关于「尾指数与样本均值的失灵:克劳塞特拟合、厚尾组合与头部决定论」,《达人经纪的幂律、抽查与买断》给出的定量结论是什么?

结论:第二章用克劳塞特、沙利齐与纽曼的方法拟合尾指数,并说明期望收益为正的孵化组合在只签 10 人时仍有 0.637 的概率亏损。

数据口径:书中参数为示意;全部计算由书中程序生成;本书不是个人化的法律或理财建议。 第六章把阶梯分成、买断金与抽查率合成三联定价试算器,买断金取剩余佣金的 1.712 倍。

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: 0.637, 10

signings needed for at least one head (rate 0.0140):
probability 0.5050 signings
probability 0.90164 signings
probability 0.99327 signings
Pareto minimum 1.731; expected return +50%
signings P(loss) median return top-1% income share
100.637 -0.302 0.626
300.570 -0.104 0.385
1000.471 0.033 0.234
3000.314 0.158 0.322
10000.117 0.230 0.380
查看完整验算程序
# Chapter 2. p23: an incubation portfolio is decided by very few heads.
# Each signing costs 1 unit of incubation spend.  With probability 0.8
# a creator never earns the agency anything; otherwise the commission
# over the contract is Pareto with tail index 1.3 and a minimum chosen
# so that the mean per signing is 1.5, an expected return of +50%.
# For portfolios of n signings (fixed seed) the program prints the
# chance of losing money, the median return and the median share of
# income from the top 1% (at least one creator).  Head rate 0.0140 is
# the reference row of p11.  All inputs are illustrative.
import math
import random

rng = random.Random(20261112)
P_HIT, ALPHA, MEAN, H = 0.2, 1.3, 1.5, 0.0140
XMIN = MEAN / (P_HIT * ALPHA / (ALPHA - 1))


def creator():
    if rng.random() >= P_HIT:
        return 0.0
    return XMIN * (1.0 - rng.random()) ** (-1 / ALPHA)


print("signings needed for at least one head (rate {:.4f}):".format(H))
for p in (0.5, 0.9, 0.99):
    n = math.ceil(math.log(1 - p) / math.log(1 - H))
    print("  probability {:.2f}: {} signings".format(p, n))
print("Pareto minimum {:.3f}; expected return +{:.0f}%".format(
    XMIN, 100 * (MEAN - 1)))
print("signings  P(loss)  median return  top-1% income share")
for n in (10, 30, 100, 300, 1000):
    rets, shares = [], []
    top = max(1, n // 100)
    for _ in range(3000 if n < 1000 else 600):
        xs = sorted((creator() for _ in range(n)), reverse=True)
        tot = math.fsum(xs)
        rets.append(tot / n - 1.0)
        shares.append(math.fsum(xs[:top]) / tot if tot > 0 else 0.0)
    rets.sort()
    shares.sort()
    loss = len([r for r in rets if r < 0]) / len(rets)
    print("{:8d} {:8.3f} {:14.3f} {:19.3f}".format(
        n, loss, rets[len(rets) // 2], shares[len(shares) // 2]))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《网红经纪机构孵化幸存者幂律、达人飞单抽样审计博弈与违约买断期权定价》第 2 章「尾指数与样本均值的失灵:克劳塞特拟合、厚尾组合与头部决定论」(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-172 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026172,
  author    = {Fan, Yafu},
  title     = {Survivor Power Laws in Influencer-Agency Incubation, Sampling-Audit Games against Off-Platform Deals and Option Pricing of Contract Buyouts},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论