Last week I began a blog series, Smaller, Earlier VCs Should Invest Differently, on how the growing array of micro-VCs investing in early-stage rounds should think differently. You should read that now before diving into this post, but TL;DR is that these VCs (including us at M25) should not utilize some of the best practices in portfolio theory that have worked well for the larger, traditional VC. These strategies, transferred to our stage and AUM, don’t take account of the vast differences in risk and reward. Today, I’ll first dig into a little-talked-about metric: portfolio variance.
At a high level, portfolio variance in VC makes intuitive sense. Investing in a single startup would be incredibly risky — while the expected value is strong, most startups don’t do well and very few of them reap huge multiples. You could do really well if that single investment is a hit — but you could easily lose it all if it fails. The spread of possible returns are endless.
If you increase the number of investments in your portfolio to have multiple “shots on goal”, you start to decrease the risk of an unsuccessful portfolio, while also capping the possible gains. Can all five startups in a portfolio fail? Yes, but it’s way less risky than a single investments. Can all five startups in a portfolio achieve incredible success? Possibly, but the odds are incredibly small — much smaller than the likelihood of you investing in a single high-returning company. The range of possible returns are clearly narrowing with a five-company portfolio. What about 10, 20, 50 or even 100 investments in a portfolio? As the numbers get higher, it is less and less likely for you to capture all losers or all winners — the variety of likely outcomes becomes more and more “tight”, with a decreasing amount of portfolio variance. To put it succinctly from Investopedia:
Portfolio variance is a measurement (or for us in VC with terrible data, an estimated measurement) of how the aggregate actual returns of a set of securities making up a portfolio fluctuate over time.
As simple as it seems, portfolio variance is a very difficult KPI to gauge, and for various reasons it seems like it is not often discussed thoroughly. Though it definitely should be — we often talk about each deal in a “risk vs. reward” scenario — so why is all of the focus in evaluating VC firms based solely on the potential reward aspect (gross returns, IRR, etc.)? LPs naturally care if their GPs can source great dealflow, effectively evaluate founders/companies and optimally execute follow-on/exit opportunities. Shouldn’t they also care if their GP can structure and manage a portfolio with strong results even if some of the companies are unfortunate? It is clear there are no guarantees in tech investing, and the brightest star can become a dud overnight for completely uncontrollable or unknowable reasons.
When I launched our early-stage VC firm, I built a model to try and assess the different risks (and rewards) that a portfolio would face given different controllable inputs like number of companies and follow-on deployment strategy. Many of my peers across Chicago (or beyond) have seen some version of this model, and we’ve had dozens of hours of debate/discussion that has (ideally) honed it’s accuracy. To be clear, the only thing I know for sure about this model is that it is wrong (much like a founder’s financial projections). While much of the data comes from CB Insights and Pitchbook, the data is not perfect, assumptions had to be made for missing data points and the good data I did have is only historical (we cannot yet know how tech investing returns will be in the future). That being said, it matches with what we both intuitively and factually know: earlier investments will have higher risk/reward, and portfolio variance shrinks as the fund diversifies into more investments. Duh.
Rather than bore you with an Excel spit out, I’ve decided to bring the ol’ box-and-whisker plot out of retirement (though you can tweet us if you want to see the model). Using our model (given the assumptions below), I compared a Series A firm (defined as investing an average of ~$5M on ~$21M postmoney) to an early-stage Seed firm (defined as investing an average of $700K on $4M postmoney) for the following questions:
Question #1: If a typical Series A firm invests in 20 companies (an approximation of a traditional VC strategy), what would it look like if a Seed firm did the same thing?
What stands out to me here, particularly around risk (the returns are interesting, but could be highly dependent on our model and the data we’re using) is the confidence interval range. With the same number of investments, a Seed fund has nearly double the range (4.5x vs. 2.5x for the Series A fund). While the Seed fund’s upside is higher, it also has a lower bound on the confidence interval 20% lower (in relation to the median) than the Series A firm. And looking at the box-and-whisker plot, the predictability of a 20-company portfolio is much more of a crapshoot than the same Series A portfolio.
Question #2: What size of Seed portfolio gives us the same size confidence interval as a 20-company Series A portfolio?
The answer is that the Seed firm needs approximately 70 companies, or 3.5 times the Series A firm, to achieve the same confidence interval range. That seems like a lot, and it is — it would probably necessitate serious strategic changes in how the Seed firm invests (e.g. it likely can’t take board seats in 70 companies). A nice consequence of an increased number of companies is that the typical, median firm actually performs better — the median has increased ~8% from 2.5x with 20 companies to 2.7x with 70 companies. How? The law of large numbers explains how the median will converge to the mean, which is higher than the median in the typical venture capital right-skewed distribution.
So what does this really mean? Well, if an early-stage investor tries to mimic the tried-and-true strategies, they will end up with a much more varied range of portfolio outcomes. They could absolutely crush it, or leave their LPs with a poor-returning fund well below the industry median. One thing they are definitely not doing is ensuring that they are taking on the same amount of risk as the traditional firms they admire. If they wanted the same level of risk, the early-stage fund would have to dramatically increase the size of their portfolio. Interestingly, with more portfolio companies the confidence interval narrows to resemble that of Series A — without sacrificing the higher-potential-reward median that Seed investing should enjoy over later-stage investors. Essentially, you are more likely to beat the larger, traditional VC firms’ returns with the larger portfolio — but that will also likely force significant shifts in strategies for capital deployment, ownership levels, management style and more. But those are conversations worth having another time!
Next week, I continue the series with part 3 on why dilution doesn’t matter to earlier, smaller VCs (and how this impacts things like ownership goals and follow-on decisions).
- Earlier-stage have higher standard deviations (risk) and IRRs (reward)
- There is enough deal flow to maintain the high-quality of all early-stage investments even as the portfolio’s size significantly increases
- Your investment decisions are not prophetic and there is a significant degree of luck/exogeneous factors in the outcomes of your portfolio companies.
Notes on the model:
- This model has each investment for a given stage as an equal-sized investment into an averaged valuation. In reality this will likely not be the case; just remember that the more you vary the size/valuation, the more portfolio variance will increase.
- This model assumes no follow-on (only initial investments) for simplicity purposes. Follow-on allocations can significantly affect portfolio variance.
- Please note that there is a round between Seed and Series A, called “Seed +” in our model, which averages $1.6M on $8M postmoney
A big “thank you” to fellow M25 director, Mike Asem, for his assistance crafting this series, providing feedback and help in editing.
About the Author
Victor Gutwein is the managing director of M25 Group, a VC firm he founded in 2015. Victor is a Kauffman Fellow (Class 22) and an active member and co-chair of the Consumer group at Hyde Park Angels. Previously he has worked in corporate strategy on a variety new businesses in retail & ecommerce. Victor has a passionate history with startups, including a vending machine business and kick scooter company, along with being on the board of the University of Chicago’s first student-run venture fund.
Victor lives with his wife on the South Side of Chicago and loves staying active with backpacking, running, biking and most water sports. If he can’t convince you to workout with him though, he’ll usually succeed in getting you to try out a Euro-style board game (like Settlers of Catan) with his friends.
M25 Group is one of the most active venture capital firms focused solely on early-stage investments in the Midwest. Their objective, analytical approach has helped support their thesis and craft what is known as an ‘index fund of Midwest startups.’ M25 has already invested in over forty companies since their inception in 2015, and continues to invest in over twenty companies each year. Their collaborative, forward-thinking approach and diverse array of investments across industries and business models throughout the region has quickly established them as a key node in the Midwest startup ecosystem.