Mongo is obviously not competitive with KDB though, if you are talking of timeseries data as you suggest.

I would think it is in fact very much the wrong tool for the job. Time series data screams structured storage, not unstructured.