Venture capital portfolios: diversification versus conviction investing

An overview of different portfolio construction models used in the U.S. venture capital market, including pros, cons, and model frequency.

Portfolio diversification within a venture fund can be a divisive subject. Do you spread your capital across a broad array of companies to help reduce risk? Or do you concentrate your bets into a small roster of deals in the hopes that if one of them succeeds, you’re better positioned to win big?  Many fund managers (GPs) fall strongly into one camp or the other. And fund investors (LPs) can be equally opinionated. So what are the pros and cons of each position?


Conviction model

A conviction investing strategy revolves around a small, concentrated portfolio of startup investments. It operates on the premise that funding one outsized winner can meet or exceed the fund’s return targets. Given the low number of deals in the portfolio, the success of the fund is incredibly hit-driven and pushes a fund to own as much of a winning company as it can.  

  • Fund Strategy: These funds often act as lead investors with their initial investments,  in order to buy as much of a company as possible early on, when valuations are cheaper.  As a company matures and raises additional capital, fund partners look to use their pro rata (follow-on investments) to maintain as high a percentage of ownership as possible until a company exits…so long as the partners deem a particular company worthy of continued investment.
  • Key Components: GPs at conviction-oriented funds are typically very hands on with their companies. By having smaller portfolios, GPs believe they are better able to support each company and add value through focus and time. Because of the heavy time commitment to each company, conviction model funds are naturally constrained in the number of investments they can make and manage. Partners at conviction models funds historically have a maximum 1:10 GP to Board Seat ratio, under the belief that higher ratios may be too time consuming for a partner to handle well.
  • Worth Bearing in Mind: Funds may not be able to maintain ownership in follow-on rounds due to a host of factors. One possible challenge is a large investor coming in at a later round and insisting on buying the bulk of that round — thereby preventing the earlier investors from exercising their follow-on rights (and reducing their ownership percentage).  Another possible challenge is lack of dry powder (where the fund already allocated its reserve capital to other companies, or did not have enough reserve capital set aside to begin with).


Diversification model

A diversification investing strategy revolves around making numerous investments across an array of companies (usually 25+) with the goal of finding more winners and mitigating the risk of a fund’s entire portfolio failing. It’s a similar investment strategy to index investing popularized by Vanguard and others, but adapted for the private markets. It operates on the premise that a large diversified portfolio of deals will give the fund more options for follow-on, and potentially more wins in the portfolio. Detractors of this model pejoratively call it “spray and pray”, while practitioners view it as “moneyball for startups.”  

  • Fund Strategy: These funds usually do not act as lead investors, in large part because they make too many investments to carry the workload and time requirements of a lead investor. Instead, they typically write smaller checks in a fundraising round. These fund managers argue that you cannot look at a young startup today and accurately predict in 10 years whether that company will be a massive hit. Therefore, it makes more sense to have a larger portfolio to improve your chances of investing in multiple successful companies. Like with conviction models, diversification-oriented funds also look to maintain their ownership percentage as companies raise more money … but their overall ownership percentage may be lower (than that of a conviction fund) because their initial check size was smaller.
  • Key Components: GPs generally do not take board seats because they do not have the time to proactively monitor such a large portfolio. These funds like to add value where they can, but that value may be more reactive or it may be focused in a particular form of advice or networks. In practice, these funds are usually low touch (once invested) and high volume (in terms of number of investments). The GPs tend to spend most of their time sourcing and vetting investments to meet their portfolio size, translating into lower value add for startup founders. Having a clean structured process for value-add is key to scalability.
  • Worth Bearing in Mind: Diversification model funds can face some of the same risks to follow-on access as conviction model funds. Given the higher number of deals, the fund may also find that a bunch of its companies reach their next funding stage — but the fund may not have enough reserve capital to invest in each one. This can force the fund to make difficult decisions around which companies to follow-on and which to pass. Additionally, this model can struggle as the overall fund size creeps up: for example, if you’re a $200M fund but write lots of small initial checks, you may not have enough ownership in winning companies to generate enough returns to meet your targets.  


Model Variations    

In practice, there’s more variation to venture capital models than simply portfolio diversification versus conviction. Venture studios and accelerators are two hybrid approaches that add to the options in the market.

Venture Studio: a venture studio model is a form of a conviction strategy as the venture firm builds business internally that it believes to be highly promising, and hopes to spin out with high ownership down the road. This model is sometimes paired with a broader venture fund approach and can be used to help diversify the fund’s portfolio. Examples include Pioneer Square Labs and Human Ventures.

Accelerators: more commonly seen as diversification strategies, accelerators invest in lots of different startups over time. That being said, some accelerators opt for very small class or ‘cohort’ sizes (as few as 4 or 5 startups in some cases). For these small cohort accelerators, the managers argue that they are incredibly hands-on and resemble more of a conviction approach. Some accelerators are purely that — accelerators — and invest only at the earliest stages. Others are used to source deal flow for a larger fund, and are designed to invest across multiple stages of a company.


Today’s Market  

In our experience, a conviction portfolio has 10-15 startups, while a diversification portfolio may range from 50 to 100 startups or more. In our 2018 State of Terms Report we found the median and average size of a venture portfolio to be 25 and 29 companies, respectively. That tends to indicate that more early stage funds are leaning towards diversification vs. conviction portfolios.

Examining the market, our data shows some natural clusterings of funds’ chosen portfolio sizes: those targeting 10-15 deals in a portfolio; 18 to 20 deals; 25 deals; 30 to 40 deals; and then a wide range of funds targeting beyond 50 deals.

With respect to portfolio size, it’s important to note that most funds target a range rather than a precise number of deals. A target range can help a fund be more flexible — investing a little more into certain deals, or across a few more deals if enough promising startups cross their desk. For our analysis, we took the middle number of a fund’s specified range.


Does the model fit?    

Venture capital funds exhibit a wide range of target portfolio sizes and approaches. Despite the contention between proponents of conviction versus diversification strategies, both carry pros and cons. And many funds add their own interpretations or nuances to a strategy’s execution, muddling the debate.

While it’s important that a fund’s model and its underlying math align, potential for success is influenced by many factors. At Different, we believe that no single model is ‘ideal’ for all situations. But rather that different models fit different industries, technologies, stages, geographies and individual GPs. So we prefer to ask the question…”Does the model fit?”