An old post worth reviewing: Warren Buffett on Diversification - 1966
I was reminded of Buffett's thoughts as I was once again thinking about optimal portfolio size for the "know something" investor. If you can consistently find ideas where you make 50% more when you are right than you lose when you are wrong, then even if you are right only 50% of the time, the Kelly Formula would say that you should be making 16.67% position sizes, or a total portfolio of 6 positions.
While I think those odds and payouts are pretty conservative for the investor that really puts in the work and has a developed a good process, even being more conservative and betting 1/2 Kelly would yield an optimal portfolio of just 12 positions. Considering most of us are looking for traits that give us better than 50/50 odds of being right, and potential upside vs. downside ratios much higher than the above example, I think a portfolio composed of 6-12 core positions also adds some protection for the difficulty, or impossibility, of estimating odds and precise payouts.
Of course, the odds and payouts change as prices change, and certain things may be being bought or sold over time, which could lead to one having a few more positions. But if you are fully invested in things that meet the general purchase criteria for most value investors, and if you are willing to put in the significant effort required to deeply understand each investment, then a portfolio of 6-12 core holdings seems about right to me.
And I think the most common danger area with having this kind of concentration--among the things that one has some control over--is being wrong about the downside risk in each investment, and so that analysis is probably especially important. As Alice Schroeder wrote about Buffett: "He will pass on huge upside opportunities if he can't get downside protection..." Or as Howard Marks often says, if you avoid the losers the winners will take care of themselves.