优化交易策略的滚动平均窗口（python）

我有一个时间序列叫做“价差”，我试图最大化交易策略的夏普比率，这取决于：

滚动平均数

滚动标准偏差的次数（在滚动平均值上创建上下秩）

问题是我无法优化，因为： “ValueError:窗口必须是整数”

如何为最小化函数指定x1（滚动周期）中的整数值

战略职能是：

def statarb(x):
    x1 = x[0]
    x2 = x[1]
    df['RollingMean'] = df['Spread'].rolling(x1).mean()
    df['Desv'] = df['Spread'].rolling(x1).std()
    df['superior'] = df['RollingMean'] + (df['Desv'] * x2)
    df['inferior'] = df['RollingMean'] - (df['Desv'] * x2)
    df['signal'] = np.where(df['Spread']>df['superior'],-1,np.nan)
    df['signal'] = np.where(df['Spread']<df['inferior'],1,df['signal']) 
    #Returns
    df['ret'] = (df['Spread']/df['Spread'].shift()-1)
    df['BnH'] = df['ret'].cumsum()
    #Position 
    df['positions'] = df['signal'].fillna(method='ffill')
    #Strategy return
    df['Stratr'] = df['ret']*df['positions'].shift()
    df['Strat_ret'] = df['Stratr'].cumsum()  
    #Sharpe Ratio
    prom = df['Stratr'].mean()*252
    desv = df['Stratr'].std()*np.sqrt(252)
    sharpe = prom/desv
    return -sharpe

致意

import numpy as np
from scipy.optimize import minimize

x0 = [50, 1]
print(statarb(x0))

n = (10, 500)
m = (1, 3)
bnds = (n, m)

sol = minimize(statarb, x0, bounds=bnds)
print(sol)