Skip to content

Applying Genetic Algorithm in Entrepreneurial Search

George Soros has a clever way for risk management. In a given year, he would treat his previous year’s principal/initial capital and his previous year’s profit separately. He would treat his principal with the usual risk profile while imposing different risk profile on the previous year’s profit depending on previous performance. He would take more risk if he was aligned with the market, or less risk if he was not aligned with the market in the previous year. With such risk-adjusted capital allocation, he would bet more when he was at sync with the market and gained even more profit or bet less when otherwise and minimize the impact if mistake bound to happen one year from now. For a black box investing model (or at least a grey box one where there is an immense amount of complexity and uncertainty makes it difficult for a precise calculation), this mechanism is truly brilliant as it makes the user doesn’t have to hopelessly grasp the internal mechanics of the black box, but just takes a proxy using an iterative refining of input and output relationship.

Now let’s translate that concept from finance into business. We know that there are numerous factors that makes a business successful.

The central argument of the book Illusion of Entrepreneurship is that not everyone can be entrepreneur. Some people are even better working in an established company because their contribution to the GDP would be higher by working there instead of running an ineffective business. However, it is also true that some entrepreneurs generate so much value that it’s better for them to run a business themselves rather than work in an established company because they will generate more value that way. So, the question is which firm/entrepreneur to support, because not all of them are effective firms/entrepreneurs (some are even subsistence enterprises). So identifying them would mean a greater efficiency of state investment and the return of that investment.

A simple business plan competition can be a search algorithm in identifying potential entrepreneurs, such as one research in Nigeria did.

However, what makes a successful entrepreneur? I like quantitative parameters because they are objective, easy to measure, and hence, also easy to iterate, refined, and improved upon. There are many parameters to choose from, including, but not limited to:

  • The classical “maximizing shareholder value”, reflected in share price
  • Profit, either in absolute value or margin percentage
  • Number of jobs created

Of course there is also the more abstracts values such as serving the community, preserving the environment, etc. While these are great values, we would like to start from the quantitative ones first.

No matter which quantitative parameters to be maximized, all of these depend on a single umbrella factor. Are the entrepreneurs and their firms performing the best that they can?

There are numerous factors that determine whether entrepreneurs and their firms would succeed or not. It’s a hassle to correctly model them upfront, and we lack the data anyway. One approach to solve this is to borrow the genetic algorithm approach.

Let there be a population of entrepreneurs with all the possible traits that could exist in this world. We know some traits is beneficial for running a business, some traits are detrimental, and some traits are irrelevant. But we don’t know which is which (and we don’t have to, at least not in the early phase). We conduct a selection based on our chosen parameter(s) or its derivatives or proxies. This can be in form of competitive entrepreneurship contest where the notion of elimination is valid (it can also shed cost for the whole research process by having not too many participants, e.g. >10,000).

We then set milestones which act as natural selection on these entrepreneurs:
* Initially, each entrepreneurs would receive seed money and sufficient materials to get started.
* The first milestone should be a sales or profitability threshold, such as $100,000 in profit within 1-2 years. If they reach this threshold, some form incentive can be utilized, such as cash grant or additional stake in their company’s equity is bought
* The subsequent milestones should be similar, with increasing sales or profitability threshold within a specified time interval with suitable incentives
* Finally, the research should wrap up by one of three things: a final sales/profitability milestone, an acquisition, or an IPO

The milestone should be designed carefully to not invoke Goodhart’s Law:
“Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes

As applied in economics, the law is implicit in the economic idea of rational expectations, a theory in economics that states that entities who are aware of a system of rewards and punishments will optimize their actions within said system to achieve their desired results. E.g. employees whose performance in a company is measured by some known quantitative measure (cars sold in a month etc.) will attempt to optimize with respect to that measure regardless of whether or not their behavior is profit-maximizing.”

The milestones should also be fine-tuned to take into account the sectoral difference. Some sectors can generate profit faster or slower. Some sectors might have difficulties in getting acquired or do an IPO.

At this point we should have gathered a trove of valuable data. We have the data on entrepreneurs and firms that is initially selected vs those which are not selected. For each milestone, we also have the data on what kind of entrepreneurs and firms that pass that milestone. We can do post-mortem analysis on what factors really differentiates the successful entrepreneurs and firms vs those who do not.

Some hypothesis on factors that affect the entrepreneur:
* Raw intellectual horsepower
* General problem-solving skills
* Emotional maturity
* Access to mentorship
* Access to capital
* Financial literacy
* Conscientiousness
* Leadership

Some hypothesis on factors that affect the firm:
* Sector
* Business model
* Product-market fit
* Logistics & Operations
* Financial planning
* Recruitment and employee development practice
* Competition

At the end of the research, we will both have the economic impact of the winners and also insightful data that can guide us to optimize the production of the next cohort of highly effective entrepreneurs and firms.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: