At first I felt the same but I realised it's only because I had gotten used to Pandas. Now for most things, Polars seems more concise.
Oh for sure, polars is way more concise, but so is Kx (kdb). It comes down to learning curves. For a once written code that's not an issue but for daily profiling it surely takes time to get used to.
Impressive, but take a look at this classic article that changed my whole view about millions of backtests.