The Exchanges

Q Where do you hope this business goes over the next couple of years?

A I really want to put a moat around it because it's so interesting and it gives us a chance to stretch our brains almost daily. There's probably two stages for it. One is the scaling up of the asset management business, putting as much capital as we reasonably can to work without reaching diminishing marginal returns. While there's no competition, eventually competition will come It'll erode our purchase discounts. And at that point, we have a lot of optionality in the business. We will be sitting on one of the best private company data sets that exists given the breadth at which we serve the startup employee base. One of the pet ideas I have for a second stage of the business, which is once the purchase discounts have been competed away and there's lots of competition is that the secondary markets for private shares right now are totally anemic. Maybe a hundred late stage companies Stock trades. The primary reason that's the case is because the buy side on these markets just don't know anything about the other 30,000 venture backed startups. Using our data to help characterize those startups at some minimal level to the buy side of the secondary markets should be a profitable thing to do and should provide a lot of value in the world. There's one world in which we run a proper index business on one hand that's a steady state and then start trying to make the secondary markets w…

AI assessment note: “There's probably two stages for it. One is the scaling up”

Q So once you found these people, or they found you, and you're offering them a deal, what are the important pieces of the deal from their perspective?

A The single most important piece is how much money they need all in to do their full exercise. The pieces of the money are what is your total exercise cost, Which is just a strike price times however many options you have across however many grants you have. Then there is the tax cost, which is the tax that is actually associated with the act of exercising. That is typically applied to the paper gain between the current board approved fair market value of the stock as of the moment you're exercising and whatever your strike price is, which is a disaster. There's no reason that that should be taxed. You can't turn your stock into money when you buy it. It's kind of messed up, but it is what it is, and it's hard to change laws. Let's just assume that that's going to continue forward for a while. If you have incentive stock options, that gain will be applied to the alternative minimum tax regime. If you have non-qualified stock options, it'll be applied to the ordinary income tax regime. And then there is the tax related to actually doing a transaction with us. So our specific transaction is that we're buying some of your shares, which means there's a capital gain event, potentially. Those are the three components. If you've been at a company for a long time, the biggest component is usually the second, which is the tax associated with your exercise. If you've been at a company a s…

AI assessment note: “The single most important piece is how much money they need all in”

Q When you have an approach that came from Such a different insight and different methodology than might think of a secondaries firm that's looking for those larger businesses. What have you heard from the LP community when you've gone out to talk to people about the strategy?

A Oh, you get all kinds of reactions. The folks who think that you need to be in the winner picking business if you are in VC, which almost definitionally means you need to be a primary VC doing rounds for companies. They were never gonna like the strategy, and one day the cash on cash returns will change their mind, and until then, probably not. For the folks that can't even get access to the asset class, this is a godsend. And then there's probably a middle layer of LPs where the most interesting thing is less the access and the return profile, and more that they know they're supposed to be in VC, but they're not sufficiently staffed to properly run a VC fund manager Sourcing diligence relationship maintenance program. So we see all kinds of reactions both substantively to the strategy and also to the practicalities of the strategy and what it means for them as an investor.

AI assessment note: “Oh, you get all kinds of reactions. The folks who think that you need”

Q Which two people have had the biggest impact on your professional life?

A One is Randy Wynn, and the other is Emilio Cejo. Randy was one of the founders and longtime CEO of Capital IQ, but relevant to me, he was my first backer and angel investor and mentor in the first business, and he's been with me all the way through, and he's taught me what it actually means to be somebody's backer. I've lived also in the venture world, and I've seen a lot of venture firms say what they were going to do for you, and not develop any relationship with the people that they invested in, and also frequently fall down in terms of Delivering any value after their capital. I hope I have the opportunity one day to do for somebody else what Randy's done for me. Emilio is the other person. Emilio and Randy and I all met each other around the same time. Emilio was my first partner in that business in which we were building the machine learning models. He is one of the smartest people that I've ever met, but in a way that really opened my eyes about what kind of things you should look for in a person that you work with. He is a mathematical statistics PhD from Columbia. It's easy to put him in a bucket on the basis of that, but when he was rebuilding our liquid asset pricing model inside of the brand name bank, he put on every hat possible and knocked down roadblocks until that thing was algo trading. He got it through legal and compliance, which you have to imagine is a nea…

AI assessment note: “One is Randy Wynn, and the other is Emilio Cejo.”

