It is hard, bordering impossible, to support unicode combining characters with pdftex.
This is not much of a problem in practice as (if using luatex is not an option) you can usually pre-process to handle the combinations.
For example if instead of using pdflatex directly you use myscript
which is
sed -e 's/\(.\)\xcc\x83/\\tilde{\1}/g' \
-e 's/\(.\)\xcc\x85/\\bar{\1}/g' \
-e 's/\(.\)\xe1\xb7\x89/\\acute{\\grave{\\acute{\1}}}/g' \
-e 's/\(.\)\xcc\x82/\\hat{\1}/g' \
$1 > $1-
pdflatex $1-
then with file.tex
\documentclass{article}
\begin{document}
$x̃, x̅, X̅, f̅, f᷉, f̂$
\end{document}
the command myscript file.tex
produces
The script is written using bash syntax but a .bat file for windows could be used with a suitable windows sed (or perl or python)