Probably best to try carving out stuff in a core-satellite structure. This is where you have a core asset allocation that is diversified among ETFs, plus a small amount (say 10%) that you then use to make stock picking decisions. Of course, depending on how much you have to invest, 10% may be a very small amount.
In stock picking, there are generally two approaches, which are not quite as distinct as people make them sound… there is a value approach, which prioritizes buying stuff at low valuations, and there is a growth approach, which prioritizes identifying companies with high growth potential. They are actually not that dissimilar, because valuation is still important in the growth approach, it’s just that one is spending much more effort trying to predict and value the growth component of a stock price rather than the value of existing operations. It is hard to avoid overpaying for growth, but essential.
As for your other questions:
Diversification can be achieved with ETFs through as little as 4 or 5 (or arguably even 2), because they diversify across asset classes. We say that diversification is about the correlation of returns, but really it is about the “synchronization” of returns. You want assets that don’t go up and down at the same time, and this is easier with asset classes. It is true that for assets in the same asset class, you generally need about 15 or 20 to protect yourself against ideosyncratic risks, but if your assets are truly unsynchronized, then a portfolio of four uncorrelated ETFs can be more diversified than a portfolio of 20 tech stocks.
However, when you are stock picking, you are basically making bets on the return vs. the level of ideosyncratic risk, so if you feel confident enough in your analysis, you arguably don’t need a super-diversified portfolio. This is what the Buffeteers mean when they say that as long as you’re buying at a big enough discount, you don’t have to worry so much about diversification. I don’t fully believe this, but that’s because I’m never fully sure that if I’ve bought at a big discount, it can’t still go substantially lower, so I do like to diversify more.
For risk tolerance, ultimately there is nothing more to do than to ask you how would you feel if your portfolio lost 10%, 20% 40% 80%, etc in one year… There’s no secret formula in the CFA curriculum that tells you how to convert your feelings to a volatility or VaR number. So here’s my my suggestion: Ask yourself: at what point would you panic, or conclude that you had no idea what you were doing and just give up (as opposed to refining your idea, which would be something different)? Then set your target volatility to something like 1/4 of that. This is based on the idea that people tend to overestimate their risk tolerance, and that normally distributed returns would have annual returns larger than the volatility figure in about 1 out of 3 years, so you need to shrink that number even further.
