I’m in the position of agreeing with both @Xero786 and @Gryphonknight on differing views.
I think Top30 is the best system for initial matching until AW has settled enough for a good application of Elo. I like Gryphon’s suggestions for an Elo system on an ongoing basis, however. Clever to bring in individual player performances to improve the granularity to make the system more responsive. Using it just on straight total alliance result wouldn’t be responsive enough.
(Hey, @Brobb… maybe Elo CAN work 
)
The reason Elo may be better long term is because ‘good heroes’ doesn’t necessarily mean ‘good AW performance’. If an alliance of very established players (with good heroes) doesn’t wholly embrace AW and are ok with that, they will be matched way above their effective weight if just using Hero Strength. Elo in this case would match them far more appropriately.
If using Top30 hero strength I really think it would have to be tempered with the ‘Opt in’ idea I already suggested so that only the ‘interested’ players in an alliance are influencing the matching.
But an Elo system that @Gryphonknight mentions is neat - nicely self-correcting. Great idea imo.
(Edit: Replacing my damned caps ELO with properly stated ‘Elo’. Tch.)