He may be getting that figure from the $950M "$ Exits" figure divided by 450 companies -- but that's incorrect as not all 450 companies have had exits.
I'd imagine that probably 45 companies have had exits, making the avg exit value around $20M, which sounds pretty realistic to me.
That's correct, except most of the companies that have not exited never will (many have already failed,) and it would take quite a few 100M+ exits to significantly skew that average.
Those assumptions could be wrong, but I don't think more than, say $1B in exits, can be expected from the existing YC companies, bringing the average YC exit to $4.5M - not much difference. Even an additional $5B in exits would only bring the average to $11.1M. Most of that would probably come from a few startups (think AirBNB), unfairly skewing the average further - we really should be talking about the much lower median.
Getting into YC brings the average from pitiful to poor. Note there is not a single 100M+ exit that has occurred from any other accelerator listed.
We also have to factor in founder dilution, which is usually massively higher in these big exits, and also that exits are split among the founders. The expected exit for an individual even going through YC is likely sub $1M taking those factors into account.
My point is not that startups are a bad choice. I am even an occasional angel investor. This does reinforce the frequently made observation that startups are not a guaranteed, or even likely, road to riches. Startups are for people who like startups.
It's pointless to talk about the median outcome, because it will almost certainly be zero. It would be very surprising if we had a success rate over 50%. In fact it would probably be bad; it would probably be a sign we were too conservative.
Your estimate of the total value of YC companies is currently off by about 10x.
Your belief that startups are a mediocre way to make money reflects a misunderstanding of probability. In fact they are a very good way to make money for some people, and a terrible way to make money for most people.
I don't think it's pointless to talk about median outcome. The median outcome or worse, by definition, is what most entrepreneurs will achieve, and as you say, that's likely zero even for YC companies who based on this data have an order of magnitude higher success record.
Sorry if my estimate is off - another $10B in exits would change the number significantly, and maybe the filter and assistance of YC is sufficient to change the expected outcome. That doesn't so far seem to be the case with other accelerators.
As you say, "for most people," startups are a terrible way to make money. Based on this data, "most people" is almost all people that get into almost all accelerators. I do think that's an important, though perhaps not new, deduction from this data. It's even more important for people hoping to make money by joining a startup as a non-founder, where the financial upside is so much lower.
I'm not sure what you mean about a misunderstanding of probability. Certainly there is a large amount of both luck and character that determine the outcome for an individual, and that can be summed up with probability, especially for any given startup, since as at a poker table, luck is relatively stronger short term, and character overtakes luck in the long run.
The misunderstanding of probability is analogous to believing that if 3% of the population is over 6' 2" tall, the probability that I, when measured, will be over 6' 2" is 3%. In fact it is 0%.
The overall success rates for founders come from averaging the rates of a few people for whom the chance of succeeding was quite high with a lot for for whom it was zero. Which means they have little predictive value for individuals.
Indeed probability is only useful when there is uncertainty. When there is certainty, probability is nonsensical.
While I think YC may be quite adept at choosing only people over 6'2", people in general are not very good at knowing if they are over 6'2" in entrepreneurial talent. We know people are very bad at judging their own competence, and generally vastly over-estimate it. There is no tape measure for entrepreneurial competence. Unfortunately, I'm not sure anyone, even the YC partner group, is able to reliably measure future success, though I'm sure you are much better than most investors.
Further, an enormous amount of luck is involved, as demonstrated by the large number of one-hit-wonder entrepreneurs, the vast range in outcome for similarly talented people, as well as the large number of impressive entrepreneurs who never really succeed.
So I do think it makes sense to apply probability to your chance of success, no matter who you are. Very talented entrepreneurs still have a high chance of failure, and there is also a good chance that someone is not as talented as they and others think they are. We all know people of both types.
I'd imagine that probably 45 companies have had exits, making the avg exit value around $20M, which sounds pretty realistic to me.