From “Volatility & Portfolio Construction” (March 2022): “My approach to position sizing has room for improvement. While I question where it lands between being an art or a science, I think my past decision-making has often felt a bit too arbitrary. That’s not to say that I see a need for a complete overhaul, but I do believe my process could benefit from an approach that was more deliberate (Sean Stannard-Stockton, the CIO of Ensemble Capital, helped to plant this seed during his appearance on The Business Brew).”
Today’s write-up is focused on portfolio construction, and specifically position sizing. As Michael Mauboussin once noted, “Almost all investment firms focus on edge, while position sizing generally gets much less attention.” That has been true in my experience, and it needs to change; my goal is to develop a more thoughtful and actionable approach to this critical component of my decision-making process (this will be a work in progress, with today’s output laying the foundation for further refinement over time). Importantly, for the output to be of any value, it must be constructed in alignment with my investment philosophy (a “quality-first”, long-term mindset, and specifically a concentrated portfolio of 10 - 15 stocks with multi-year holding periods).
Over the past 15+ years, my evolution as an investor has greatly benefited from standing on the shoulders of giants; put less glowingly, I will copy from others when I believe they’ve reached logical answers to the questions that are critical to long-term investment success. In the case of position sizing, I believe that the aforementioned Sean Stannard-Stockton, and the team at Ensemble Capital, have shared some great insights (see “Our Approach to Portfolio Construction”, as well as the seven-part series published in 2019).
At a high level, here’s how I’d summarize Ensemble’s views on portfolio construction (which I’ll use later in the post as a starting point for my own approach to position sizing): first, they believe 20-25 stocks is the sweet spot, supported by academic research on a sufficient number of holdings to realize a large percentage of the associated portfolio diversification benefits, while remaining positioned to place meaningful bets behind their highest conviction ideas; second, their individual position sizing methodology is inspired by the Kelly Criterion, with particular focus on expected returns (price versus target) and relative conviction levels among the positions; third, qualitative factors, such as the strength of the moat or the quality of the management team, are ranked / quantified to distinguish between shades of conviction (quantify the probability of success); fourth, those two (quantified) inputs – the return potential and the relative conviction score – inform an initial target weight for each position (with individual weightings scaled back on a pro rata basis if the totals exceeds 100%); and finally, each position is given a drift range / band - when the position weight crosses either side of the limit, a trade will be executed to bring it back in-line with the target weighting (“in the absence of any increase in our assessment of a stock’s fair value, about 15 - 30% outperformance will typically cause us to trim the position”).
On page seven of “Our Approach to Portfolio Construction”, the Ensemble team shows an illustrative example of what this looks like in practice. If a position has minimal upside to their price target (+13%), but with a high conviction rank (10, meaning it’s among the highest quality ideas they’ve found in terms of management quality, the sustainability of the moat, etc.), that position would have a target weighting of ~3.6%. At the same time, an idea with a significantly higher potential return (+76%) but a much lower conviction ranking (1) would receive a similar target weight. In my mind, those numbers are revealing: Ensemble’s approach to position sizing (in terms of the output) clearly places a large emphasis on certainty / conviction. (From the Business Brew podcast - “The Kelly Criterion demonstrates mathematically that it is the likelihood of being correct that is the far bigger driver of how big your bet should be… Position sizing should be driven far more by the likelihood that you will be correct than by how much you’ll make if you’re correct; they’re both important, but that’s the key piece.”)
My own investment approach has evolved over time to align with that mindset; my first filter when considering an investment idea is an assessment of business and managerial quality, with the valuation and expected returns being an important – but secondary – consideration. On that point, I think this discussion between Warren Buffett and Charlie Munger at the 1997 Berkshire Hathaway shareholder meeting is noteworthy; specifically, when talking through his thoughts on opportunity costs, it’s clear that Buffett’s first filter is business quality and certainty, not the valuation or return criteria:
Charlie Munger: “I would argue one filter that’s useful in investing is the simple idea of opportunity costs. If you have one opportunity available in large quantity, and you like it better than 98% of the other things you see, you can screen out the other 98% because you already know something better… People who have very good opportunities, using the concept of opportunity costs, can make better decisions about what to buy. With this attitude, you get a concentrated portfolio, which we don’t mind. That practice of ours, which is so simple, is not widely copied. I do not know why…”
Warren Buffett: “If somebody shows us a business, the first thing that goes through our head is would we rather own this business than more Coca-Cola? Would we rather own it than more Gillette? It’s crazy not to compare it to things that you’re very certain of. There are very few businesses that we’ll find where we’re as certain about the future as we are at companies such as those. And therefore, we will want companies where the certainty gets close to that. And then we’ll want to figure that we’re better off than just buying more of those. If every management, before they bought a business in some unrelated deal that they might not have even heard of a short time before it was being promoted to them, if they said, ‘Is this better than buying in our own stock? Is this better than buying Coca-Cola stock or something’, there’d be a lot fewer deals done. But they tend not to measure it against what we regard as close to perfection as we can get… You simply look around for the thing that you feel the surest about, and that promises the greatest return weighted for that certainty.”
Using Ensemble’s portfolio construction framework as a starting point, I’m going to force rank the TSOH portfolio across four variables: (1) five-year expected returns; (2) strength and sustainability of the moat / competitive position; (3) predictability of business profitability over the next five years (as opposed to the company’s ability to keep growing revenues or other KPI’s); and (4) the quality of the management team (i.e., proven ability over time, turnover in the C-suite, etc.) As shown below, I’ve graded each portfolio position on these variables with a score of one to three (with 1 being the lowest and 3 being the highest), and with an equal mix of 1’s, 2’s, and 3’s.
The output, shown in the right hand column, is the cumulative score for each of the 12 portfolio positions; note that only ~25% of the total is from expected returns, which is in alignment with my “quality first” / business owner mindset.