[Add]新增CRR期权定价模型,针对国内美式商品期权

This commit is contained in:
vn.py 2018-01-07 12:34:46 +08:00
parent ebc748259e
commit c320ce844f

181
vnpy/pricing/crr.py Normal file
View File

@ -0,0 +1,181 @@
# encoding: UTF-8
'''
Cox-Ross-Rubinstein二叉树期权定价模型主要用于标的物为期货的美式期权的定价
变量说明
f标的物期货价格
k行权价
r无风险利率
t剩余到期时间
v隐含波动率
cp期权类型+1/-1对应call/put
n: 二叉树高度
price期权价格
出于开发演示的目的本文件中的希腊值计算基于简单数值差分法
运算效率一般实盘中建议使用更高速的算法
本文件中的希腊值计算结果没有采用传统的模型价格数值而是采用
了实盘交易中更为实用的百分比变动数值具体定义如下
delta当f变动1%price的变动
gamma当f变动1%delta的变动
theta当t变动1天时price的变动国内交易日每年240天
vega当v涨跌1个点时price的变动如从16%涨到17%
'''
from __future__ import division
import numpy as np
from math import (isnan, exp, sqrt, pow)
# 计算希腊值和隐含波动率时用的参数
STEP_CHANGE = 0.001
STEP_UP = 1 + STEP_CHANGE
STEP_DOWN = 1 - STEP_CHANGE
STEP_DIFF = STEP_CHANGE * 2
DX_TARGET = 0.00001
#----------------------------------------------------------------------
def generateTree(f, k, r, t, v, cp, n):
"""生成二叉树"""
dt = t / n
u = exp(v * sqrt(dt))
d = 1 / u
a = exp(r * dt)
uTree = np.zeros((n+1,n+1))
oTree = np.zeros((n+1,n+1))
# 计算风险平价概率
p = (a - d) / (u - d)
p1 = p / a
p2 = (1 - p) / a
# 计算标的树
uTree[0, 0] = f
for i in range(1, n+1):
uTree[0, i] = uTree[0, i-1] * u
for j in range(1, i+1):
uTree[j, i] = uTree[j-1, i-1] * d
# 计算期权树
for j in range(n+1):
oTree[j, n] = max(0, cp * (uTree[j, n]-k))
for i in range(n-1,-1,-1):
for j in range(i+1):
oTree[j, i] = max((p1 * oTree[j, i+1] + p2 * oTree[j+1, i+1]), # 美式期权存续价值
cp * (uTree[j, i] - k)) # 美式期权行权价值
# 返回期权树和标的物树结果
return oTree, uTree
#----------------------------------------------------------------------
def calculatePrice(f, k, r, t, v, cp, n=15):
"""计算期权价格"""
oTree, uTree = calculatePrice(f, k, r, t, v, cp)
return oTree[0, 0]
#----------------------------------------------------------------------
def calculateDelta(f, k, r, t, v, cp, n=15):
"""计算Delta值"""
price1 = calculatePrice(f*STEP_UP, k, r, t, v, cp, n)
price2 = calculatePrice(f*STEP_DOWN, k, r, t, v, cp, n)
delta = (price1 - price2) / (f * STEP_DIFF) * (f * 0.01)
return delta
#----------------------------------------------------------------------
def calculateGamma(f, k, r, t, v, cp, n=15):
"""计算Gamma值"""
delta1 = calculateDelta(f*STEP_UP, k, r, t, v, cp, n)
delta2 = calculateDelta(f*STEP_DOWN, k, r, t, v, cp, n)
gamma = (delta1 - delta2) / (f * STEP_DIFF) * pow(f, 2) * 0.0001
return gamma
#----------------------------------------------------------------------
def calculateTheta(f, k, r, t, v, cp, n=15):
"""计算Theta值"""
price1 = calculatePrice(f, k, r, t*STEP_UP, v, cp, n)
price2 = calculatePrice(f, k, r, t*STEP_DOWN, v, cp, n)
theta = -(price1 - price2) / (t * STEP_DIFF * 240)
return theta
#----------------------------------------------------------------------
def calculateVega(f, k, r, t, v, cp, n=15):
"""计算Vega值"""
vega = calculateOriginalVega(f, k, r, t, v, cp, n) / 100
return vega
#----------------------------------------------------------------------
def calculateOriginalVega(f, k, r, t, v, cp, n=15):
"""计算原始vega值"""
price1 = calculatePrice(f, k, r, t, v*STEP_UP, cp, n)
price2 = calculatePrice(f, k, r, t, v*STEP_DOWN, cp, n)
vega = (price1 - price2) / (v * STEP_DIFF)
return vega
#----------------------------------------------------------------------
def calculateGreeks(f, k, r, t, v, cp, n=15):
"""计算期权的价格和希腊值"""
price = calculatePrice(f, k, r, t, v, cp, n)
delta = calculateDelta(f, k, r, t, v, cp, n)
gamma = calculateGamma(f, k, r, t, v, cp, n)
theta = calculateTheta(f, k, r, t, v, cp, n)
vega = calculateVega(f, k, r, t, v, cp, n)
return price, delta, gamma, theta, vega
#----------------------------------------------------------------------
def calculateImpv(price, f, k, r, t, cp, n=15):
"""计算隐含波动率"""
# 检查期权价格必须为正数
if price <= 0:
return 0
# 检查期权价格是否满足最小价值(即到期行权价值)
meet = False
if cp == 1 and (price > (f - k) * exp(-r * t)):
meet = True
elif cp == -1 and (price > k * exp(-r * t) - f):
meet = True
# 若不满足最小价值则直接返回0
if not meet:
return 0
# 采用Newton Raphson方法计算隐含波动率
v = 0.3 # 初始波动率猜测
for i in range(50):
# 计算当前猜测波动率对应的期权价格和vega值
p = calculatePrice(f, k, r, t, v, cp, n)
vega = calculateOriginalVega(f, k, r, t, v, cp, n)
# 如果vega过小接近0则直接返回
if not vega:
break
# 计算误差
dx = (price - p) / vega
# 检查误差是否满足要求,若满足则跳出循环
if abs(dx) < DX_TARGET:
break
# 计算新一轮猜测的波动率
v += dx
# 检查波动率计算结果非负
if v <= 0:
return 0
# 保留4位小数
v = round(v, 4)
return v