关于「希腊值随剩余期限的非线性变化」,《隐含波动率、偏度与垂直价差》给出的定量结论是什么?
结论:希腊值由布莱克—斯科尔斯方程联系在一起,在模型中时间损耗是伽马的价格。平值期权的伽马与时间损耗随到期临近按 \(1/\sqrt{T}\) 加速,实值与虚值期权的则先升后降。熊市看跌价差的净希腊值随标的位置改变符号:标的在多头行权价附近时净损耗为负并迅速放大,在空头行权价附近时为正。时间止损的合理性因此取决于状态,而不只是天数。
数据口径:一个熊市看跌垂直价差最多亏多少、最多赚多少,交易者在开仓时就知道;可「持仓 7—10 个交易日没有起色就平仓」这条经验规则,能被证明吗?本书是教学与研究用的数学模型,不是投资建议。 书中以标注框区分已发表的证据、可运行的 Python 计算(16 段程序)、原典与作者的模型和假说;书中所有价格与交易都是虚构或示意的。
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
| finite-difference check: largest relative error 3.3e-06 (gamma limited by step size) | |
| put K | 100, vol 0.22 DTE: 60 30 14 7 3 1 |
| ITM S | 95 delta -0.683 -0.771 -0.873 -0.950 -0.995 -1.000 |
| ITM S | 95 gamma 0.042 0.051 0.051 0.035 0.008 0.000 |
| ITM S | 95 theta/day -0.019 -0.024 -0.023 -0.013 0.003 0.008 |
| ITM S | 95 vega/pt 0.137 0.082 0.039 0.013 0.001 0.000 |
| ATM S | 100 delta -0.460 -0.472 -0.481 -0.486 -0.491 -0.495 |
| ATM S | 100 gamma 0.045 0.063 0.092 0.131 0.200 0.346 |
| ATM S | 100 theta/day -0.025 -0.038 -0.057 -0.083 -0.128 -0.226 |
| ATM S | 100 vega/pt 0.161 0.114 0.078 0.055 0.036 0.021 |
| OTM S | 105 delta -0.259 -0.199 -0.119 -0.051 -0.007 -0.000 |
| OTM S | 105 gamma 0.035 0.042 0.044 0.033 0.009 0.000 |
| OTM S | 105 theta/day -0.023 -0.029 -0.031 -0.023 -0.007 -0.000 |
| OTM S | 105 vega/pt 0.138 0.084 0.041 0.015 0.002 0.000 |
| ATM theta, 7 vs 30 DTE: ratio 2.19 (sqrt(30/7) | 2.07) |
| bear put spread, long 100 (vol .22) / short 95 (vol .245) | |
| DTE S value delta gamma theta/day vega/pt | |
| 30 | 92.0 3.811 -0.244 -0.0259 +0.0244 -0.050 |
| 30 | 95.0 2.983 -0.299 -0.0091 +0.0166 -0.026 |
| 30 | 97.5 2.225 -0.300 +0.0085 +0.0052 +0.004 |
| 30 | 100.0 1.518 -0.260 +0.0219 -0.0057 +0.031 |
| 30 | 103.0 0.849 -0.184 +0.0269 -0.0129 +0.046 |
| 7 | 92.0 4.698 -0.177 -0.0806 +0.0584 -0.032 |
| 7 | 95.0 3.743 -0.464 -0.0883 +0.0746 -0.039 |
| 7 | 97.5 2.392 -0.575 +0.0102 +0.0117 +0.000 |
| 7 | 100.0 1.096 -0.425 +0.0953 -0.0539 +0.038 |
| 7 | 103.0 0.254 -0.150 +0.0706 -0.0473 +0.031 |
| 3 | 92.0 4.928 -0.077 -0.0709 +0.0504 -0.012 |
核心公式
\[\Delta = N(d_1) - 1, \qquad \Gamma = \frac{\varphi(d_1)}{S\sigma\sqrt{T}}, \qquad \mathcal{V} = S\sqrt{T}\,\varphi(d_1),\]
