The Rise of Intelligent and Data-Driven Venture Investing

Ravijot Singh Narang
5 min readAug 14, 2018

--

VCs have brought very important innovations to the world, but there have been hardly any important innovations in the VC world, to name- ICOs, Crowdfunding, Venture builders, etc.

However, with advancements in Artificial Intelligence and availability of data at almost no cost, the VC Industry is a fertile ground for adopting data-driven strategies to ramp up the deal sourcing and to train gut with intelligence gained from data.

As an investor, there are three important things that you need at your tips most of the time: Where the best companies are? Whether to invest in a particular company or not? What is the optimal distribution of reserves for follow-on investments?

Where the best companies are?

Scouting abilities are often refined through years of networking and branding oneself in different markets and also require VC Associates to spend a lot of time in the field.

However, this is probably not the only way to source good companies, let’s take Hone Capital as an example, Hone Capital created a machine learning model by partnering up with AngelList and drawing data from sources such as Crunchbase, Mattermark (Now FullContact) and Pitchbook Data to explore about 400 different characteristics for each deal and identify 20 common characteristics for seed deals that predict future success (they defined future success as making it to series-A round). Based on the characteristics, the model browses the web to identify and recommend the best startups to reach out to and this model has increased their deal flow up to 20 deals per week.

VC Associates can also refine their process of scouting and analyzing different ecosystems by using intelligence gained from predictive models to learn where to play and why.

By coming up with different factors and quantifying them as shown in the charts below, an investor or VC Associate can understand in which ecosystem the firm needs to channelize its resources, build more relationships and why.

Note: The data used in the graph aren’t real data.

Whether to invest in a particular company or not?

Though data-driven models have great potential to become efficient and bias-free decision-making tools for investors, it’s still far from being a full-stack solution. But what it can readily do is to help investors check their biases revealed through data and help increase operating efficiencies leading to human time being spent on the parts where gut and judgment matters most.

How? Predictive Engine.

A predictive engine can assess the possibility of a startup success (success in VC is subjective and has to be quantified by the VC) by using different predictive modeling techniques and by analyzing multiple factors that take into account making a startup success. This predictive engine will work just the way a credit score is calculated and will assign a score to the founders, their ability to execute and the possibility of success.

A few of the important factors to be taken into account pre-analysis will be: Are the founders technical or have relevant useful experience? Do they have significant domain expertise? Do they have previous startup experience? And have they managed people and budgets before?

Also, the engine has to be able to identify causality (not only correlations) between successful companies and attributes/factors that can be spotted at the time of investing.

Examples of insights that Predictive Engine will generate (apart from the overall score):

  1. The firm should invest in startup A if the management team can show the relevant background in the field
  2. The firm should invest in startup B
  3. The firm should invest in startup C and form an investment syndicate
  4. Startup D doesn’t meet the minimum requirements and hence should be passed

More the number of startups that the predictive engine consumes, stronger will be the accuracy and better will be the insights that the predictive engine will generate.

How to optimize the distribution of funds for follow-on investments?

Enhancing the above predictive engine with basic VC guidelines and using models such as Monte Carlo Simulations, an engine can be built that calculates the optimal distribution for follow-on investments, identify which portfolio companies need more help, etc. . Like an inventory optimization platform, but applied to VC instead of retail.

Future of decision making in VC to achieve the best performance is going to be a combination of the old fashioned gut based judgment and AI-enabled platforms, and firms have already started using data-driven strategies to make investment decisions.

A few that I could find are:

  • Hone Capital: the Palo Alto-based US arm of CSC Group, it partnered with AngelList to create their proprietary model;
  • Correlation Ventures: A real data-driven investor, it reaches a decision on whether to invest or not in 2 weeks, plus other 2 for extra due diligence.
  • Social Capital: Has recently they built analytics due diligence tool (which is hosted on their webpage) to help them invest in early-stage companies
  • GV (Google Ventures): Is using AI and machine learning to inform their investment process

What are some other areas where you think data-driven strategies could impact Venture Capital?

Do share your ideas in comments or connect with me on LinkedIn.

Follow me on Twitter for ideas on Startups, Venture Capital, Creativity & Human Brain.

--

--

Ravijot Singh Narang
Ravijot Singh Narang

Written by Ravijot Singh Narang

Founder @ Acify Technologies. Past: Product, Data Science & Creativity: @meQuilibrium, @Wayfair, @Personify, @Musigma.

No responses yet