Preseed VC mindset for public markets

I was explaining Ilay how I look for outlier founders on preseed and seed. The argument was around whether those same instincts could help perform in public markets.

Ilay immediately modeled this. He is my partner at Binomial Technologies, a hybrid long short equity fund built around quantitative research, fundamental signals and AI. The research uses hundreds of signals and ML models to source trades, with Ilay making the final decisions.

Can you invest in public companies like a preseed VC?

Our model defines a “moonshot” as a company that is top 10% of its sector by return AND compounds revenue by at least 15% annually over the next 3 years.

Instead of trying to predict any stock or the portfolio average, it looks for rare outcomes, just like a VC.

One exceptional company matters much more than the rest of the portfolio

And it worked pretty well. Only 5.8% of eligible companies became moonshots. Among the model’s top decile predictions, that jumped to 15.5%, a 2.65x hit rate improvement. The top 1% did even better!

A VC like strategy, pick 20 companies and hold them for 3 years, compounded at 22.5% annually versus 15.2% for the broader eligible universe.

When we allowed the model to replace companies as its conviction changed, similar to a public markets portfolio, the alpha became even larger.

The strongest signals weren’t valuation formulas

The best signals were mostly around where a company was in its life cycle: relatively small, young, recently financed, still founder led and growing quickly.

The model ended up looking for public companies that behaved like startups.

The traits we used were things like how small the company was relative to its industry, how long it had been generating revenue, how recently it had raised equity, whether it is still founder led, how recently it had gone public and how quickly revenue was growing.

The model missed NVIDIA, Palantir and Enphase

It is great at identifying companies that are still early, rather than mature companies suddenly entering a new acceleration era.

The model cannot understand the personal part. The founder story.

It reads financials, growth rates, funding history and press releases. But it does not assess who Jensen Huang or Alex Karp are. It cannot understand why a company that has been public for a while still has a founder capable of creating a 10x outcome.

That soft side is extremely hard to teach the model. And that is where the alpha in VC comes from. The model can narrow thousands of companies down to the potential 100. But the best version of this is not fully quantitative.

Model first, human second.

Let the model create the shortlist. Then use human judgment to understand the founder, the story and what may have fundamentally changed.

It also got badly hurt by the 2021 IPO cohort.

We found some evidence that the mental models venture investors use to recognize companies before they become mature may continue to have value even after those companies go public.

And public markets give you one advantage we don’t have in VC. You can change your mind and adjust your portfolio along the way.

Link to the paper.

‍