rlucas.net: The Next Generation Rotating Header Image

July, 2010:

The Secret Trick of Portfolio Effect Dominance

Smart-as-a-whip VC Sim Simeonov did some math and some simulations and came up with the conclusion that a large “portfolio effect” has a major, almost overwhelming, effect on the financial returns of e.g. angel and seed VC portfolios.

English translation: make a LOT of small bets in order to win. (Even if most of those bets are losers.)

Now, this sounds somewhat counter-intuitive to a lot of folks who have been trained to “think like investors:” after all, the more deals you do, the less of a special snowflake you must be, right? (And we all know, Private Equity Professionals are the Specialest Snowflakes of All.) Furthermore, doing more and more “losers” in order to scrape together more winners (rather than trying to avoid losers altogether) just grates the wrong way at the investor mindset.

Well, Sim Simeonov has rejected that mindset with his Simeonov sim. (Forgive me.) But the point of my post isn’t just an emulation of Sim’s simulation. It’s that he’s right despite his simulation math. Rational VCs should be doing as many deals as possible, true, but it’s not due to portfolio IRR; it’s in spite of IRR.

No, the real reason that rational VCs should be doing as many deals as possible is that, to a large extent, VC firm survival has been dependent more on appearance than on financial reality. Specifically, assume two similar firms on their 2nd or 3rd funds, and assume they both, 5 years in, go out to raise funds with a 20% IRR. However, Firm A has earned that IRR through a risk-averse, lower-beta type strategy, while Firm B earned that IRR with a much larger “shotgun” portfolio chock-full of duds, but with a tiny sliver of a Google or similar mega-hit. Which firm is going to be able to raise the next fund with more certainty?

How things have worked to date is that LPs fall in love with great stories, and so it’s more important (in general) that you have one great story in the portfolio (even if it’s for a small investment) than that you have a bunch of sleepy, boring stories that average out to the same return.

Now, I’d love to see what savvy upstream investors like the Super LP, for example, think of this theory: it’s certainly not very flattering to the LP community if you suggest that they make investments solely on “stories” and not on “math.” In the LPs’ defense, I don’t think they’re being lazy or stupid; on the contrary, for the first several decades of the VC industry’s life, the market was so inefficient, and the data were so very sparse, that “stories” were the only reasonable data to look at.

However, we now have a VC (and super-angel, and micro-VC, etc.) industry that is chock-full of history, overflowing with data, and crowded with participants in the marketplace who will (relatively) quickly compete for new niches (especially as the squeeze of a denominator effect and general anti-VC enantiodromia are felt). I no longer feel as strongly as I once did that we’d soon have algorithmic VC decisions, and I doubt that we’ll see a “robot uprising in venture capital.” But I do hope and believe that we will see a more disciplined industry, and one where VCs’ incentives get rearranged to align better with actual financial returns (rather than with “stories” that drive fundraising).

(I do happen to know that there are institutional investors who realize the “great story bias” and are seeking to exploit the inefficiency it creates… let’s just say that if you were going to exploit it, you might look quite hard at Revenue-Based Finance as the way to do so 😉

Can a VC sit on more than 10 boards without f***ing up?

[Update: the conversation referenced below was with Scott Austin of the Wall Street Journal; he has written an article about the subject. In it, he quotes some VCs that have well over ten board seats, including Forest Baskett of NEA. I know Forest only through his and his firm’s reputations (both stellar), and I emphatically do not mean to impugn his board service, commitment, or professionalism. That said, I can’t back down from the math below and I must suggest that overloaded VCs necessarily give some of their boards short shrift.]

I had an interesting conversation today about VCs and board seats. The essential question is: how many board seats is too many?

Happily, the math involved here is pretty low-key (algebra, the highest level of mathematical reasoning to which VCs are required to aspire).

One of the best VCs I know always aspires to be the lead investor. Being the lead investor means you’re the de facto coach, quarterback, and ringleader for that round’s investor syndicate (and likely the entire board). He also tries, both as lead investor and as board member in general, to be a coach, if not confidant and mentor, to the CEO. This is all in addition to the normal duties of a good board member: meetings themselves, prep for meetings, often board dinners the night before, regular if less frequent contact with non-CEO executives (diligence and prudence), recruiting etc., and of course, any audit/comp committee work.

