The Importance (and Shortcomings) of a Data-Driven Approach in Venture Capital

Introduction

We live in a world where the quantity and quality of data continue to compound at an exponential rate.

Within asset management, data-driven or quant strategies are certainly not new yet the opportunity to better leverage data (for fund managers and for allocators) remains significant.

Data-driven strategies first emerged in public market (non-access constrained) strategies with early pioneers such as Bridgewater Associates and Renaissance Technologies dating back to 1975 and 1982, respectively. Private market (access constrained) strategies however have generally been slower to follow suit.

At Pattern, we believe that venture capital is uniquely well-positioned relative to other private asset strategies to allow for robust data-driven approaches at both the GP and LP level. As we know, venture capital is a power-law-driven strategy where winners drive a significant share of aggregate returns. These large outcomes share a generally reliable path dependency on their way to an exit which is often tied to a series of subsequent financing rounds. These potential indicators of success are not as widely available in other strategies.

Equally, the quantity and accuracy of such fundraising data continue to improve at the same time that the means through which this data can be accessed and contextualized continues to be streamlined. For example, one can build direct data integrations into various sources of data which thereby offers the potential for robust, efficient and proactive data analysis across large quantities of data in ways that previously would have been impossible to conduct manually.

We’re convinced that the venture capital industry (whether that’s GPs or LPs) will increasingly integrate various forms of data-informed frameworks into their investment processes. GPs were the first movers but LPs are starting to catch up quickly.

This note focuses on the importance (and shortcomings) of a data-driven approach from the perspective of the LP when allocating to venture capital funds.

TL;DR

• Venture investing is uniquely well-positioned for robust data-driven approaches: Data has the potential to enhance every component of the investment process (for GPs and LPs) and allocators, in particular, can unlock significant benefits from better leveraging data.

• Back-tested research and relevant KPIs: Our research suggests that at least a handful of KPIs are highly correlated with future fund performance and we are excited to share more details below. For example, funds generating a score of greater than ~30% on our KPIs are 8x more likely to generate a 5x+ return than the absolute probability of a fund generating a 5x+ return.

• Unlocking and expanding breadth of coverage: Our anecdotal experience suggests that ~40% of funds that we are actively engaged with and who we consider to be in the top decile of funds either have no institutional LPs or are not actively or broadly marketing their funds.

• Shortcomings of a data-driven approach: Data-driven approaches will increasingly help allocators get to the “right” funds for the right reasons (and hopefully at the right time). From there, it will be up to their internal underwriting processes and (GP-facing) products to determine whether or not they can be granted access to the best funds. Data analysis cannot replace these critical components of the fund investment process alone.

• Path forward: If you are a fund manager curious to learn how your prior angel investments or funds rank feel free to reach out to us directly or submit relevant investment information to us via HARDCAP. If you are an allocator interested in learning more about how to integrate a data-driven approach into your fund investment process, feel free to reach out to our team.

Disclaimer: Pattern is exclusively focused on small funds with early-stage strategies. Our frameworks are designed to help us analyze the venture industry through this lens and recognize that these metrics and frameworks might not be appropriate for folks who have a preference for later-stage strategies.

Why Should LPs Have a Data-Driven Approach?

Data has the potential to enhance every component of the investment process for allocators. This is not limited to the process of making a new investment (sourcing > diligence > accessing) but also includes portfolio management (streamlined look-through portfolio exposures) and portfolio maintenance (earlier performance signals when evaluating manager re-ups).

This report will focus primarily on the components that go into making a new investment with a particular emphasis on sourcing. The utility of a data-driven approach will vary depending on the profile and the strategy of the allocator. We share our perspectives on the ways in which leveraging a data-driven approach has been instrumental for our particular strategy.

As a reminder, Pattern Ventures is a specialist fund of funds exclusively focused on partnering with and empowering exceptional small venture funds, $5-50M in size. (See our recent article “ACCESS AND OUTLIERS: The Appeal of Small Venture Funds” to learn why we seek to partner with small funds.)

