# Returning The Fund

I have always felt that every investment in a venture fund should be able to return the fund.

That doesn’t mean that they all will.

In fact, for many funds I have worked on, only one or two investments work out well enough that each of them can return the fund.

So if you have a \$100mm fund, you need to look at each and every investment and ask yourself if the company delivers on everything they are seeking to do will that return \$100mm to your fund.

It’s a tall order and doesn’t happen that frequently.

But if it never happens, you won’t be in the venture capital business for long.

1. Yoors

Yes. He is the master !

1. sigmaalgebra

The “master”?I recall: “Yes, he’s the master. It’s closed, but I may be able to get you in.”.All from Madoff.That video shows that Dalio is at best a D- student of his subject. In simple terms, he doesn’t know even “dip squat”.

2. sigmaalgebra

There Dalio shows that he is at best a D- student of his subject. What he said is JUNK.What he is considering, really making a mess of, is the weak law of large numbers I discussed here, yesterday?His first mistake is to regard correlation as a percent. Nope. Correlation is essentially a cosine so has a value from -1 to +1. No percents needed or relevant.His second mistake is to consider several investments but just one correlation for each — total nonsense. Yes, with just two investments there is just one correlation relevant. But with 3 investments, we need 3 correlations.In general for positive integer n and n investments we need(n^2 – n)/2correlations.What he is driving at, making a mess of, is just the first parts of the first H. Markowitz work.And Dalio is not SOLVING the problem but just making a mess out of the FORMULATION of the problem. In particular, after we have the (n^2 – n)/2 correlations, we still have to decide how much to invest in each of the n investments. So, we have a rate of return in mind, and the problem is to invest to minimize the variance of the portfolio while achieving the expected portfolio rate of return.That problem is a relatively simple one in optimization with a quadratic objective function and essentially just one constraint which is linear.Markowitz got a Nobel prize in economics for that work. But some of the Nobel prizes are for some remarkably simple and elementary applied math.Here I will give poor student Ray a little help: Once he goes for the (n^2 – n)/2 correlations, he will usually have trouble getting enough good data to estimate all of them, and he will wonder, rightly so, about how much error in those correlations yields how much error in the final portfolio return and variance.But there is an approach that can be much simpler and much better: In addition to correlation, there is probabilistic independence. And if two random variables X and Y are independent in this sense, then their correlation is easy to find, it’s 0. There are some techniques, e.g., using the Chi squared distribution to test the statistical null hypothesis that there is independence, but in many cases we know about independence — sit down for this — just from what we know generally, intuitively, no arithmetic required. The leading case is the correlation between the first flip of a coin and the next flip. Collect all the data you want to find the correlation; meanwhile it’s essentially just obvious that the two flips are (probabilistically) independent, and that means that the correlation is 0.For more, for two random variables X and Y to be independent it is necessary and sufficient for knowing the value of one of them doesn’t help predict the value of the other one. So, the first flip of a coin does not help predicting the result of the second flip. Knowing if a butterfly flapped its wings or not in Japan does not help predict the weather in NYC.For a little more, in stock investing, we believe that there is a driver common to nearly all the stocks, the market. Soooo, stocks are correlated because they are all driven by the market driven by the GDP, inflation, etc. Soooo, hedge against the market and, presto, bingo, Apple and GM are conditionally independent given the hedge and, thus, have (conditional) correlation 0.Soooo, how to get that dream portfolio with all (n^2 – n)/2 correlations 0? Sure: Pick stocks that are all obviously conditionally independent given the market average. No arithmetic required.Any questions, Ray?

1. David Spiva

whoa, calm down. Did Ray steal your girlfriend or something?

1. sigmaalgebra

Dalio seems to be a big guy in stock market investing. So, he should pass out good stuff. Don’t expect to find bad hamburger at Le Cirque.But, then, I did him a favor: Explained to him to think first about independence, not correlation. Then with independence, the correlations are 0, no arithmetic required.On “calm down”, that’s like a recent song/dance by Taylor Swift? Soooo, she’s 29 now but has a very pretty face and has her weight, figure, and hair like many young men would like to see in a girl of about 13?As a singer, her voice range seems to be four notes within some one octave. So, apparently the boys/men like her as a fantasy girlfriend, and the girls like her as a recipe to be for boys a fantasy girlfriend?

1. awaldstein

So likewise more, smaller funds I presume.

2. William Mougayar

Right, and you can only do that by having the right % ownership in given companies, and following a portfolio approach.This also reminds me of Brad Feld’s famous analogy- “my job is to take a box full of money from investors and return a bigger box with more money in it.”

3. Semil Shah

Maples has mentioned he uses a “RTF Math” framework when considering an investment, just to be clear about what kind of return would be needed at a given ownership level at exit, simply to do the exercise as part of diligence. For me, coming from a non-finance background, I’d say it took a good four years for this concept to finally sink in. It may be obvious to most, but when new folks are starting out, it wasn’t easy to notice or grasp.

1. jason wright

Yes, never put the cart and what it carries before the financial horse that pulls it and the VC jockey that rides it.

4. jason wright

So the bigger the fund is the fewer potential investments there are to be found for it?Big, bigger, biggest (but not necessarily superlative).

5. Michael Aronson

How do you look at initial check size and hoped for x return vs later rounds? Eg if you were willing to invest 5% of the fund in the seed/a a 20x would do it but that would be a very large bet if you had a limit of 10% of fund in anyone company which is typicalBest we have done is 80% of a fund with jet.com by being willing to max out early, although we did an spv in later rounds that allowed investors who participated to get well more than 1x their entire investment in fund and spvThanks

6. Dan Frommer

1. Mike

If you return 3x the fund in 7 years that is an IRR of ~ 17%. A standard pension or retirement fund is probably looking for 7% in aggregate which returns ~2x over 10 years.

7. Tom Labus

This should apply to hedge funs also!

1. Patrick

Not sure how you come to that conclusion for an asset class that is both very diverse and extremely different than VC.

1. Tom Labus

Hedge funds are market neutral I was referring to “not lasting long” without that long ball.

8. pointsnfigures

We do it on every single investment. Game it out all the way to the end. We know our assumptions will probably be wrong, but at least we have some rationale for investing the \$ at the price we invest at with an assumption that the entire fund will be returned with this one check and our LPs will get paid.

9. Richard

Don’t worry, Yahoo got your back.

10. Mcmillan44

this interesting