(You can read more of our original ideas in our archive. You can order a business plan of this idea here.)
Problem: Getting data on the performance of Venture Capital funds and fund managers is extremely difficult.
Subscribe here to get access to the first 500 ideas from our blog. For just one coffee a month, you can have access to more than $500 billion dollars of ideas. What's not to love?
Solution: Working in venture, I’m always a bit inspired when I come across an idea that relates directly to the industry. Recently, I came across one by Yohei Nakajima, an ex-Techstars venture capitalist who now runs Untapped Capital (a pre-seed/seed VC firm investing in unexpected founders). As Yohei describes, “Hear me out, a fund of fund that cuts $100k-$250k into almost every VC fund sub $15m and then spends all the mgmt fees on a data science team to analyze performance of managers and underlying portfolios.”
The Positives of this idea:
It exists in other countries: We have that in Poland. It's called @Grupa_PFR but they cut $5-$20M checks into ~80% of local funds (thanks Borys for the call out!). As Yohei recognized, “That’s really cool and I bet provides great insight for the country. Would love to see more strategies like this.”
Would hold the industry accountable: This could essentially be a trusted, industry-wide “report-card” for Venture Capitalists. Like a CrunchBase or PitchBook, individuals or funds could pay for access to these details.
The Negatives of this idea:
Data quality from LPs is poor: As Mike Joslin describes, “interesting idea but having been at a large VC FoF with data science team using predictive models to identify promising underlying companies I’m skeptical due to data quality. Information rights/reporting varies a lot by fund which rarely includes useful metrics for ml model.” DS Weisman piled on by adding “Can confirm this. Data quality at the LP level is shockingly poor. Less so for PE but certainly for VC.”
Is there even enough data? Adam Feral commented, “Dude. Data scientists need data points. We so better with more data. How many are you talking? Sub 1k points?” In response, Yohei said “I’m imagining 100-250 funds with 20-80 port cos each, each changing valuation every 6-18 mo. Layer on all Crunchbase/Pitchbook type data plus Similar Web, Social, LinkedIn, parse LP updates and standardize data, maybe sentiment/personality assessment of managers/founders… But yeah… a solid data analyst might be enough.”
Q&A with Yohei
(From Josh Taub) What data would need to be captured? How long would you have to wait before you mark a manager or startup as a success or failure? If it was me, I’d parse LP updates and standardize data, layer on Crunchbase/Pitchbook, then SimilarWeb, BuiltWith, AppAnnie, press, social (sentiment analysis, follower analysis), LinkedIn (categorization of past cos/roles), etc as some inputs. Ongoing metrics (IRR, TVPI, “graduation rate”, avg valuation) available in a list, filterable by round or underlying portfolio industry. Eg. an LP could ask, who has the best IRR if you only look at B2B SaaS? Or who has the best graduation rate from preseed to seed rounds?
(From Evan Hynes) Would you consider layering on a marketplace for LP's/etc to discover and invest in funds that are outperforming? Yeah, I think the whole pitch to LPs would be insights and opportunity surfacing for follow-on into the funds and portfolio cos.
(From Andrew Parker) Correlation Ventures is sorta this. And the PA investing into funds by partners at A16Z, sequoia, etc is this, informally. Any LP that does a lot of directs like Industry is also kinda this. Right? Yeah, I’m thinking kind of like Correlation but an FoF so analyzing two layers, & more formalized to provide LPs insight into who to look at for fund 2. Eg Imagine a dashboard where you select an industry and it sorts all the funds by return on investments in that category only.
Monetization: Fees from managing the fund
Contributed by: Michael Bervell (Billion Dollar Startup Ideas)