Q So you had this insight that employees don't really understand their stock options. What did you do with that insight?

A The original version of Vested was just a startup equity education platform where there was a website, and it had free content and tools, and the free content was basic stuff. An article about the difference between stock and stock options, as an example. A tool that would help you calculate the alternative minimum tax associated with your incentive stock option exercise. That's a complicated tool, but everybody needs to know what the tax associated with their exercise is, just like my old guy did. That was the original version of the business, and it wasn't a complicated thing. The idea was going to be that we were going to educate startup employees more than they were, which is not hard. Eventually, we'll have three million startup employees running around our website, and Once they were all here and loved us, we were going to figure out what we could sell them. Mortgages, wealth management referrals, et cetera. That was the original thought. About a year and a half into the business, a lot of the users that were there for the education had started to come back inbound us looking for capital, but they asked questions in these very squishy ways, and so we needed to talk to them to understand what it was they really needed. It was a super interesting experience. We start talking to these startup employees who are really just thinking of us as their Sherpa. In the world of start…

AI assessment note: “The original version of Vested was just a startup equity education platform”

Q I'd love to walk through the different aspects of what you just gave at a high level. The model teasing out what you would have exposure to, and then this other data that comes from the employees. What are those two streams that feed into the model?

A I would characterize it as three streams. I munged two of them into the non-proprietary stuff. There's what I view as the table stakes data, which is financing trajectory for a private company. Just is it going up into the right or not? It's very important that it is, because in the venture asset class, you don't know how the company's gonna do until the music stops, but before the music stopped, it better be going up and to the right. Financing terms. The terms at which investors buy preferred stock in a private company matter a lot for how much risk the common stock is at. Investor quality. By investor quality, I'm not saying that we need to know that the name of the investors backing this company Are Sequoia and Andreessen, but we need to know that they have historically produced good cash on cash returns. Investor behavior, are they continuing to do their pro rata in subsequent rounds and stay involved with the company and put more capital into it and double down on their winners, or are they running away and does every round look like it's a new set of people? That I view as the table stakes stuff. There's a set of differentiated data which lives in the middle, which is not ours, is not proprietary, but To the extent that somebody wanted to build this stuff on their own over the next few years, it's doable. One is a set of financial performance estimates that we have for p…

AI assessment note: “I would characterize it as three streams. I munged two of them into the non-proprietary stuff.”

Q When you put this all together, how do you think about what a portfolio looks like?

A It's a really interesting question. At the beginning, we said to ourselves, in our pilot fund, let's see what happens. What we ended up seeing as the natural shape of our portfolio is we saw that the stage waits, so private companies raise funding rounds and usually, you know, like, angel, pre-seed, seed, series A, B, C, D, E, F, G, until forever. We saw that the natural weight and distribution by stage was roughly uniform from A through, call it F plus, with a little bit underweight in the earlier stages and a little bit underweight in the later stages. Just to fix an intuition on why that is, Series A is usually your first scaling round. You don't have a lot of employees, but also nobody's leaving. Series F plus, those are the companies that are more likely to have proper secondary markets that OpenAI, Stripe, and SpaceX, and they will do tenders for their employees, and so it really is few and far between that those deals make their way into our portfolio. So we're mostly BCD&E is where a lot of the weight is. We're actually comfortable with that. The thing that was scary to us from a stage perspective when we were starting this was SoftBank, Tiger, KOTU, D-One had bid up the late stage companies significantly for a while, especially in We were nervous about the companies whose liquidation preference stack was so high that the common stock of the company should be at risk, s…

AI assessment note: “we're mostly BCD&E is where a lot of the weight is.”

Q You alluded to underground poker games in New York. Any stories to share there?

A Yes, plenty of stories. There was a guy that came and sat at one of the games that I used to play at when I lived in Manhattan, so this is 11 or 12 years ago. Perfectly nice guy, well-dressed. He sits down with a lot more money than everybody else was bought in for at the table. And this is not a small game, so I'm trying to say something. He proceeds to make large blind bets before he even checked his cards out over and over and over again. So the folks that are regulars at these poker games know that now is the moment where you just go into the wait for good cards and then see what happens. He blew through all the money that he came with within two hours. And I sat there totally card dead, waiting for my turn and never having had it come. He was a really nice guy. He wasn't being a pain in the butt at the poker table or anything. He didn't have any problem with losing the money that he came with. And afterwards, when he left, apparently everybody but me knew that he was part of the mafia. And so I got to play with one of those guys once.

