According to the ROTOR FAQ - http://www.rotorbikeusa.com/support.html - the recommended orientation of the chainring maximizes the effective chainring diameter at the point of maximum power (consistent with theory 1 in Daniel Hicks response). Contrary to Daniel Hicks reply, theory 1 requires the longer axis of the ellipse/oval to be PERPENDICULAR to the cranks rather than parallel. The effective diameter at any point in the pedal stroke is at the point where the chain wraps onto the chainring at the top.
They claim that this happens at just past 3 o'clock. They actually vary this slightly for the inner and outer rings, based on the notion that that varies a bit when you're climbing, presumably in the smaller inner chainring. This is what the default starting position (position 3 in their setup) sets as the peak:
18º for the 53t (outer ring); 23º for the 40t (inner ring)
This is degrees past horizontal/3 o'clock. Rotor chainrings allow you to vary this in 5º to whatever you want. So you could more or less recreate the biopace effect, putting the peak closer to 6 o'clock if you choose.
Rotor does encourage riders to adjust that based on their riding position and their pedal dynamics within a 8º-33º range. The recommended starting point for triathletes/time trialers sitting way forward in the saddle moves the peak position down by another 5º for both chainrings.
Conversely, as noted, you could mount a Biopace ring 72º or 144º off how it is "supposed" to be mounted and achieve something like the Rotor Q-ring effect.
There have been some studies based on actual comparisons of power output. Studies for a 1K time trial show about a 3% increase in average power with Rotor rings. Another study for a 10K time trial with the same Rotor rings show improvement for some of the test subjects, but overall, no significant improvement. There is another study that does computer modeling of muscles and force through the pedal stroke that concludes that the "optimal" shape is like the Rotor ring but with higher eccentricity. They conclude that increasing the effective gearing where you are able to generate more force increases the average power output by increasing the time you spend in your high force zone.
Practically speaking, I am now experimenting with the combination of an outer 53T Q-ring mounted in position 3 (18º peak) and a rotated Biopace inner 42T ring. The inner ring puts the peak closer to a 10º peak (again, angle measured down from horizontal), but it's not easy to compare directly as the shapes are different. The biopace ring has a broader zone that stays at the larger diameter and then the transition to a smaller diameter is quicker.
I found that both were pretty easy to get used to compared to a round ring and that the effect is relatively subtle. With both, I find that it does seem easier to maintain a smooth pedal stroke all the way around.