Network centrality in VC
It's not what you know, but who you know.
For 30 years, Christoph Waltz was a brilliant actor no one cared about - until he met Quentin Tarantino. He went on to win two Oscars for his work with Tarantino in Inglourious Basterds and Django.
Waltz tells the story here of how he met Tarantino and of how he met a producer 35 years ago who told him: “‘I can see you’re a fabulous actor. No-one gives a shit. The only thing that matters is who do you know and who you don’t know.’ She was right & now I know Quentin.” Twas ever thus.
The same logic can be applied to entrepreneurship. Upekkha’s Rajan Maruthvan wrote this piece about the invisible networks in Silicon Valley based on education, experience and ambition. Networks enable serendipity. He talks about how “roots beat revenue” when it comes to building momentum (including fundraising).
Network is obviously also critical in VC. Without the best founders selecting a VC, returns will be subpar. 20VC’s Harry Stebbings puts it succinctly here: “network -> data -> decision.”
This topic was recently illuminated by Pitchbook’s report on exits, where they built a network graph for late stage investors (Figure 1) to demonstrate how well connected investors deliver superior returns and have lower failure rates (Figure 2).
Figure 1. Co-investor network for late-stage VC (2022-YTD)
Figure 2. Failure rates & annualized VC returns by lead investors
Yet Equidam’s Dan Gray went on to argue that Pitchbook’s data is a mirage, citing a paper which suggested that network centrality is beneficial to VCs in a hot market and actually negative in a cold market. He gives this as an example of how “venture has been cleaved into the financialised “venture bank” game of riding momentum, versus the traditional venture game of hunting outliers.”
This framing by Dan Gray resonates with our approach. Alpha seeking VCs need to be non-consensus. Beta-like (including smart beta plays like SignalRank) are indeed looking for network centrality.
This post will review some of the academic papers on the benefits/challenges of network centrality that Gray highlights, before considering how SignalRank thinks about network development in VC.
Academic thoughts on network centrality in VC
The argument for network centrality
This paper (“Whom You Know Matters”) by Hochberg, Ljungqvist & Lu examines how the network relationships of VC firms influence their investment outcomes. The study finds that VC firms with more extensive and better-connected networks achieve significantly higher fund performance, as indicated by a greater proportion of successful exits through initial public offerings (IPOs) or acquisitions.
According to the paper, network centrality enables:
Better deal flow: VCs invite others to coinvest in their promising deals in the expectation of future reciprocity
Superior due diligence: by checking each other’s willingness to invest in potentially promising deals, VCs can pool correlated signals and thereby select better investments in situations of often extreme uncertainty about the viability and return potential of investment proposals
Better co-investors: individual VCs tend to have investment expertise that is both sector-specific and location-specific. Syndication helps diffuse information across sector boundaries and expands the spatial radius of exchange
Additional resources: strong relationships with other VCs likely improve the chances of securing follow-on VC funding for portfolio companies, and may indirectly provide access to other VCs’ relationships with service providers such as headhunters and prestigious investment banks
There is an “economically large” impact of network centrality on returns: a one-standard-deviation increase in network centrality increases exit rates by around 2.5 percentage points from the 34.2% sample average. we estimate that this is roughly equivalent to a 2.5 percentage point increase in fund IRR from the 15% sample average.
The argument against network centrality
However, Dan Gray cites this paper (“Temporal Issues in Replication”) by Shi, Sorenson & Waguespeck which critiques the Hochberg et al analysis. The sociology paper by Shi et al. challenges the idea that network centrality is a universal truth. Instead, they argue that network advantages are shaped by time-specific industry structures and that replication of any academic study must account for temporal shifts:
The positive effects of network centrality appear to hold only during the boom periods associated with the 1980s and the late 1990s. By contrast, during the bust periods, the results suggest that centrality might even become a liability.
In particular, this second paper argues that it is only “during the waves of excess liquidity do the valuations become frothy and the competition to invest in certain companies intense” which gives founders the luxury of choosing which VC to select.
How SignalRank navigates this debate
The idea that highly networked VCs perform better in boom times (and worse in downturns) is worth considering from a SignalRank perspective, especially as our investor-level data program favors highly networked VCs. It is especially relevant given the recent excesses of the 2020/21 bubble, exaggerating paper gains and pumping up the whole ecosystem via capital gavage.
We do believe in persistence of returns in venture capital but our system is also dynamic with daily ingestion of data. We mitigate the risk of static network centrality (and specifically the issues of 2021) in various ways:
Our investor ranking model is now based on three year windows of data, meaning that we rank investors based on their investments from 2022 onwards (after the correction).
We are price takers, requiring a lead investor to set the round. Each round is benchmarked against all other Series B rounds in the last 90 days, to ensure that we only invest in rounds in the top 5% from the last quarter. This dynamism protects us from a repeat of 2021 leading to SignalRank spraying capital into too many companies.
SignalRank aims for diversification across 30 companies per year, which will lead to no reliance on any one single manager, whether they are networked or not.
More broadly, we question Gray’s positioning of the Shi paper as refuting the core conclusions of Hochberg. In fact, Shi is simply arguing that there are temporal shifts in the relative importance of network centrality.
How SignalRank thinks about network centrality
SignalRank is a smart beta product that offers our investors diversified access to up to 30 high quality Series Bs per year.
We therefore aim for network centrality by design in order to deliver consistent access to companies backed by high quality active investors.
SignalRank is now demonstrating network centrality in Series Bs. This chart shows how Series B investors with 10+ Series B investments in 2024 co-invested (Figure 3).
Figure 3. All investors with 10+ 2024 Series B investments, with edges showing 2+ co-investments
In order to access these high quality Series B, SignalRank supports the pro rata of our seed partners. We analyze network data on the c.1,200 post A / pre B companies that we believe have signal (“candidates”) to identify which seed managers are existing investors (Figure 4).
Figure 4. Network of SignalRank candidate companies & their existing investors (February 2024)
We then seek to identify how these seed managers co-invest together in order to identify distinct clusters. This chart (Figure 5) shows micro VCs and angels who are in at least five candidates, with edges showing where groups have co-invested. It suggests distinct sub-networks of investment groups. SignalRank should aim to have at least one partner in each cluster to be able to support distinct sub-networks with their respective pro rata in qualifying Series Bs.
Alpha from unique seed managers can enable active Series B investors to deliver beta.
Figure 5. Differentiated micro VCs & angel clusters with 5+ candidate company investments
Some concluding thoughts
GPs & LPs alike instinctively understand the importance of network in the venture industry.
The emergence of differentiated VC products for investors means that LPs can now select for GPs who offer high (eigenvector) centrality or for GPs who offer a non-consensus path.
Greater data on venture and the arrival of superior AI tools also enables investors to better map out their networks, which could be a source of incremental returns.