关于「磨损曲线,给授权定价」,《暗投授权与钩子归因》给出的定量结论是什么?
结论:第二章按磨损曲线给授权定价:一季只值首月 1.70 倍,按月价 2.5 倍报季度价高估 1.47 倍;磨损阶梯合约让品牌期望剩余升到 30,619。
数据口径:书中参数均为示意,全部计算由书中程序生成;本书是非个人化经营分析,不构成经营或法律建议。 第六章用二叉树为跳单期权定价,合理预估 20.82 已拦住示意模型中全部跳单。
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: 1.47, 1.70, 2.5
| check k = 1, 30 days: Simpson 37989.61, closed form 37989.61 | |||||
| k | half-life (M imps) | 30 days | 90 days | 360 days | 90/30 |
| 0.7 | 4.74 | 34,195 | 64,248 | 89,091 | 1.88 |
| 1.0 | 5.55 | 37,990 | 64,411 | 71,991 | 1.70 |
| 1.5 | 6.27 | 42,421 | 63,984 | 64,998 | 1.51 |
| no wear-out, 90/30 would be 3.00; a quarterly quote at 2.5x | |||||
| the fair monthly fee overpays by a factor 1.47 at k = 1 |
查看完整验算程序
# Chapter 2. p21: what a whitelisting licence is worth under wear-out.
# Model: incremental margin per 1,000 impressions starts at v0 and
# decays with cumulative impressions x as v0 * exp(-(x / lam) ** k);
# k = 1 is exponential, k < 1 decays fast early, k > 1 holds then
# drops. Licence value over I impressions is the integral of that
# rate, by Simpson's rule with math.fsum, checked against the closed
# form for k = 1. Delivery 200,000 impressions a day, v0 = 9 per
# 1,000 and lam = 8,000,000 impressions are illustrative. A month-3
# to month-1 value ratio of 0.8 or more would mean little wear-out.
import math
V0, LAM, DAY = 9.0, 8.0e6, 2.0e5
def rate(x, k):
return V0 / 1000.0 * math.exp(-(x / LAM) ** k)
def value(imps, k, steps=2000):
h = imps / steps
terms = []
for j in range(steps + 1):
w = 1 if j in (0, steps) else (4 if j % 2 else 2)
terms.append(w * rate(j * h, k))
return h / 3.0 * math.fsum(terms)
def closed_k1(imps):
return V0 / 1000.0 * LAM * (1.0 - math.exp(-imps / LAM))
if __name__ == "__main__":
i30 = 30 * DAY
print("check k = 1, 30 days: Simpson {:.2f}, closed form {:.2f}"
.format(value(i30, 1.0), closed_k1(i30)))
print(" k half-life (M imps) 30 days 90 days 360 days"
" 90/30")
for k in (0.7, 1.0, 1.5):
half = LAM * math.log(2.0) ** (1.0 / k) / 1e6
v30, v90 = value(i30, k), value(90 * DAY, k)
v360 = value(360 * DAY, k)
print(" {:>3.1f} {:>18.2f} {:>7,.0f} {:>7,.0f} {:>8,.0f}"
" {:>5.2f}".format(k, half, v30, v90, v360, v90 / v30))
v30, v90 = closed_k1(i30), closed_k1(90 * DAY)
print("no wear-out, 90/30 would be 3.00; a quarterly quote at 2.5x")
print("the fair monthly fee overpays by a factor {:.2f} at k = 1"
.format(2.5 / (v90 / v30)))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《暗投白名单授权经济学、三秒钩子生存分析与单层星探归因博弈母典》第 2 章「磨损曲线,给授权定价」(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-333 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026333,
author = {Fan, Yafu},
title = {Dark-Ads Whitelist Licensing Economics, Three-Second Hook Survival Analysis and Single-Tier Scout Attribution Games},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}