On the eery predictability of YC startup valuations
While startup valuations are often closely held secrets, Y Combinator recently published a tantalizing glimpse into the value of its top 102 portfolio companies.
The world’s top startup accelerator didn’t divulge the actual valuation of these 102 companies, but by listing them in descending order by valuation, they provided enough clues to make accurate estimates when combined with publicly available data on some of the companies and a bit of data science.
Since Rebel Fund invests exclusively into YC startups, we must deeply understand the outcomes of the YC portfolio from an investor standpoint. So, I asked our brilliant data scientist, Ajay Saini of MIT, to develop a formula that attempts to predict the shape of YC’s portfolio company valuation curve and ‘fills in the blanks’ of missing individual company valuations.
The result was quite stunning: YC company valuations follow a nearly perfect power law distribution¹
The blue dots on this curve represent companies with known valuations from public sources, and the red line represents the predicted valuation of all companies based upon these known data points. Notice how the blue dots fall nearly perfectly on the red line.
The implication of this is that YC startup valuations are eerily predictable. Not in the sense that we can predict the ultimate value of any given company a priori, but in the sense that in aggregate, they fall upon a very predictable power law distribution curve. Perhaps this shouldn’t surprise us because power law distributions come up often in complex systems, including wealth distribution in a market economy.
This curve also arms us with an accurate valuation estimate for companies without publicly known values, and allows us to arrive at some interesting conclusions for investors:
- It’s a mega hits driven business. The top 10 YC companies drive about 75% of the value of the entire YC portfolio of 2000+ startups
- Top 5% YC startups create massive value. The average valuation of the top 100 YC startups today is about $1.5B², and if we assume an average seed stage valuation for these companies of $7–8M, that’s a 200x difference.
- Silicon Valley dominates. 9 of the top 10 (and 18 of the top 20) startups are based in the San Francisco Bay Area.
This data further explains both the allure of seed stage investing and the infamous Y Combinator premium. It also explains high San Francisco rents…
Now, back to investing into YC startups.
 For statistics nerds, the pattern is roughly y = 1/x. Specifically, we fit y = k/(x^c) to the data and found that k = 90500616348.8 and c = 1.38596648346 as constants and the mean percent error is 16.53%
 The median value is “only” $400M, further emphasizing how sharply skewed the distribution is