That's an obvious thing, but the problem is the wording, because like I explained already, psychological-behavioural traits are in a way part of the phenotype.
The phenotype is already the result of genetic instructions + environmental shaping = phenotypical realisation.
So the question should rather be, which hints are more important for being able to making guesses about an individual's personality, physical traits or genetic analysis?
That depends largely on the quality of both - what can you analyse of the physique (including real data, probably hormone levels and the like, mimic, gesture etc.), what can you analyse of the genome (if you don't know what genes do what or can't analyse the important genes in question, you can say nothing about psychological traits)...
This means currently you can say more about a persons psychological traits if just analysing her with your eyes in most cases, unless there are known genetic factors of great importance, already known, you can analyse.
The more the genetic knowledge grows, the more we will be able to conclude from the genome directly - even though this will be never complete, because of the complicated nature-nurture situation and possible new/rare variants.

