Question: How do you construct portfolios and determine position sizing (weights) of individual stocks?
Answer: We have wanted to discuss this topic for a long time, so here is a very in-depth answer.
For a while in the value investing community the number of positions you held was akin to bragging on your manhood– the fewer positions you owned the more macho an investor you were. I remember meeting two investors at a value conference. At the time they had both had “walk on water” streaks of returns. One had a seven-stock portfolio, the other held three stocks. Sadly, the financial crisis humbled both – the three-stock guy suffered irreparable losses and went out of business (losing most of his clients’ money). The other, after living through a few incredibly difficult years and an investor exodus, is running a more diversified portfolio today.
Under-diversification is dangerous, because a few mistakes or a visit from Bad Luck may prove to be fatal to the portfolio.
On the other extreme, you have a mutual fund industry where it is common to see portfolios with hundreds of stocks (I am generalizing). There are many reasons for that. Mutual funds have an army of analysts who need to be kept busy; their voices need to be heard; and thus their stock picks need to find their way into the portfolio (there are a lot of internal politics in this portfolio). These portfolios are run against benchmarks; thus their construction starts to resemble Noah’s Ark, bringing on board a few animals (stocks) from each industry. Also, the size of the fund may limit its ability to buy large positions in small companies.
There are several problems with this approach. First, and this is the important one, it breeds indifference: If a 0.5% position doubles or gets halved, it will have little impact on the portfolio. The second problem is that it is difficult to maintain research on all these positions. Yes, a mutual fund will have an army of analysts following each industry, but the portfolio manager is the one making the final buy and sell decisions. Third, the 75th idea is probably not as good as the 30th, especially in an overvalued market where good ideas are scarce.
Then you have index funds. On the surface they are over-diversified, but they don’t suffer from the over-diversification headaches of managed funds. In fact, index funds are both over-diversified and under-diversified. Let’s take the S&P 500 – the most popular of the bunch. It owns the 500 largest companies in the US. You’d think it was a diversified portfolio, right? Well, kind of. The top eight companies account for more than 25% of the index. Also, the construction of the index favors stocks that are usually more expensive or that have recently appreciated (it is market-cap-weighted); thus you are “diversified” across a lot of overvalued stocks.
If you own hundreds of securities that are exposed to the same idiosyncratic risk, then are you really diversified?
Our portfolio construction process is built from a first-principles perspective. If a Martian visited Earth and decided to try his hand at value investing, knowing nothing about common (usually academic) conventions, how would he construct a portfolio?
We want to have a portfolio where we own not too many stocks, so that every decision we make matters – we have both skin and soul in the game in each decision. But we don’t want to own so few that a small number of stocks slipping on a banana will send us into financial ruin.
In our portfolio construction, we are trying to maximize both our IQ and our EQ (emotional quotient). Too few stocks will decapitate our EQ – we won’t be able to sleep well at night, as the relatively large impact of a low-probability risk could have a devastating impact on the portfolio. I wrote about the importance of good sleep before (link here). It’s something we take seriously at IMA.
Holding too many stocks will result in both a low EQ and low IQ. It is very difficult to follow and understand the drivers of the business of hundreds of stocks, therefore a low IQ about individual positions will eventually lead to lower portfolio EQ. When things turn bad, a constant in investing, you won’t intimately know your portfolio – you’ll be surrounded by a lot of (tiny-position) strangers.
Portfolio construction is a very intimate process. It is unique to one’s EQ and IQ. Our typical portfolios have 20–30 stocks. Our “focused” portfolios have 12–15 stocks (they are designed for clients where we represent only a small part of their total wealth). There is nothing magical about these numbers – they are just the Goldilocks levels for us, for our team and our clients. They allow room for bad luck, but at the same time every decision we make matters.
Now let’s discuss position sizing. We determine position sizing through a well-defined quantitative process. The goals of this process are to achieve the following: Shift the portfolio towards higher-quality companies with higher returns. Take emotion out of the portfolio construction process. And finally, insure healthy diversification.
Our research process is very qualitative: We read annual reports, talk to competitors and ex-employees, build financial models, and debate stocks among ourselves and our research network. In our valuation analysis we try to kill the business – come up with worst-case fair value (where a company slips on multiple bananas) and reasonable fair value. We also assign a quality rating to each company in the portfolio. Quality is absolute for us – we don’t allow low-quality companies in, no matter how attractive the valuation is (though that doesn’t mean we don’t occasionally misjudge a company’s quality).
The same company, at different stock prices, will merit a higher or lower position size. In other words, if company A is worth (fair value) $100, at $60 it will be a 3% position and at $40 it will be a 5% position. Company B, of a lower quality than A but also worth $100, will be a 2% position at $60 and a 4% position at $40 (I just made up these numbers for illustration purposes). In other words, if there are two companies that have similar expected returns, but one is of higher quality than the other, our system will automatically allocate a larger percentage of the portfolio to the higher-quality company. If you repeat this exercise on a large number of stocks, you cannot but help to shift your portfolio to higher-quality, higher-return stocks. It’s a system of meritocracy where we marry quality and return.
Let’s talk about diversification. We don’t go out of our way to diversify the portfolio. At least, not in a traditional sense. We are not going to allocate 7% to mining stocks because that is the allocation in the index or they are negatively correlated to soft drink companies. (We don’t own either and are not sure if the above statement is even true, but you get the point.) We try to assemble a portfolio of high-quality companies that are attractively priced, whose businesses march to different drummers and are not impacted by the same risks. Just as bank robbers rob banks because that is where the money is, value investors gravitate towards sectors where the value is. To keep our excitement (our emotions) in check, and to make sure we are not overexposed to a single industry, we set hard limits of industry exposure. These limits range from 10%–20%. We also set limits of country exposure, ranging from 7%–30% (ex-US).
In portfolio construction, our goal is not to limit the volatility of the portfolio but to reduce true risk – the permanent loss of capital. We are constantly thinking about the types of risks we are taking. Do we have too much exposure to a weaker or stronger dollar? To higher or lower interest rates? Do we have too much exposure to federal government spending? I know, risk is a four-letter word that has lost its meaning. But not to us. Low interest rates may have time-shifted risk into the future, but they haven’t cured it.