# The best way to make a fortune investing in startups

The strategy of investing in hundreds of early stage startups per year appears, at least in one case, to allow a return of more than 80% year on year. This is the first of a series of blogs addressing this type of strategy, pioneered by Y-Combinator, Techstars, amongst others, and known as the "Accelerator" or "Incubator" model for startup investment. In this blog, I take a look at the annual performance of accelerators through the lens of Y-Combinator and show that its performance is absolutely outstanding compared to any other type of investment or portfolio.

*What is Y Combinator?*

Y Combinator, or YC for short, is a company that, in my understanding, specialises in making large numbers (~200 per year) of small early stage investments (as of late 2014 they typically invest roughly $120,000 for 7% of each company) in startups. YC then provides 3 months of intensive coaching, followed by an opportunity for the startups to pitch to other investors to raise a seed round of funding (typically low seven digit range in exchange for 10-40% of equity as I understand). In the end, the result is that YC ends up owning a small, but not insignificant, stake in a very large number of startups, some of which reach very large valuations.

*How much money is YC making?*

The key to understanding profitability rests in identifying in the average annual growth rate of a $1 dollar investment that an investor/owner might make in YC. To do this I have taken a Warren Buffett type approach and calculated the average *per share* return achieved by YC. The whole reason for using this *per share *approach is because new investments are made in startups by YC every year. This means that as time passes, more and more money is invested in YC (and by YC) and the profits/returns therefore have to be split amongst a greater number of initial dollars invested. The *per share* approach works so that, as more dollars are invested in YC, YC issues more shares as a way to figure out how to split the profits. Therefore, through time, both the number of shares issued and the book value of YC increases, and, what matters for investors is not just growth in book value but growth in book value per share**. To understand more about how the growth in shares is calculated you can read below***. To understand how the book value of YC increases you can consider the following:

For every year I have calculated the book value of Y Combinator. There are two ways in which this can increase.

The first way is that when YC invests in a new batch of startups then its book value increases by the very amount that it invests in these startups. This is straightforward addition.

The second way that book value can increase is if the valuation of any of YC startups increases. To do this with complete accuracy we would have to track every single company. However, a shortcut is possible. Four YC startups (Zenefits, Stripe, Dropbox and Airbnb) account for the majority of YC's value. By considering the increase in valuation of four companies alone, it is possible to get a good estimate of the increase in book value of Y Combinator through time*.

Of course book value could also decrease. Indeed, when startups fold then the money that was initially invested would have to be written off. However, this fact is of little importance for our calculation because the book value of YC is dominated by the value of the large successful startups. The amount invested upfront in failed startups pales by comparison.

[googleapps domain="docs" dir="spreadsheets/d/1KFvTQdLcWsqXyEXWrAhRxngKuyqJfcn1srXf9ulgiks/pubhtml" query="gid=0&single=true&widget=true&headers=false" width="600" height="400" /]

You can download the spreadsheet here: Return on Book Value Calculations.

*Why are the returns offered by YC impressive?*

Annual returns above 80% are absolutely sensational. For comparison, a typical annual return for venture capital over a 10 or 15 year period is around 10% - eight* times lower than what Y Combinator is currently doing*! Of course, the big question here is whether there is enough data to determine whether what Y Combinator has done is more than a stroke of luck. That question will form the subject of the next blog.

*For now, here are some questions to think about (and comment about below):*

Do you think there is enough data to conclude that Y Combinator is truly beating other asset classes by a large margin? Why or why not?

Is there a way to tell whether returns will fall as the program becomes larger? Why or why not?

In 2014, Y Combinator increased the amount invested in each company (and also the valuation at which money is invested) - how might that affect returns moving forward?

*If you look at all of YC's large companies (above $100 million) you find that considering the top four (in fact really the top two) accounts for most of its valuation.

**It is also worth noting that, in practise, investment funds may be set up so that a certain amount of money is committed over 5 years and then profits are returned to investors in the next 5-10 years. This might initially seem inconsistent with the method of considering book value per share where money is effectively *raised *by the investing company as investments are required. However, this is approximately the case because monies that are committed to funds are often not transferred until they are immediately required and so I would argue that the book value per share method is a decent one to use.

***The calculation of the number of (fictitious) shares is best described by an example. Say, in year 1, COMPANY X raises $10 from its owner and invested $1 in 10 startups. The book value of COMPANY X would therefore be $10. Furthermore, let us assume that COMPANY X issued 1 share to its owner (i.e. that is, the owner of COMPANY X holds one share worth $10). Say, now, that by year 2, each of the stakes held in year 1 startups had risen in value to $2. The book value of COMPANY X would therefore have risen to (10 X $2 =) $20 and the price per share would have risen to $20 per share (since there is just one share). Now, let us say that COMPANY X wishes to raise an additional $20 so it can invest $1 in each of 20 different startups. In that case, COMPANY X would issue one new share in return for $20 of investment. The total book value would then be $20 (from the ten *year 1* startups) and $20 (from the twenty *year two* startups). If you think through this carefully, you'll see that if there is no appreciation in the value of startups that are held by COMPANY X then the price per share will stay constant. If the startups appreciate in value the price per share will increase. As previously mentioned, if startups fail they will reduce the book value of COMPANY X because the investments need to be written off. However, this tends to be negligible if there are one or more startups that appreciate to very large valuations (as is the case for YC) that dictate YC's overall valuation.

*This blog should not be interpreted as financial advice. The calculated book values are estimates and the numbers of shares are fictitious. The level of uncertainty involved in dealing with data sets of this size and volatility is high and any conclusions must be subject to careful skepticism.*