What Is a Black Box Model? Definition, Uses, and Examples

What Is a Black Box Model?

In science, computing, and engineering, a black box is a device, system, or object which produces useful information without revealing any information about its internal workings. The explanations for its conclusions remain opaque or “black.”

Financial analysts, hedge fund managers, and investors may use software that is based on a black-box model in order to transform data into a useful investment strategy.

Advances in computing power, artificial intelligence, and machine learning capabilities are causing a proliferation of black box models in many professions, and are adding to the mystique surrounding them.

Black box models are eyed warily by potential users in many professions. As one physician writes in a paper about their uses in cardiology: "Black box is shorthand for models that are sufficiently complex that they are not straightforwardly interpretable to humans."

Understanding a Black Box Model

Many things can be described as black boxes: a transistor, an algorithm, and even the human brain.

The opposite of a black box is a system made up of inner workings that are available for inspection. This is commonly referred to as a white box, although it is sometimes called a clear box or a glass box.

The Black Box Model in Finance

Within financial markets, the increasing use of black box methods poses a number of concerns.

A black box model is not inherently risky, but it does raise some governance and ethical questions.

Investment advisors who use black box methods can conceal the true risk of the assets they recommend under the guise of protecting proprietary technology. That leaves both investors and regulators without the facts that they need to accurately assess the risk that is being undertaken.

Do the benefits of black box methods offset the drawbacks? Opinions differ.

Who Uses Black Box Financial Models

The use of black box models to analyze investments has gone in and out of style over the years, usually depending on whether the financial markets are up or down.

During volatile patches in the financial markets, black box strategies are singled out for their potentially destructive nature. The risk levels being undertaken may not be evident until extreme losses reveal them.

Advances in computing power, big data applications, artificial intelligence, and machine learning capabilities are increasing the use and adding to the mystique surrounding black box models that use sophisticated quantitative methods.

Hedge funds and some of the world’s largest investment managers now routinely use black box models to manage their investment strategies.

Black Box Blowups

There have been several notable instances that included extreme losses in portfolios devoted to black box strategies. Black box strategies were not to blame for these events. However, investors who were dependent on those strategies suffered from them, as did many other investors who were caught in the storm.

These events include:

• Black Monday, on Oct. 19, 1987. when the Dow Jones Industrial Average dropped about 22% in one day.
• The collapse of a hedge fund, Long-Term Capital Management, in 1998. The fund made huge profits using an arbitrage strategy to buy bonds until a bond default by Russia's government caused it to collapse, nearly bringing the global financial system with it.
• The "flash crash" on Aug. 24, 2015. Flash crashes, which now occur periodically, involve a short uncontrolled drop in an asset's value, followed by an immediate recovery in its price. An increase in computerized orders is usually blamed. There were actually two flash crashes in 2015. The August event involved the S&P 500 Index and another involving trading in U.S. dollars on March 18.

The Black Box Model in Computing

Machine learning techniques that have greatly contributed to the growth and sophistication of black box models are closely related, particularly relevant to machine learning.

In fact, it has been argued that the workings of black box predictive models that are created from algorithms can become so complex that no human could work through all of the variables involved in making a prediction.

The Black Box Model in Engineering

The black box model is used in engineering to build predictive models that exist in computer code rather than in physical form.

The variables can then be observed, analyzed, tested, and revised without the expensive and time-consuming process of actually building them in the real world.

The Bottom Line

Black box models are increasingly being used to create software not only for applications in the investing world but for use in healthcare, banking, engineering, and other fields.

The black box model is developing in tandem with machine learning capabilities, and both are increasing in the complexity of their processes.

In fact, they are becoming more opaque. That is, we are relying on their results without understanding how those results are produced.