Most experienced domain investors -- not all but probably most -- would agree with much of what sdsinc is ssaying. The safest bets for investors are .COM and the established ccTLDs.
Most of my sales during February were .COMs. One was actually a .XXX sale for $x,xxx. There were some ccTLDs, .NETs, and .ORGs also.
I think that will remain the balance for the foreseeable future -- mostly .COM, followed by ccTLDs and some good .ORG / .NETs, with a smattering of vanity extensions.
Personally, I think there are strong arguments for .XXX. Adult webmasters are stubborn and cheap. But the ice is gradually melting.
One thing I can't emphasize enough, though, is this:
If you're a new domainer who hasn't yet had a bunch of end-user sales, then you should NOT be investing in the new extensions right now.
I'm buying some. But I've learned how to pick them (to some extent) from my sales in .COM and other extensions. Like all domainers, I started out buying mostly crap. But crap in .COM is a lot cheaper than crap in .WHATEVER. So I could experiment with more. .COM sells faster than anything else. There are more buyers for it. They're familiar with it. Domainers buy it from other domainers, whereas they really don't for the new extensions. And .COM / .NET / .ORG / ccTLDs have proven for years that they sell in high volume for the highest prices.
THAT is what you need to try. That's where you'll learn from sales. If you're only holding risky bets in .WHATEVERs, you might go 2 years without ever experiencing a sale. How can you learn that way?
Start with the least risky domains. Sell those. Then gradually venture in riskier territory if you feel like maybe losing 100% of your money.
Find a .COM domain that has 20+ businesses out there who specialize in what's to the left of the dot. Make sure it sounds right. Make sure it's as good or better than the domains they already have. Find out what domains like that one sell for. If you can get it for 1/10th that amount or less, then considering buying it.