AI assessment note: “Yes, plenty of stories. There was a guy that came and sat”

Q How do you think about who the VC backers were as a signal?

A We went back and forth on this when we were building the model in the first place. Originally we were like, let's just take the tier one VCs. And then there's a pause. I was like, okay, who are they? I know some names, but really who are the tier one VCs and are there results and are there performance out there? First, you have to actually agree on who Top quartile VCs are, which is not that obvious and easy a thing to do. Second, around the time that we were doing this, the markets were heating up and then crashing. And we heard whispers that particular funds within even some of the top tier firms that everybody would agree with the top tier firms were battling pretty significant go to zero risk. And we were just like, I don't know if we should be incorporating brand names as a major predictor in our algorithm. Instead, what we did was we split the baby, and we said, we do have a sense for what their cash-on-cash returns have been historically. Let's analyze what their cash-on-cash returns were historically. Let's sort them by that, and let's let the algorithm figure out where the natural cutoff is.

AI assessment note: “Let's analyze what their cash-on-cash returns were historically.”

Q In a power law business, you wouldn't think from the outside you could put a pricing model together and it would have any accuracy, because what you really care about are those power law winners, so how does it work?

A That's right, and we don't believe you can either. One question That follows that theme that you might ask is, why don't you just take your five favorite companies? Why does it need to be the top 20%? And the answer is because I wouldn't believe that the top five was really the top five. At a decently high level of Zoom, All of the work that we've done to date, both on benchmarking our actual portfolio and on robust backtesting, has suggested that we can believe that the top 20% of companies, as determined by our selection model, are really the top 20% of companies. The way that we think about the power law is, one, it's real. Most of the returns are gonna be driven by the big generational companies. Two, picking winners is very hard. Combining those two points, you need to be in every credible deal. We think our selection model does a really good job helping us to identify the pond to fish in, and it's our job to go try to get one unit of everything.

AI assessment note: “our selection model does a really good job helping us to identify the pond”

Q How do you ensure that that delivery occurs?

A We totally obsessed about this as a problem at the beginning, and we've now become so comfortable with it, we don't think about it twice. Originally, we were like, okay, delivery risk is going to be the main risk in this business. We need to get our head around this type of contract that we're using and how it's historically been used and what its delivery and non-delivery rates have been. We found only two examples in the entirety of all the canvassing that we did where non-delivery became a thing. One was an example where somebody didn't actually own the options in the first place. Whoever bought from them didn't validate they were the proper option holder. Which is a super easy thing to get around. You just need people to show you their option grant management account and go to the source, which we do. The second was, instead of doing option exercise, somebody forward sold all of their stock in a particular private company, and then it had a monster liquidity event, and there was a fifty million dollar delivery that was owed, and it was just economically rational for that person to fight it tooth and nail and disappear. Those were the two things that we saw as potential issues, but knowing that those were not issues for us, we proceeded forward and lived our lives. We have had thus far a 100% delivery rate on enough liquidity events at the sample sizes. You can trust the num…

AI assessment note: “You just need people to show you their option grant management account and go to the source”

Q How did you think about accessing capital to address that need?

A Let me walk backwards for a second. There are many different transactional structures that have evolved to help people who need funding for their stock option exercise. All of them have one commonality, which is that you usually end up with exposure to the underlying common stock that you're helping the person buy. You're getting some access to the equity of a venture backed startup. The first thought that we had was there's probably people that had been locked out of the venture asset class writ large. That might think this is an interesting way to get exposure that they otherwise couldn't get exposure. Now, that thought did not go too much further at that moment, because to us, the real trick was figuring out how we could be good investors. The access is interesting, but it's a parallel asset class. A lot of companies go to zero. You gotta figure that out first. We ended up with two evolutions In the business that get us to where we are today. The first evolution was we realized that there were some reasonable purchase discounts that were available to us in helping these startup employees. Specifically, we can get exposure to the common stock that they're buying usually at the independently produced company board approved fair market value of that company's common stock, which tends to incorporate a Discount for lack of marketability because common stock of private companies …

AI assessment note: “people that had been locked out of the venture asset class writ large”