关于「稀缺溢价半衰期:指数衰减拟合、重复交易指数与配货模型」,《擒纵误差与名表溢价半衰期》给出的定量结论是什么?

结论:第六章说明拟合溢价半衰期必须先估长期水平,否则会把 5.69 个月误估成 8.61 个月,并用重复交易指数与供求模型区分配货放松决定的溢价终点与市场结构决定的塌缩速度。

数据口径:书中参数为示意;全部计算由书中程序生成;本书是非个人化的经营与理财分析,不构成个人化法律或理财建议,涉及法规之处以上架前人工复核为准。 第一章并排摆出两本账:天文台机械表的容差比月差 15 秒的石英表宽 12.0 倍,比天文台石英标准宽 85.7 倍;拍卖样本里的收藏表有正收益(已发表研究报告实际年化 5.5 %),而示意模型中全部增值的 0.816 集中在一成型号上。

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: 5.69, 8.61

true P0 0.950 Pinf 0.050 lam 0.120 half-life 5.78 mo
fitted P0 0.994 Pinf 0.0500 lam 0.122 half-life 5.69 mo
bootstrap 95% interval for half-life5.31 to 6.12 mo
naive fit with Pinf0: half-life 8.61 mo

核心公式

\[P(t) = P_\infty + (P_0 - P_\infty)\,e^{-\lambda t}\]
查看完整验算程序
# Chapter 6. p61: fitting an exponential premium decay and its half-life
# 模型:月度溢价序列 P(t) = Pinf + (P0 - Pinf) exp(-lam t) 乘以对数正态
# 噪声(标准差 0.08),真值 P0 = 0.95、Pinf = 0.05、lam = 0.12/月,
# 36 个月,全部为示意合成数据。拟合:对 Pinf 做网格搜索,每个 Pinf 下
# 对 ln(P - Pinf) 做最小二乘直线,取残差平方和最小者;
# 置信区间用对数残差自助法 400 次(固定种子)。
import math
import random
import statistics

rng = random.Random(20261112)
TRUE = (0.95, 0.05, 0.12)
T = list(range(36))


def curve(t, p0, pinf, lam):
    return pinf + (p0 - pinf) * math.exp(-lam * t)


obs = [curve(t, *TRUE) * math.exp(rng.gauss(0.0, 0.08)) for t in T]


def line(xs, ys):
    mx, my = statistics.fmean(xs), statistics.fmean(ys)
    sxy = math.fsum((x - mx) * (y - my) for x, y in zip(xs, ys))
    sxx = math.fsum((x - mx) ** 2 for x in xs)
    b = sxy / sxx
    return my - b * mx, b


def fit(ys):
    best = None
    for i in range(0, 41):
        pinf = i * 0.0025
        if min(ys) <= pinf:
            break
        a, b = line(T, [math.log(y - pinf) for y in ys])
        p0 = pinf + math.exp(a)
        sse = math.fsum((y - curve(t, p0, pinf, -b)) ** 2
                        for t, y in zip(T, ys))
        if best is None or sse < best[0]:
            best = (sse, p0, pinf, -b)
    return best[1:]


p0, pinf, lam = fit(obs)
res = [math.log(y / curve(t, p0, pinf, lam)) for t, y in zip(T, obs)]
boot = []
for _ in range(400):
    ys = [curve(t, p0, pinf, lam) * math.exp(rng.choice(res)) for t in T]
    boot.append(math.log(2) / fit(ys)[2])
boot.sort()
print("true   P0 {:.3f}  Pinf {:.3f}  lam {:.3f}  half-life {:.2f} mo".format(
    TRUE[0], TRUE[1], TRUE[2], math.log(2) / TRUE[2]))
print("fitted P0 {:.3f}  Pinf {:.4f}  lam {:.3f}  half-life {:.2f} mo".format(
    p0, pinf, lam, math.log(2) / lam))
print("bootstrap 95% interval for half-life: {:.2f} to {:.2f} mo".format(
    boot[9], boot[389]))
naive_a, naive_b = line(T, [math.log(y) for y in obs])
print("naive fit with Pinf = 0: half-life {:.2f} mo".format(
    math.log(2) / -naive_b))

唯一数据源与引用

范雅福(Fan, Y.). (2026). 《机械擒纵等时性误差、游丝温度系数与名表二级市场稀缺溢价半衰期母典》第 6 章「稀缺溢价半衰期:指数衰减拟合、重复交易指数与配货模型」(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-207 · v1.0.0 · CC-BY-NC-ND-4.0

BibTeX
@book{fan2026207,
  author    = {Fan, Yafu},
  title     = {Escapement Isochronism Errors, Hairspring Thermal Coefficients and the Half-Life of Scarcity Premiums in the Secondary Watch Market},
  year      = {2026},
  publisher = {5D Library},
  edition   = {1.0.0},
  note      = {DOI pending}
}

本书其他章节结论