos, the 3-second survival of the hook correlates with the
# profile-visit rate per view at r >= 0.3. Fisher z, two-sided 5 %, power 0.8.
# H2 (chapter 5): the no-hop report page raises survival after the tap from 0.70 to at
# least 0.80. Two-proportion test, per arm.
# H3 (preface, the central claim): payers per view on "viral" days (views >= 5 x the
# 90-day median) are below half of those on ordinary days. Ratio of two Poisson rates,
# solved for the expected payers needed on viral days.
import math
ZA, ZB = 1.959964, 0.841621
if __name__ == "__main__":
r = 0.3
n1 = ((ZA + ZB) / math.atanh(r)) ** 2 + 3
print("H1. videos needed to detect r = {:.1f}: {:.0f}".format(r, math.ceil(n1)))
p0, p1 = 0.70, 0.80
pbar = (p0 + p1) / 2
n2 = (ZA * math.sqrt(2 * pbar * (1 - pbar)) + ZB * math.sqrt(p0 * (1 - p0) + p1 * (1 - p1))) ** 2 \
/ (p1 - p0) ** 2
print("H2. taps per arm to detect 0.70 -> 0.80: {:.0f}".format(math.ceil(n2)))
# log rate ratio has variance about 1 / x_viral + 1 / x_ordinary (x = payer counts)
target = math.log(0.5)
rows = []
for xv in (10, 20, 40, 80, 160):
xo = 4000
se = math.sqrt(1.0 / xv + 1.0 / xo)
rows.append((xv, se, math.exp(target + ZA * se)))
print("H3. if the true ratio is 0.5, upper 95 % bound of the measured ratio")
print(" payers on viral days se of log ratio upper bound")
for xv, se, ub in rows:
print(" {:>20} {:>17.3f} {:>13.3f}".format(xv, se, ub))
print()
print("Refutation lines: H1 fails if the 95 % interval of r lies below 0.1 after {:.0f} videos;".format(
math.ceil(n1)))
print(" H2 fails if the no-hop arm is not above 0.75 after {:.0f} taps per arm;".format(math.ceil(n2)))
print(" H3 (the book's claim) fails if the ratio's lower 95 % bound exceeds 0.8.")唯一数据源与引用
范雅福(Fan, Y.). (2026). 《零露脸案例录屏短视频量产、前三秒痛点钩子赛马与私域收费站导流母典》第 8 章「实证闭环:每日转化体检、蒸馏小工具与预注册证伪」(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-350 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026350,
author = {Fan, Yafu},
title = {Faceless Case-Screen Short-Video Production, Three-Second Hook Tournaments and Direct-to-Paywall Traffic Funnels},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}