Blind Decisions

The extent to which we fail to choose good companies to invest in is often understated. Instead, it seems to be accepted that VC, grant, angel, or any other type of funding method just can't do that much better then other markets. Sure, a hotshot VC may outperform the stock market, but they are not going to get a 10x return on every investment they make. On average, VCs only get that kind of return on 20% of their chosen companies and the rest given them more or less no return at all. So overall, most give boring returns in the 10% range. In cleantech especially, investments has really failed to return the types of returns LPs were expecting (see mass exodus of LP support for cleantech based VCs) 

However, the simple fact that some investments do give yield massive returns should still encourage us to pursue better ways of choosing the companies that are invested in. This should be even more compelling in the cleanteh space where the potential markets are well established and gigantinourmous. But before we can really invent a new investment decision mechanism we need to take the current method, stab it in the heart with a wooden stake and put it in the ground.

To that end, I have recently been experimenting with very simple model that illustrates just how bad the current decision mechanism is at picking companies that will go on to be successful. The model is based on an allegory for our decision mechanism based on the "blindness" of the people choosing which companies should be invested in and which should not. The model gives some interesting insights into how terrible the current system is at really finding good companies to fund. 

The premise of the model is this: There exists a large pool of potential investments that inventors and entrepreneurs are pushing. Think of these like green and blue boxes in a bag where green boxes represent potential investments that will go on to make huge returns and blue boxes will go on to fail. The investors reach in a remove a box and study it, trying to determine if it is green or blue. If they are completely blind, then their eyes are closed every time and they will just randomly pick some and throw away others. If they are 50% blind, then when they pick out a new box then 50% of the time they will open their eyes and be able to see if the box is blue or green and easily either pick the box or throw it away. If they are 0% blind, then they would always have their eyes open and always throw away the blue boxes. 

Clearly, investors never deliberately close their eyes when looking at potential investments, but the analogy is not unfounded. There are many situations where investors are blinded by many factors such as: internal/external biases, lack of technical rigor brought out by the pitching environment, size of market, etc. This "blindness" is really just a figure of merit representing the skill an investor has at seeing a company for its real potential. 

The Model

The model can be downloaded from google drive here: 

https://docs.google.com/file/d/0B2QCuKN1K9FuRURTaWJJUjNYWEk/edit?usp=sharing

The model works using the known failure rates of investments to back calculate the population of the total pool of potential investments using the blindness factor. In other words, it answers the questions "If 80% of all investments are failures and 50% of the time the investors are blind to whether the company is really a good or bad investment then what does the total populations of ideas look like". 

This model is helpful, because right now we have no way of knowing how good investors really are. Yes, 80% of their investments fail, but maybe they are actually really good and 95% of the time when they look at a company they know whether it is a good or bad investment, but there simply are so many bad ideas that the times they do get fooled add up to 80% of their portfolio.  By quantifying the population of good and bad ideas before it goes through the investor screening process, we can get some idea of what blindness factors make sense. In other words, we can start to bound the problem and see how much room for improvement there really is in the investor realm. 

Results

Above, this figure shows the results of running the model assuming 0% eyes open up to 90% eyes open. The total population is based on 10 companies being chosen for funding in each group. So, if an investor is 90% "eyes open" or, in other words, and investor might say "90% of the companies that come across my desk I can tell you if they are good or bad", then for a portfolio where 10 companies are funded and 2 are grand slam hits, there are more than 100 companies in the total population. The figure below gives more detailed data for the 90% case. In the bar graph to the right, the required difference between the total pool and the funded pool is illustrated. 

 

Conclusions

So why should we care about this? The system by which we pick cleantech companies is broken. And if we don't fix it, there wont be any more LP money for cleantech. That means it will be up to grant money to carry cleantech development going forward. And that is not going to work. The first step in fixing this is to show that this system is broken. It is shocking how little discussion there is going on about the fundamental system by which we choose which companies to invest in being broken. 

How does this model help? This model helps illustrate how blind the investment process is to whether a company is good or bad. 75% is pretty bad for a process with such high stakes. but is our system even 75% "eyes open"? If it were, there should be a 60 company total pool for every 10 funded companies and 2 successful companies. If we go by the old verbiage that 90% of startups fail, this would be mean that for 2 successes there would be only 20 companies in the total pool. Using the model, that would mean that investors are "eyes open" only 50% of the time. 

50%... sucks. And yet, is anyone talking about changing the system? If at all, we are talking about meaningless incremental improvements like "capital light". Lets face the facts, our current decision mechanisms are operating blind. They are not telling us what companies are worth investing in. I don't know what the perfect system looks like, but 50% seems like a pretty low bar. I think if we had a realistic dialect about the issues in the current system we could do better. But first things first, the current system is terrible and needs to go.