Doesn't that cut both ways though? If there are serial correlations in the data then modeling and accounting for the variance explained by those correlations sh

Doesn't that cut both ways though? If there are serial correlations in the data then modeling and accounting for the variance explained by those correlations should help with future predictions, no?