收益率波动率的计算.doc

NUMPAGES 13 - PAGE 1 收益率波动率的计算 观测值： 收益率： 历史波动率： 移动平均模型：用过去M天的收益率的历史波动率来估计今天的波动率 ，比如取 % using moving average model to estimate volatility % date from 1990-12-19 to 2003-9-19 clear M=[20,60]; m=length(M); z=xlsread(index);z1=datenum(1990,12,19)-z(1,1);z2=z(:,1)+z1; x=z(:,2); x=price2ret(x); n=length(x); for i=1:m for j=1:n-M(i)+1 a=x(j:M(i)+j-1); b(M(i)+j-1,i)=std(a); end end io=find(b==0); b(io)=NaN; plot(z2(2:end),b) datetick(x,23); legend(20 days,60 days,1) xlabel(date) ylabel(volatility) title(volatility of shanghai composite daily return with MA model) 指数加权平均模型： 经验表明：一般对于日收益率，对于月收益率 % using exponential weighted moving average model to estimate volatility clear lamda=0.94;m=20; z=xlsread(index);z1=datenum(1990,12,19)-z(1,1);z2=z(:,1)+z1; x=z(:,2); x=price2ret(x); n=length(x); s=1:m;nlamda=lamda.^(s-1); slamda=sum(nlamda); for i=1:n-m+1 a=x(i:m-1+i);a1=a-mean(a);a2=a1.^2; b(m-1+i)=sqrt(sum(a2(end:-1:1).*nlamda/slamda)); end io=find(b==0);b(io)=NaN; plot(z2(2:end),b) datetick(x,23); xlabel(date) ylabel(volatility) title(volatility of shanghai composite daily return with EWMA model) % x=xlsread(index); y=x(:,2); z=price2ret(y); autocorr(z) parcorr(z) 因此收益率过程可以如下刻画： ARCH模型： % test the arch effect for daily return of shanghai composite clear z=xlsread(index); x=z(:,2); x=price2ret(x); [H,P,Stat,CV] = archtest(x, [1 2 3 4 5 6 7 8 9 10], 0.050); [H,P,Stat,CV] ans = 0 0.2518 1.3135 3.8415 0 0.0525 5.8939 5.9915 1.0000 0.0173 10.1543 7.8147 1.0000 0.0110 13.0564 9.4877 1.0000 0.0224 13.1015 11.0705 1.0000 0.0322 13.7789 12.5916 1.0000 0.0465 14.2728 14.0671 0 0.0718 14.4040 15.5073 0 0.0964 14.8075 16.9190 0 0.1351 14.9173 18.3070 spec = garchset(P,0,Q,7); [Coeff, Errors, LLF, Innovations, Sigma, Summary] = garchfit(spec,x); a=Coeff.ARCH; b=Coeff.K; for i=1:n-6 c=x(i:i+6);c=c.^2; d(i)=a*c+b;