Alex Kirk, Vice President at Verdis Investment Management, is this week’s guest on Swimming with Allocators with Alexa Binns and me. Shane Goudey of Sidley Austin joins for the fund formation segment. Listen here: https://swimmingwithallocators.com/podcast/can-you-index-venture-capital/
Alex Kirk listed the things allocators like before he said it. The manager with a strong prior fund. The spin-out from a brand-name firm. The track record that suggests some kind of repeatable edge. Then he said all of that feels like discipline, and nobody should be undisciplined, and turned the whole thing over. At an early stage, he said, that instinct stops being a tool and starts being a structural bias.
I’ve been sitting with that word since we recorded.
I started in equity research covering REITs, and discipline there had a clear shape. You built the model on your own assumptions, then your boss told you to go check consensus, and if you came in within a cent or two of the street, you felt good. Alex’s framing explains why that worked. Most private market asset classes, he said, look like the classic bell curve, where mean and median sit close together and that middle is just the beta return. When the answer lives near the middle, narrowing is the job.
Seed is a different shape, and Alex has run the arithmetic on it. Because of the power law, the mean return at seed sits far above the median, and historically it lands right around top quartile for the asset class. His conclusion follows directly: “if you back every single seed fund and every single seed company, you really get that top quartile asset class return.” Public markets already solved this, he said: you buy the whole distribution for a few basis points and own every outlier by construction. Nothing like that existed at seed, so Verdis built it. They sit in Philadelphia, started inside a single family office, and write roughly a fund commitment a month, usually into sub $50 million funds, often fund ones. Since 2017: 70 funds, 45 GP relationships, more than 4,000 companies, something like 72 unicorns.
The quotable line is the index math. His real argument is narrower and more useful, and it’s about what you can actually underwrite. Results and track record, he said, are far less controllable, and so harder to underwrite. Process is the controllable part, so that’s where he spends his diligence: portfolio construction, sourcing, the network, how the manager expects to deploy. His favorite question for a fund one is basically homework. Your fund is 25 million, walk us backwards through the math. He called it a back of the envelope conversation, and it’s the fastest way I’ve heard to learn whether a GP has thought about their own business.
Here’s where I pushed, and where I’m still not settled. One of my mentors told me years ago that as a GP it’s critical to understand where you fit in your LP’s portfolio stack, and I’ve repeated that more times than I can count. So I asked the natural follow-on. Are you underweight in one strategy within seed, overweight in another? He said no, not as a dodge. They deliberately don’t tip the portfolio in any direction, they prefer generalists over sector specialists, and they let managers steer exposure based on what founders are building. Then they let the thing cap weight itself. I pressed once more on AI, since a shift that size has to show up as concentration, and he agreed. That’s how the cap weighting works, and a large portion of the portfolio is in AI today.
That answer is clean, and it breaks the advice I’ve been handing out. If you walk into Verdis with a pitch about the gap you fill, there’s no gap. The whole design is to not have one.
The part that stayed with me came later. Alex said something I hadn’t heard put that way: at fund one, your surface area for luck and your surface area for skill are both unusual. Skill may climb across later funds, but luck decays, so fund one is mathematically when a manager is at their luckiest. On air my mind went to music. A band waits its whole life to make the first album, and the second one is brutal, because now they have to produce on a deadline and whatever worked the first time has quietly become the only thing they think they’re allowed to do. Same mechanism Alex described at the top, just pointed inward: pattern matching narrows the LP’s funnel, and early success narrows the GP’s.
What I can’t resolve is how much of this travels. Alex said almost in passing that thankfully he doesn’t have to make the selling decisions. It runs on a very long time horizon and a mandate that doesn’t force liquidity. Alexa asked whether there’s a crew out there running the quant version of venture, some WhatsApp group of ex hedge fund types, and he just said no. My read is that the scarce thing isn’t the insight; it’s the permission. I also went in expecting iteration, and kept asking how the approach had changed as the power law steepened and AI reset pricing. He kept saying it’s been remarkably consistent since they started, and what changed was the sourcing and the data around it. I tried that door three ways, and it didn’t open.
With gratitude,
Earnest