EVT 金融时间序列实现的极值理论图 R 语言图

$$
\begin{aligned}
\gamma_{t} &= \sum_{j=1}^{\lfloor s/2 \rfloor}\gamma_{j,t}, \\
\gamma_{j,t+1} &= \gamma_{j,t} \cos\lambda_j + \gamma^{\ast}_{j,t} \sin\lambda_j + \omega_{j,t}, \\
\gamma^{\ast}_{j,t+1} &= - \gamma_{j,t} \sin\lambda_j + \gamma^{\ast}_{j,t} \cos\lambda_j + \omega^{\ast}_{j,t}, \quad j=1,\ldots, \lfloor s/2 \rfloor,
\end{aligned}
$$

$$\begin{aligned}
c_{t+1} &= c_t\cos\lambda_c + c_t^\ast\sin\lambda_c + \omega_t,\\
c^\ast_{t+1} &= - c_t\sin\lambda_c + c_t^\ast\cos\lambda_c + \omega^\ast_t,
\end{aligned}
$$

where $\mu_t$ is the trend component, $\gamma_t$ is the seasonal component and $c_t$ is the cycle component. The seasonal component with period $s$ can be defined in a dummy variable form

$$
\begin{aligned}\label{structural}
y_t       &= \mu_t + \gamma_t + c_t + \epsilon_t, \quad \epsilon_t \sim N(0, H_t),\\
\mu_{t+1} &= \mu_t + \nu_t + \xi_t, \quad \xi_t \sim N(0, Q_{\textrm{level},t}), \\
\nu_{t+1} &= \nu_t + \zeta_t, \quad \zeta_t \sim N(0, Q_{\textrm{slope},t}),
\end{aligned}
$$

For computing the marginally standardized residuals, multivariate versions of $F_t$ and $v_t$ are needed, whereas the Cholesky standardized residuals can be computed directly from the sequential Kalman filter as
$$
v_{i,t} F_{i_t}^{-\frac{1}{2}}, \quad j=1,\ldots,p,\quad t=d+1\ldots,n.
$$

$$
\begin{aligned}
\VAR(y_t|y_{t-1},\ldots,y_1)&= \VAR(\E(y_t|\theta_t,y_{t-1},\ldots,y_1)|y_{t-1},\ldots,y_1)\\
&+ \E(\VAR(y_t|\theta_t,y_{t-1},\ldots,y_1)|y_{t-1},\ldots,y_1)\\
&= \VAR(\E(y_t|\theta_t)| y_{t-1},\ldots,y_1) + \E(\VAR(y_t|\theta_t)|y_{t-1},\ldots,y_1) .
\end{aligned}
$$

有R 代码嘛，求

gonglaijing9 发表于 2021-3-29 16:04
有R 代码嘛，求

有啊， 没有怎么绘制出图？

tulipsliu 发表于 2021-3-30 10:03
有啊， 没有怎么绘制出图？

可以发一下R代码吗

gonglaijing9 发表于 2021-3-30 16:31
可以发一下R代码吗

不可以； 我现在的代码都是加QQ后收红包后再发的。
不好意思。
你自己问其他人吧。
如果你也发红包， 可以加我的QQ，我发给你。
280201722