YC has now funded enough companies that if it has the data (likely in a bunch of spreadsheets) to do a more accurate version of this. Expenses aren't really flat, growth isn't purely exponential and today's estimates aren't reality. A simple model that fixes those three things, should improve accuracy though it would still be subject to enormous volatility which could also estimated based on the same history. This also ignores optionality and probability of a failed raise which would further shift the answer.