The appeal of small funds is met with meaningful challenges with respect to every component of the investment process across sourcing, underwriting and accessing. Amongst other factors, there have never been more small and emerging funds than there are today and the size of this ecosystem has grown ~18x over the last decade alone. We outlined some of the challenges presented by this part of the market in our previous report:

As we discuss further below, data can unlock significant value by enabling allocators to efficiently and proactively screen large components of the market in ways that would otherwise be impossible to do manually. However, despite the value it unequivocally provides, data cannot, and most likely will not, replace the other important components of the fund investment process. These “shortcomings” do not diminish the relevance of a data-driven approach but rather highlight the importance of combining such an approach with true experience and expertise in underwriting and accessing the most exceptional funds.

Establishing Relevant KPIs

Our team has built integrations into data providers which has allowed us to create and back test KPIs that we believe are helpful and early indicators of fund performance. Our research suggests that at least a handful of KPIs that are highly correlated with future fund performance. We are excited to share more details below.

These metrics can be considered as early drivers of the more traditional metrics used to measure fund performance (TVPI, DPI, etc.) and can offer a reliable and early signal (observable within 2-4 years of a fund life) into future fund returns.

This compares favorably to what has otherwise been a 6-8+ year manager attribution cycle in venture capital.

Source: Cambridge Associates.

At a high level, the allocator has an opportunity to build any number of KPIs using inputs at the underlying portfolio company level (or otherwise) that are relatively easily accessible from either data providers or funds themselves.

These include but are not limited to:

1. The entry point of the investment by the GP in question.

2. The most recent fundraising round conducted by the portfolio company.

3. Time between financing rounds.

4. The entry point of any “Tier 1” investors on the cap table of the portfolio company.

These “scores” can then be analyzed over a particular period of time that is deemed relevant by the allocator to establish a “hit rate” on any particular KPI.

The Need for Purpose-Built Tools

While the underpinnings of this analysis are certainly not revolutionary, it has been powerful to evaluate and benchmark a large universe of funds on these metrics. We have found that a handful of KPIs appear to offer a particularly strong correlation with fund returns for funds with early-stage investment strategies.

These results are derived by back-testing the correlation between KPI scores and fund returns for fully distributed funds with vintages from 1996-2014, where at least 50% of investments were made at or prior to the seed round. These KPIs are highly relevant for the funds in our “buy box” which all run early-stage strategies. We believe this analysis scales with fund size (i.e., it is relevant for early-stage strategies of all fund sizes) but would suggest the allocator rely on other KPIs for alternative venture strategies.

As an example, below, we show the directional relationship between fund returns and one of our internal KPIs.

We observe relevant thresholds at the absolute score level (~30%) below which the probability of an outperforming fund is extremely low. Similarly, the likelihood of an outperforming fund (3x+ or 5x+ returns) has a strong directionally positive relationship with the absolute score level. The probability of an outperforming fund within these score bands is significantly higher than the aggregate probability of an outperforming fund observed at the industry level.

Funds generating a score of greater than ~30% are 8x more likely to generate a 5x+ return than the absolute probability of a fund generating a 5x+ return.

Pattern Match

What we’ve done at Pattern is to integrate these insights into our sourcing process by building out a proprietary and purpose-built tool called “Pattern Match” that allows us to efficiently and proactively evaluate and analyze a large (~2,000+) universe of funds that we believe to be in our addressable coverage universe.

From here, we can dive deeper into any fund by clicking through to their fund tear sheet which provides a host of relevant information (deployment cadence, entry point breakdown, co-investor analysis, “look-through” portfolio exposures) in addition to their relative ranking across several KPIs we actively track.

This allows us to get up to speed very quickly on the investment profile of a particular fund which then informs our decision to engage with the manager.

This approach provides us with a powerful proactive outbound sourcing function that we can use alongside our more traditional network-driven and other inbound sourcing strategies. For our small fund strategy in particular, this approach provides us with a broader top of funnel than what we believe is available to most allocators.

This is not just a function of the robustness of one’s network. It is driven by the growing evidence on our side that shows that there is a meaningfully sized cohort of small funds with significant fund outperformance who either have no institutional LPs (it’s all their own capital and/or capital they have raised from friends and family) or else they are highly selective about who they wish to market their funds to. These are GPs who are intentionally keeping their fund size small and not widely or actively engaging with the LP community.

