NavList:

The "native frequency" of the oscillator doesn't necessarily have a direct impact on the quality of the clock, though.  For example, you could take the cesium signal and divide it down to ~1000 Hz, and it would still have the same long-term stability (you'd be adding some noise with the digital dividers).  In the case of the pulsars, the rotation has very high quality, with Allan deviations of something like 1e-14 at time intervals of days.  A little worse for longer intervals, and quite a bit worse for short intervals because of the blurry pulse.  At one time pulsars were thought to be interesting for timekeeping, but now the modern fountains and optical clocks outperform them by several orders of magnitude.

Empirically, though, the higher-frequency signals do seem to be correlated with performance, for a set of not-quite-understood reasons, so I agree with you that far.  The optical standards at ~1 PHz are next in line; they are better than cesium (even cesium fountain).  Also, higher Q (quality factor, narrowness of resonance) also seems to be associated with higher clock performance, but again sort of empirically and not directly.

Cheers,

Peter