Follow me @samirkaji for my random, sometimes relevant thoughts on the world of venture and start-ups..
In my last post, I illustrated the exit math necessary for a seed fund to generate a 3X net return multiple for its investors (LP’s). The portfolio construction of the fund used in the example is one that is most often seen for seed funds: ~30 portfolio companies, with about half of the fund reserved for doubling down on portfolio companies that show the most promise early-on.
Having met or seen over 300 seed fund managers over the past 5+ years, we’ve observed that greater than 65% of seed funds have portfolio construction models that are very similar to the one noted above.
This statistic is not all that surprising given the sound rationale that accompanies the model above along with the axiomatic nature of venture capital investing. While fund managers go to great lengths to differentiate on the basis of sector, stage, geography, brand, and value-add, they rarely steer from traditional structural items such as portfolio construction and economics.
This is likely in part driven by LP’s, who historically have favored convention over innovation when it comes to the structural elements of venture funds (sidebar: I think LP behavior plays prominently in reinforcing some of the perverse behavior that permeates in venture investing, but that is to be saved for another time).
However, in an industry where risk adjusted return performance for 75% of funds is middling to poor, it begs the question whether standard seed fund portfolio construction should be further evaluated as perhaps one of the contributing factors of poor performance.
It’s well known that venture investing is inherently risky and many controllable and non-controllable variables determine fund success. As such, it’s imperative that venture investors not just what these variables are more importantly how they specifically correlate to their own investment models — — With the number of exogenous factors that affect startup success and the right skew of the value creation curve, venture investing is certainly a probability based vocation. The best VC’s are ones that construct methodologies that continually tilt the odds in their favor.
Given that optimal portfolio construction is an unmistakably driver of fund success, let’s come back to it now.
Despite the fact that non-standard portfolio construction models are often criticized (i.e. terms such as “spray and pray” are derisively used for highly diversified/low ownership portfolios), it stands to reason that standardized portfolio construction model may sub-optimize the probability and magnitude of fund performance if other specific fund factors aren’t considered first.
Fund managers should instead decide on portfolio construction as a function of a thoughtful analysis of firm strengths/weaknesses, investment philosophy, and market realities.
For example, a manager with deep domain expertise and an extremely hands-on operating philosophy may be better served with a concentrated, high-conviction/ownership model than a manager whose primary “superpower” is network relationships. Higher beta consumer funds may have a better probability of success with 50 companies (with lower ownership per company) than an enterprise fund that employs the same portfolio model.
Firms like 500 startups, Cervin Ventures, Wavemaker Partners, Aligned Partners, Precursor Ventures, and SV Angel are examples of firms that have embraced portfolio construction models that have been designed based off their firm’s unique characteristics rather relying on conventional standards.
While we cannot yet measure the exact impact these portfolio construction models will have, positively or negatively, on performance for those firms, it is folly to summarily dismiss portfolio construction models that don’t fall into the “standard” camp.
The point is that in an ultra-competitive seed fund market, every variable should be carefully considered — it might just mean the different failure and success.
And for new managers thinking about fund portfolio construction, remember:
One size certainly most certainly not fit all.