Our anecdotal experience suggests that ~40% of funds that we are actively engaged with and who we consider to be in the top decile of funds either have no institutional LPs or are not actively or broadly marketing their funds.

Why Data-Driven Strategies Alone Are Not Enough

While this early traction has been validating, we are the first to admit that data-driven strategies alone are not enough.

Most allocators would benefit from better leveraging data but equally believe that such an approach cannot solve other critical components of the investment process, specifically underwriting and access.

Underwriting Challenges

The limitations of a data-driven approach to underwriting are at least two-fold.

1. False Positives

The first is that even if data suggests a strong correlation with returns, there will inevitably be false positives in the data. Many portfolio companies that would screen well on most trackable metrics can ultimately prove to be a false signal. This is particularly apparent in the current market environment where companies who have raised a series of financing rounds from extremely reputable investors are being significantly marked down or worse.

A complementary approach might be to track non-investment related metrics such as networks and co-investors. A consideration here, however, is that there is a limited opportunity to back-test these particular metrics with realized fund returns. This results in either only tracking a meaningfully dated set of inputs or otherwise making a series of assumptions that would similarly be subject to the prevalence of false positives. This is driven by the conflict between the rate of change in the venture ecosystem and the amount of time it takes (~10+ years) for funds to be distributed.

Nevertheless, we do not believe that the prevalence of false positives alone comprises the integrity of the signal one can observe from a data-driven approach. Instead, it reinforces the need for a more fundamentally driven underwriting process to be used alongside it.

2. Manager Underwriting is Extremely Challenging

Manager underwriting is highly complex and is arguably impossible to solve with a data-driven approach alone. Even if there was irrefutable data to suggest outperformance in a prior fund, there are no guarantees (while recognizing the potential for the persistence of returns) that future funds will similarly outperform.

Underwriting funds warrants a post of its own (for now, Craig Thomas’ post is a great place to start) and requires a combination of significant experience as well as strong networks to do well. Experience provides the allocator with process-driven underwriting frameworks and networks that enable important, primarily “off-list” referencing of a manager and their fund product.

Access is King

Arguably the most critical factor in fund investing is having the ability to access the funds that you wish to commit to and to be able to get that access in the size that’s required for your strategy. Part of the appeal of small funds is their relative accessibility compared to larger, more established funds. That’s certainly not to say that accessing exceptional small funds is easy though.

By their very nature, small funds have less capacity than their larger counterparts and that supply can move quickly. Funds we have committed to have all been heavily oversubscribed reinforcing the need to 1) be early and 2) have a right to win access. While data-driven strategies can help with the former, unfortunately, there is not much they can do for the latter.

We’ve found that the awareness of our data-driven approach drives engagement amongst leading GPs. However, this engagement alone is not enough to rely on to get access in size. While data-driven tools can be helpful to GPs, we don’t expect the most impressive GPs to be willing to provide coveted fund access in exchange for data. Instead, the allocator needs to rely on their “GP-facing product” (i.e., what can they offer GPs beyond capital), their reputation, and their networks to unlock their desired access to the most promising funds.

GP-Facing products can include a range of factors. Some factors that are favored include the potential longevity of the relationship, the mission of the LP organization, the ability to participate in co-investments, and the ability to streamline future fundraising initiatives through LP connectivity or otherwise.

Similarly, reputations and networks are essential because the best GPs are generally hard to reach and are highly selective about whom they wish to partner with. In both cases, they are going to do unsolicited reference checks on their prospective LPs, in which case having the requisite reputations and networks is critical in enabling the LP being referenceable to the GP.  

Some of these factors are structural. The others take a signficant amount of time to accrue.

Conclusion

Even though we are currently in a challenging market, venture will continue to attract growing amounts of capital from an increasingly broader LP community as allocations to private strategies continue to grow over time, particularly in the family office and retail channel. This will most likely drive up the competition to partner with leading GPs, which raises the bar in terms of what it will take for allocators to “win” going forward.

Data-driven approaches will help allocators get to the “right” funds for the right reasons (and hopefully at the right time). From there, it will be up to their internal underwriting processes and (GP-facing) products to determine whether or not they can be granted access to the best funds.