Out of Sample Forecasting Performance

Dear All,

Im having some problems with my out of sample performance. Im calculating the out of sample r2 relative to the historical mean but im getting the opposite signs as to what I should expect looking at the literature. Can someone point me in any direction as to what I might be doing wrong? Thanks in advance, see partial code below.

 %%Out-Of-Sample Predictive Regression (1)
%Construct matrix of independent variables
    X1_US=[us_svar lty us_infl tms tbl us_lpay us_lep us_slep us_ldp us_ldy];
%Construct specifications    
    whichstats={'yhat','r','standres','rsquare','mse','beta','covb','tstat','hatmat'};
    sample=80;
    total=152;
    variables=10;
    time=(length(us_p))-sample;
    increments=1;
%Regression log return on matrix   
    for k=1:variables
        for i=1:increments:total-sample
        STATS1_US(k,i)=regstats(us_lret(2:80+i,:),X1_US(1:79+i,k),'linear',whichstats);
        end    
    end 
%Display MSE for each variable    
    for j=1:variables
        for l=1:increments:total-sample
        MSE1_US1(j,l)=(STATS1_US(j,l).r(end,end)).^2;
        MSE1_US(j,:)=mean(MSE1_US1(j,:));
        end
    end 
%R Squared calculation
    for j=1:variables
    R2_1_US(j,:)=1-(MSE1_US(j,:)./US_MSE0(1,1));
    end

The Historical Mean I have defined as follow

 %%Historical Mean
for i=1:72
    us_reth(i,:)=nanmean(us_lret(1:79+i,:));
end 
    US_MSE0=mean((us_lret(81:end,:)-us_reth).^2);

So even though I should expect negative r2 values, Im getting positive ones. Any help would be greatly appreciated!