Edited: Your question actually consists of two different aspects.
Why you should stick with the lowered subscripts with tensors
When there are no superscripts, e.g., $A_\mu$, the subscripts will be in a position you may find “not too low”. However, if there are superscripts present, e.g., $A_\mu^\nu$, then the subscripts will be lowered to give room to the superscripts.
Please note that the \indices and \tensor commands from the tensor package intentionally put the subscripts in a lowered position. So when you type something like
$\tensor{A}{_\mu} \, \tensor{M}{^a_b^c_d}$
The subscripts \mu, b and d will be sitting on the same baseline, as they should be for these two tensor objects.
If all you want is the “not lowered” version, then just type
$A_\mu$
But this, to me, destroys the meaning that A is a tensor, and IMHO leads to bad alignment.
\documentclass{article}
\usepackage{tensor}
\linespread{1.5}\selectfont% for better visual
\begin{document}
The good: $\tensor{A}{_\mu} \, \tensor{M}{^a_b^c_d}$
The bad: $A_\mu , \tensor{M}{^a_b^c_d}$
\end{document}

Why you shouldn’t abuse the \left and \right pair
Please don’t think that always using \left and \right is good. In fact, it is quite the opposite.
This answer by @rgrig quoted from the TeXBook, in which Donald Knuth, the inventor of TeX, explains why manual sizing delimiters are provided.
This other answer showcases examples where \left...\right is too small and when it is too big.
\documentclass{article}
\usepackage{tensor}
\linespread{1.5}\selectfont% for better visual
\begin{document}
The good: $\bigl[ (\tensor{A}{_\nu}) - (\tensor{A}{_\mu}) \bigl]$
The bad: $\left[ \left(\tensor{A}{\nu}\right) - \left(\tensor{A}{\mu}\right) \right]$
\end{document}

\documentclass{...}and ending with\end{document}. We cannot offer much help just with a picture, or some code fragments. – Ruixi Zhang Sep 04 '18 at 04:18