查看完整验算程序
# Greeks of a European put (closed form, checked by finite differences) as days to
# expiry (DTE) shrink, and the net Greeks of the 100/95 bear put spread. Theta per
# calendar day, vega per 1 vol point. Illustrative parameters.
import math
def ncdf(x):
return 0.5 * (1.0 + math.erf(x / math.sqrt(2.0)))
def npdf(x):
return math.exp(-0.5 * x * x) / math.sqrt(2 * math.pi)
def put(S, K, T, r, s):
d1 = (math.log(S / K) + (r + 0.5 * s * s) * T) / (s * math.sqrt(T))
d2 = d1 - s * math.sqrt(T)
price = K * math.exp(-r * T) * ncdf(-d2) - S * ncdf(-d1)
delta = ncdf(d1) - 1
gamma = npdf(d1) / (S * s * math.sqrt(T))
theta = -S * npdf(d1) * s / (2 * math.sqrt(T)) + r * K * math.exp(-r * T) * ncdf(-d2)
vega = S * npdf(d1) * math.sqrt(T)
return price, delta, gamma, theta / 365, vega / 100
def fd_check(S=100.0, K=100.0, T=30 / 365, r=0.03, s=0.22, h=1e-4):
p, d, g, th, v = put(S, K, T, r, s)
fd = ((put(S + h, K, T, r, s)[0] - put(S - h, K, T, r, s)[0]) / (2 * h),
(put(S + h, K, T, r, s)[0] - 2 * p + put(S - h, K, T, r, s)[0]) / h ** 2,
-(put(S, K, T + h, r, s)[0] - put(S, K, T - h, r, s)[0]) / (2 * h) / 365,
(put(S, K, T, r, s + h)[0] - put(S, K, T, r, s - h)[0]) / (2 * h) / 100)
return max(abs(a - b) / abs(b) for a, b in zip(fd, (d, g, th, v)))
if __name__ == "__main__":
r = 0.03
print(f"finite-difference check: largest relative error {fd_check():.1e}"
" (gamma limited by step size)")
print("put K = 100, vol 0.22 DTE: 60 30 14 7 3 1")
for S, lab in ((95, "ITM S=95"), (100, "ATM S=100"), (105, "OTM S=105")):
g = [put(S, 100, d / 365, r, 0.22) for d in (60, 30, 14, 7, 3, 1)]
for i, name in ((1, "delta"), (2, "gamma"), (3, "theta/day"), (4, "vega/pt")):
print(f" {lab:10s} {name:10s}" + "".join(f"{x[i]:8.3f}" for x in g))
th = [put(100, 100, d / 365, r, 0.22)[3] for d in (30, 7)]
print(f"ATM theta, 7 vs 30 DTE: ratio {th[1] / th[0]:.2f}"
f" (sqrt(30/7) = {math.sqrt(30 / 7):.2f})")
print("bear put spread, long 100 (vol .22) / short 95 (vol .245)")
print(" DTE S value delta gamma theta/day vega/pt")
for d in (30, 7, 3):
for S in (92.0, 95.0, 97.5, 100.0, 103.0):
a, b = put(S, 100, d / 365, r, 0.22), put(S, 95, d / 365, r, 0.245)
n = [x - y for x, y in zip(a, b)]
print(f" {d:3d} {S:5.1f} {n[0]:6.3f} {n[1]:+6.3f} {n[2]:+7.4f}"
f" {n[3]:+7.4f} {n[4]:+6.3f}")唯一数据源与引用
范雅福(Fan, Y.). (2026). 《期权隐含波动率曲面、偏度溢价与垂直价差结构的定量风控》第 4 章「希腊值随剩余期限的非线性变化」(版本 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23796765
DOI: 10.5281/zenodo.23796765 · 阅读本章所在专著 · 5DL-B-038 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026038,
author = {Fan, Yafu},
title = {Implied Volatility Surfaces, Skew Premia and Quantitative Risk Control of Vertical Spreads},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
doi = {10.5281/zenodo.23796765},
url = {https://doi.org/10.5281/zenodo.23796765}
}