Conviction, Diversification, and Portfolio Construction

Sometime within the last 48 hours, I tweeted about a potential test for an investor’s true conviction in a company he or she invests in. Specifically, I wrote:

Investment “conviction” is a really overused term. If you want to test it, most investors’ LPAs permit them to invest 10%, in some cases 20%, and for some funds, no restriction, into one startup. Percentage of a fund into one company is the true test.

Luckily, a number of people replied to disagree, including some VCs who have real experience in driving returns across funds.

The first wave of reactions centered around the importance of diversification in a portfolio. I totally agree diversification is an important concept to help protect LPs and maximize returns for the GP. Of course, a fund can over-diversify — like many newer seed funds do — in order to increase the surface area to find a great investment or, simply, because they don’t know any better. I know this is true because I lived it myself.

Ok, sure, so portfolio diversification is wise, yet behind closed doors, we hear many LPs grumble about a fund they’ve backed that’s found 1 or even 2 huge winners but where the check size is so small the returns don’t even return the fund at a multi-billion dollar exit. That may be too much diversification. Then there’s the other end of the spectrum, over-concentration, which also puts the pool of capital at risk.

There are countless posts on portfolio construction, or how many investments are ideal in a venture portfolio, and so forth. I am not qualified to opine on the topic. In fact, I’m hoping folks read this and respond with their own experience and data, as I’d love to learn more about. What I do is that the topic of VC fund portfolio construction is not an easy way to grasp, and is likely even harder to master in practice.

If you open my original tweet and scroll through the thread, you’ll see lots of VCs pushing back on how much of their fund would be allocated to one company. I am stuck here because I am having a hard time seeing how the math shakes out. Sure, if you have a very small fund, or if there’s high ownership for the fund early, or if the fund hits 2-3 monsters, I can see how the math works — absent of that, it’s hard to model out.

So, we have this portfolio construction tension between concentrating money into the best companies in a portfolio versus protecting the GP and LPs via diversification. Things are even more complicated with the fact that top-tier seed funds now boast growth or opportunity funds. My questions for readers would be — how much ownership should be targeted for an initial investment as a function of fund size? How diversified or concentrated are the best fund vintages? What percentage of fund allocated to one company begins to constitute reckless behavior? At what point should a fund introduce a separate side vehicle or process for SPVs for follow-ons to preserve early-stage diversification but also reap the benefits of concentration in things that are more de-risked?

Thanks in advance for your answers and comments. It’s interesting that this whole thread started around the emotional concept of conviction, but veered into the limits of how portfolios are actually constructed.