Say you would want to build an MVP for a site similar to covestor.com or wealthfront.com. How could you do this without paying huge legal fees?
I think the true definition of MVP in this case is not a fully functional site but rather strong indications of customer intent, which based on specific regulatory issues might only be able to be "sign-up to learn more" at which point you could do some customer interviews. Depending on what exactly is the nature of the site, you might be able to test some other key behaviors and assumptions without tripping-up into regulatory issues.
I've seen companies in this space spend their entire $1m+ initial funding almost entirely on lawyers.
Another answer I wrote on Clarity advised a SaaS entrepreneur to spend as little on legal fees as possible. While the same advice applies here during your MVP, once you want to start building a real service, it's likely that you will have to spend significantly on legal. Ideally, you, a co-founder or a close adviser have a background that reduces some of your legal outlay.
Happy to talk through MVP experiments with you that can maximize your learning.
By definition, an MVP exists to prove some assumptions you have. It may not even need to be fully functional. Depends what assumptions you're trying to validate. A lot of MVPs (including my original one) were just clickable prototypes using Photoshop, HTML, CSS, etc. with not much more behind it.