Investing really has two dynamics: picking companies and allocating dollars towards them. Investors often talk about the former, but I’d argue the latter is incredibly underrated. If you own 30 companies but the majority of your dollars are put into underperformers, that’s poor portfolio construction/management. The only thing that matters is how much you own of a stock that goes up a lot. If you only had a 1% position and the stock quadruples, that’s a world of difference from having an original 7% position. Let’s do the math to see how much alpha you would’ve lost out on. Assume that other part of the portfolio compounds at 15% for the year.
Scenario 1
1% position quadruples
Other 99% goes up 15%
Total return: 17.9%
Original 1% is now 3.4%
Scenario 2
7% position quadruples
Other 99% goes up 15%
Total return: 41.9%
Original 1% is now 19.7%
That is a gigantic difference! $1 million turns into $1.18 instead of $1.42 million, just because you allocated 6% more initially. Now 6% more is a good chunk! I usually don’t take starter positions that are that big. Typically, I like to let positions “compete” for extra space in the portfolio. In other words, I kinda let the winners run and naturally become a bigger part.
And that brings me to my current portfolio construction framework. I like to think of it on two vectors: conviction and return profile. A company I have very high conviction on with a stellar return profile will likely have a high allocation. But first, let’s break down these two vectors a little more, starting with the more abstract concept of conviction.
This is the artsier side of investing. It’s a combination of how well I know the company, how much I believe in management, how much of a leader in the industry the company is, how long it’s been in the portfolio, etc. The higher the conviction, the lower the return hurdle I need. Here’s why: the return profile is just a simple earnings model (I explain this in the free financial analysis course if you haven’t taken it yet). Basically I’ll just extrapolate out revenue and earnings and find a reasonable multiple and then see what the return estimate would be. If I have a really high conviction stock like CRWD and it has a lower return estimate than say PTON, I usually default to the strength of conviction because it is so much more comprehensive. And I’ve found that the return profile isn’t usually very precise. Honestly, I’d be fooling myself if I think I could guess where multiples and margins will be in 5 years with any degree of certainty. Doing deep research to understand where my conviction lies is likely a better use of time because I can adapt the model as I learn more about the company.
Let’s make this a little more practical. PTON is only about a 4% position whereas CRWD is a 23% position, yet PTON is growing faster, has higher margins, and a lower multiple. Why is that? Well, it could be ineptitude on my part but the reasoning is that I have more conviction in CRWD and it has earned the right to be the biggest position. On the other hand, PTON has had some manufacturing delays because hardware is inherently much less scalable. I also am more unsure about where growth goes from here with PTON as they have had a lot of demand pulled forward from COVID. This greater uncertainty in future growth usually equates to less conviction. If I knew for sure that a company was going to grow at 100% for 5 years, I can triangulate into a multiple that I’d be willing to pay. But if I had no idea, this exercise becomes much more difficult.
Once again, this is as much an art as a science. In fact, I’ve tried multiple times to create a numerical system for portfolio construction. Maybe it’s a lack of skills but I haven’t really made much progress. It really comes down to weighing a ton of information and trying to decide which pieces will make the most impact on the stock price. But you also need to weigh those important pieces against the companies you already own. That’s why I’ve talked in the past about always having a gold standard company. Right now, that is CRWD for me. For a new company to get in the portfolio, it needs to have a huge return profile or else I’ll just allocate more dollars to CRWD. Either that or I have to do a TON of work to get conviction on this new company. That’s why I’m not always adding new companies to the portfolio. It takes so much time to build conviction and know everything about a company. If I don’t have conviction, then the return profile needs to compensate. It might be a good idea to put some numbers to the relationship between return profile and conviction. It’s probably 75%+ conviction that is responsible for the allocations in the portfolio and the rest is return profile.
I don’t know if that’s the correct mix but that’s what I’ve been doing. The problem with a high return, low conviction stock is that I likely won’t be able to hold it if it underperforms or if it tanks of some terrible, unforeseen news. On top of this, I’m usually skeptical because a high return profile implies that the valuation is pretty low which means I need to do even more work to understand why. On the flipside though, I have pretty high conviction in Snowflake but haven’t bought any because I’m finding it hard to believe in its return profile. I guess what I’m trying to say is that you can think of return profile like valuation. That’s why I trimmed TTD and MDB a little bit ago. I have high conviction on them but I thought the new additions (DOCU, etc.) provided a little better return profile with not much of a drop-off in conviction since I’ve studied them for a while.
The other piece of portfolio management is the number of companies you own. Owning more means you have fewer dollars to allocate to your top ideas. Right now, there are 13 companies in the portfolio and that’s the highest there has ever been. Usually, it’s around 8-10. Mathematically, you get roughly 70% of the diversification benefits from owning just 8 stocks. After that, the incremental benefit drops off quite quickly. This has been a weird time obviously because there are so many companies that are growing very fast and I’m sort of waiting to see how they will do post-COVID. For instance, FVRR may grow 90% next quarter but next year, it’s pretty tough to know where growth will be. Sure, we can guess based on sequential numbers, but the uncertainty is higher and therefore, a lower conviction score.
In hindsight, during the peak of COVID, Zoom was a no-brainer combination of conviction and return profile. Too bad I was focused more on the return profile than having the imagination to think about how integral it would be in a COVID world. That’s one area I want to improve on in terms of portfolio construction: not being afraid to average up. This has cost me serious alpha. I find myself anchoring to stock prices and worrying about buying more when the price has increased. Rather, I could view it as the market validating what I’ve been seeing all along. I am currently trying to improve at this and likely need to build some sort of system to remind myself to do this.
Portfolio construction is hard. I’d like to hear your thoughts or things that work for you surrounding this topic. I’m always eager to learn!
Interested in Infuse Asset Management? Let's talk.