What kind of time does this take, on a monthly basis (4.3 weeks/month)?

  • Weekly CEO calls: 4.3 * 30 min = 2.15 hrs
  • Board meetings q 6 weeks, plus prep: (3 hrs + 1 hr) * (4.3 / 6) = 2.87 hrs
  • Massaging the egos of other board members before/after: 30 min * (4.3 / 6) = 0.36 hrs
  • Monthly CFO or other exec calls: 1 * 30 min = 1 hr
  • Executive recruiting (2 major searches a year, taking 10 hours min. each): 2 * 10 hrs / 12 = 1.67 hrs
  • Committee work (2 meetings a year, taking 3 hours min. each): 2 * 3 hrs / 12 = 0.5 hrs
  • Go to one industry conference a year: 8 hrs / 12 = 0.67 hrs

This is the bare minimum theoretical lower bound that you can consider as the time requirement to be a good, lead investor VC board member: 9.1 hours per board, per month, or just about 2.1 hours a week.

That is for the perfect, steady-state, frictionless world: board meetings are in your town at your firm’s offices (no travel), you do not raise a round, and there are no crises. A more realistic assumption would be to add:

  • Meeting travel q 6 weeks (MINIMUM, even driving up the 280 from Menlo Park to the city takes some time): 1.5 hr * (4.3 / 6) = 1.1 hrs
  • One crisis OR new round per year: 20 hrs / 12 = 1.7 hrs
  • Actually “adding value” like you said you would (soliciting customers, buyers, investors, etc.): 2 hrs /month = 2 hrs

So the real-world minimum adds another 4.8 hrs / month, bringing us to 13.9 hrs /month or 3.2 hrs / week.

How much do VCs really work? I think it’s fair to suggest that VCs work at least as much as other ambitious but affluent, socially-encumbered, and non-hourly-billable professionals: probably on the order of 50-60 hours a week. Let’s call it 55, which would reflect the combination of 10 hour days, 5+ hours each weekend, and a long and/or exotic-enough vacation each year to brag about with the other nouveau-affluent in your social circle.

The real-world catch here is that VCs have to spend a minimum of 5 (and as high as 12) hours at weekly partnership meetings. Let’s call it 6 hours/week to be charitable.

55 hours total – 6 hour partner meeting overhead = 49 workable hours.

49 workable hours / 3.2 hours per board (real world minimum) = 15.3 boards.

So there we have it: 15 boards is the upper bound of what a VC can probably sit on. HOWEVER, this assumes 100% of his working capacity is devoted to board work — nothing here for new deals or fundraising (or for other exotic and occasional pursuits, like strategic planning, learning and research, or leadership and mentoring of junior personnel). That estimate of 15 boards also has what I call the “conceit of optimality,” or inverse-Murphy: it assumes that the crises, new rounds, etc. do not overlap and create impossible time-crunches.

Given that fundraising is THE existential requirement of VC firms, and given that new deal work does have to happen somehow (after all: how did those 15 investments get made??), you’ve got to make significant provision for the working time of a VC to those other, non-board priorities. I personally think that non-board work is at least HALF of the workable hours, but I could be convinced that a board-seat-heavy partner might spend 2/3 of his time on board work.

Therefore, I think that 10 is the maximum realistic board seat capacity of a VC partner who wants to do a reasonably diligent and good job on boards, while also doing the minimum to stay in business as a VC. In practice, I think many boards will take more-than-average time, and I think most VCs will need to spend more time on non-board work, so 7 or 8 is probably a better number to set as a prescriptive maximum.

One could exceed ten board seats without f***ing up in exceptional cases:

  • Independent angel investor (no “firm” overhead)
  • Exceptional geographic and/or industry concentration
  • Evergreen fund, wind-down of a fund, or other nontraditional partner role

Otherwise, you are going to be dropping packets on the floor like a Cogent router in a SQL Slammer epidemic.

One probably needs to hold significantly FEWER than ten (or even fewer than 7) if the following hold:

  • Geographic diversity (have to fly to board meetings)
  • Industry diversity (trying to stay up to date and mine contacts in diverse fields)
  • Series A/B/C rather than later stage weighting (hypergrowth, “chasm,” hiring, and fundraising challenges).