I started investing in startups with the assumption that I don't know what I'm doing (which is always true), but that the only way to actually learn anything is through experience. Therefore, my goal was to invest in a variety of startups, learn from the experience, help some startups, and hopefully not lose too much money while doing so. I don't have any single criterion for investing. Sometimes, the idea seems good, sometimes the people seem good, and other times I'm just curious to see what will happen. No matter what though, I want to be helpful and learn something interesting from the experience.
I've definitely learned a lot, but the recent Heroku acquisition (for a reported $212 million) made me wonder if I've also reached the point of "not losing too much money", so I did the math. From 2006 (when I started investing) through 2008, I invested about $1.21 million in 32 different companies. About half of those companies were either acquired or are dead/mostly-dead. The other half are still alive and seem to be doing well. The current acquisitions total about $1.34 million, only about a 10% gain, which isn't great, but at least I'm not losing too much. Fortunately, the "still alive" category includes a number of very promising investments, such as Meraki, Weebly, and Wufoo, so I expect the final return will be much better than 10% (which is all gravy anyway, since my primary goal was to learn more and be helpful).
Of the current acquisitions, only two have yielded a greater than 10x return: Heroku and Mint. Unfortunately they were also two of the smaller investments, proving that I don't know what I'm doing, or at least showing that I need to make a point of investing more money into the best companies (Mint was oversubscribed, but I don't remember why I didn't put more into Heroku -- edit: apparently it was also oversubscribed).
It's worth noting that the highest return was from a Y Combinator company (Heroku, winter 2008). I've been investing in the YC companies almost from the start (Wufoo was winter 2006), and they keep getting better. YC is attracting the best founders, and Heroku is just the tip of the iceberg. The more great YC companies there are, the more reasons there are for other smart founders to join YC, and I find myself asking non-YC companies why they aren't yet in YC. This trend definitely contributed to my decision to join YC as a partner.
There were also a couple of medium-sized acquisitions (AppJet/Etherpad and 280North) and several smaller but still nice (2x-3x) exits such as Auctomatic, Parakey, and Zenter. Sometimes people complain about these deals, but as much as I'd like to get a nice 10,000x exit, I'm certainly not going to complain when someone doubles my money!
A few companies (such as ScanScout) were acquired by other private companies, so I include those in the "still alive and doing well" category, since it was not an exit from the investor perspective (no liquidity). Only two companies are officially dead, but there are at least four "zombies" that still exist, but in a minimal form and are almost certainly worthless.
The two bits of advice that I always give to people looking to get started with angel investing are: 1) Assume you'll lose your money and 2) Plan on investing in a large number of companies. (my third bit of advice is to attend YC demo day) I think my experience so far validates this advice. It's important that investors have the right motivations (helping out startups and learning how to be a better investor) and the right expectations. Anyone doing it with the idea that they can simply find the next Google, invest a bunch of money, and then get super-rich is going to be very disappointed. That said, finding the next Google and buying a 1% stake is my current billion dollar plan :)