Question

我遇到了以下问题。我有两个带有每日拦截估计的向量，我假设拦截的行为都分别遵循OU过程。但是，拦截之间存在很强的相关性，因此为了处理两个OU进程，我想关联两个OU进程的维纳进程。问题：我不知道如何在我的代码中实现相关系数，可能有人可以帮助我：我正在运行以下代码：

 %model  intercept   
 x = inter_w;
 dx = diff(x);
 dt = 1/252; % Time in years (252 observations per year)
 dxdt = dx/dt;
 x(end) = []; % To ensure the number of elements in x and dxdt match

 % Fit a linear trend to estimate mean reversion parameters 
 coeff = polyfit(x, dxdt, 1); 
 res = dxdt - polyval(coeff, x);
 revRate   = -coeff(1)
 meanLevel = coeff(2)/revRate
 vol       = std(res) * sqrt(dt)

 OUmodel = hwv(revRate, meanLevel, vol,  'StartState', inter_w(507,1), 'Correlation', corr)

第二个OU过程

 y = inter_s; dy = diff(y); 
 dt = 1/252; % Time in years (252 observations per year) 
 dydt = dy/dt; y(end) = []; % To ensure the number of elements in x and dxdt match

 % Fit a linear trend to estimate mean reversion parameters 
 coeffy = polyfit(y, dydt, 1); 
 resy = dydt - polyval(coeffy, y);
 revRatey   = -coeffy(1)
 meanLevely = coeffy(2)/revRatey
 voly       = std(resy) * sqrt(dt)

 y = inter_s;
dy = diff(y);
dt = 1/252; % Time in years (252 observations per year)
dydt = dy/dt;
y(end) = []; % To ensure the number of elements in x and dxdt match

% Fit a linear trend to estimate mean reversion parameters
coeffy = polyfit(y, dydt, 1);
resy = dydt - polyval(coeffy, y);

revRatey   = -coeffy(1)
meanLevely = coeffy(2)/revRatey
voly       = std(resy) * sqrt(dt)

OUmodely = hwv(revRatey, meanLevely, voly, 'StartState', inter_s(507,1), 'Correlation', corr)

如果我理解这个解释，我的维纳过程现在与相关系数相关＆＃34; corr＆＃34;。但是，如果我模拟1000个inter_w和1000个inter_s，那些进程之间的相关性甚至不接近＆＃34; corr＆＃34; ..我在某个地方犯了一个错误，但不知道在哪里..谢谢你的回复

如何在维纳进程中运行具有关联的OU进程

0 个答案: