how to select outgroup for phylogenetic analysis

[bagaric]

is there some strategy to follow

[mbblack]

There is no rigorous defined method for selecting an outgroup. You obviously want something that is not part of your ingroup but is not so distantly related that inferred homology is difficult or practically impossible.

As to the first criterion, that depends entirely on what your ingroup actually is and the diversity it already covers.

So one way to pick an outgroup for sequence data would be to build a tree with just your ingroup sequences. Take your most divergent branch from that, and do a BLAST search to find something that clearly is not part of your ingroup sequence family, but still retains some evidence of clear homology to them.

Or, if you are inferring an organismal tree, you can just do a literature search for a potential common ancestor to your ingroup, and then pull a sequence for a member of that taxon from available databases.