I've always been afraid of building shopping carts myself. We can do it, but I would lie awake at night thinking about the liabilities involved: Imagine when a customer has $10,000 in orders in a day, and your cart crashes, or something similar... hello lawsuit!
Anyway, when it comes to shopping carts for the few clients I still consult for:
1) Search engine usability is number one. If it doesn't have friendly URL's, forget about it. That knocks out 90% of the stuff out there, I swear. I know you can re-write URL's, etc. in a lot of carts, but that's not the only element important in a well designed page for SEO.
2) Ease of use for the store-owner: calculating taxes and shipping are hugely important. Next comes actually adding items. Then finally integrating things with existing systems.
3) Ease of use for the customer. I hate the carts that make you sign in or get a password. Make the sale easy the first time! So many buyers are first-time buyers that want things immediately.
4) The actual cost of the cart is here. I use to recommend a Freeware cart, but then found one that costs $100 a month to use that's great for SEO... Freeware is useless if you can't make good use of natural search traffic.
But really, there's a lot I *don't* know (anymore) about the ins-and-outs of what's necessary from a day to day level in a good cart. All I know for sure is there's no perfect software I've come across.
Finally: I'm in touch weekly with a guy who's doing $300k of sales a year in a one-man operation, and he's been grunting it out in HTML and Americart (although he's transitioning now to something automated). It's crazy hard work, a lot of duplicating, but it has really worked for him, despite my misgivings.
The point is, cost is one of the least important elements. Yes, when you're starting up its tough to bear.
Keep thinking, though, and keep pushing. And come back and let us know how you do with it in a while.