I'm so glad you're writing this yourself and not trying to adapt some larger open source project that does too much, or a theme that does too little (except look like an auction site).
My thoughts:
Yes, get a holding page together now, capture email addresses and domain sales submissions.
Then get a Minimum Viable Product out asap - something you can handcrank with perhaps with just one auction a week. During this period, say listing is free if you tweet about it, tell X friends and link to the site.
Then build the perfect system taking on board what you learn from the one-off event auctions.
On pricing, you will always be starting auctions higher than you'd expect, and ending lower, so DomainLore's percentage fee for non-selling non-excellent domains (which drives down starting prices) isn't really applicable at either end.
I'd favour a membership model, with the possibility to pay for promotion (eg get featured in regular email).
The moderation is the greatest challenge, time-wise. You want to prioritize good domains, but you need a certain regular quantity, and you do need some things that sell for low because that will attract people too. (And indeed, some buyers may prefer to play with buying a cheaper domain before they are confident to buy a premium domain). If you can devise a way to crowdsource the moderation that can't be gamed by sellers, you'd save a lot of your time.
I strongly suggest that you build in the ability to bundle things in from an early stage.
Sure, keep a headline domain, but allow users to specify what else is included - eg web content, matching .co.uk, matching hyphenated, associated Twitter account etc etc. This sort of thing is really difficult to do on most domain marketplaces and would be another point of difference.