Diversification and Position Sizing

For the past several years I’ve been grappling with the issue of diversification and, by extension, position sizing .  My favorite value investors seem to be all over the place on this issue, some opting for very concentrated portfolios, others for relatively diversified portfolios. The only generalization that I arrive at from looking at their portfolios is that the smaller the average size of the companies represented in the portfolio, the more positions held, i.e. the more diversified. Thus, the microcap Hummingbird Fund has up to 100 positions while Chieftain Capital Management, investing only in mega and large caps, has under 10. Of course, this could have more to do with liquidity issues than diversification since funds have to be ready to meet redemptions and microcaps don’t always provide a lot of liquidity. Megacaps, on the other hand, are almost always easy to dispose of.

This generaliztion, however, isn’t very helpful for me as I have no liquidity issues with my investment portfolio; I keep a cash cushion that, along with current dividend income, should be enough to cover my living expenses for up to three years. Therefore ‘liquidity’ is not a factor in my portfolio construction.

The most straight-forward comments on diversification and position sizing that I have run across are from Monish Pabrai. Originally, that is before the 2008 bear market, Pabrai laid out a simple diversification strategy for his portfolio; He took his ten best ideas and invested in each equally. I like this approach. Its simple, neat, not pretentious and relatively easy to implement. Furthermore it avoids what I consider a theoretical quagmire.  I have always had an issue with using some kind of ranking system to size portfolio positions; I don’t see why, if one is really able to rank one’s investment ideas with any sort or accuracy, one shouldn’t simply go with the best couple of ideas, or even the  best single idea. Isn’t the purpose of investing to maximize after tax returns?  and isn’t one implicitly making a probabilistic statement with a ranking system? Why then fall back on diversification if you believe in your own ranking system? I find it contradictory.

So I like the concept of simply dividing investment ideas into good and bad, and then taking a certain number of the good ones, preferably the best ones, and implementing them. What, then, is the right number of positions? I don’t think there is a right number. Somewhere between 8 and 20 positions should provide one with all the theoretical benefits of diversification – In fact, Pabrai’s post crash strategy seems to increase his number of optimal positions from 10 to around 20. And more than 50 or 60 positions in a portfolio its seems to me begins to bring  back the specter of average performance (unless one is microcap territory where not only liquidity but availability may be a factor). From a behavioural perspective I view my own tendency to overdiversify as a contra indicator. I find I want to invest small amounts in a multitude of positions – I am always finding something attractive or potentially attractive – and I have to continually fight (generally with little success) this tendency and bring the number of positions in the portfolio down into my target 20-25 range. To help in this process, I try to keep in mind the potential impact each position could have on my overall portfolio return. Thus, position size is also related to potential return. What impact can an investment totalling less than 1% of investable funds have on the portfolio unless it has the potential for a 10 fold return?  So I do tend to size my positions based on potential return, but based more on some nebulous rationalization having to do with market capitalization than on any forecasting of cash flows.

Then there is the difference between theory and practice. I tend to want to buy into positions in thirds or fourths. Like all value investors I have come to expect that I will be early to the party and begin investing long before the valuation of a security reaches its nadir. The practical application of this strategy, however, leads one to start positions that may remain in the process of accumulation  for quite some time or are simply never completed, either because the share price turns unexpectedly up, or because one is waiting for the price to reach the second, third or fourth buy-in point (which I lay out at the time of  the initial investment). Likewise on the sell side, I tend to cash out step-wise. Thus there are almost always more than the 20-25 target number of  positions in my portfolio at any one time. The real issue here is when to pull the plug on the partial positions; I go through a weening-out process at least semi annually.

I think the value in creating a diversification strategy is not really in determining the exact number of positions to be held in the portfolio, but the discipline that a strategy imparts to the manager. It forces one to continually evaluate each position against another existing or potential opportunity, reduce one position when adding another and size each position for potential impact.  Perhaps it also leads to too much turnover – I tend to think that anything over 25% is too much (implied 4 year average holding period) – but I find this is a small price to pay if it brings a kind of rigorous discipline to ones investing. In the end it is only the discipline of a defined strategy that can help counterbalance the natural human tendency to invest badly.