His attention was drawn to something unusual while he was visiting the Van Gogh Museum in Amsterdam some time ago. Wang, a professor of management and organizations at Kellogg, says Van Gogh’s style and subject matter restricted after he moved to southern France in the late 1800s. Van Gogh’s most successful phase began with this change, which included classic works like Starry Night and Van Gogh’s Chair.

Wang was finally able to answer a topic he had pondered for a long time: What causes a hot streak?

From Albert Einstein’s “Miracle Year” in which he discovered the Theory of Relativity to Peter Jackson’s box-office triumph with the Lord of the Rings trilogy, “hot streaks,” or career-defining moments of remarkable success, have been noted in almost every creative field. Wang’s prior research revealed that hot streaks happen at random during a career, like a comet with no predictable timing.

Wang believes that “hot streaks” happen to most people. “However, it appeared as if they were brought about by’magic’ or simply by chance.” In other words, there was no credible precedent for these periods of tremendous performance.

Because of their visit to the Van Gogh Museum, Wang and his colleagues devised a study that utilized artificial intelligence to examine millions of works by painters such as Vincent van Gogh, directors such as Steven Spielberg, and scientists such as Charles Darwin.

Hot streaks are often preceded by a combination of exploration and exploitation, which narrows the focus on a certain type of activity to produce outsized performance across other domains.

As a coauthor of the study, Jillian Chown, an associate professor at Kellogg, explains, “We’ve discovered one of the first identifiable regularities associated to the beginning of a hot streak. Individuals and organizations can utilize it to better understand what activities to engage in and how to use them in the most effective way possible.”

A lack of standardization in the way creative works are described has been a major barrier to understanding what causes a hot streak until this study. As an example, what makes up an artist’s style? Is it possible to classify a filmmaker’s different works?

For this task, the researchers created a deep neural network for three domains: art, cinema and science, with the help of Lu Liu, a PhD student at Kellogg’s Center for Science of Science and Innovation; Nima Dehmamy; and C. Lee Giles, a research assistant professor at Kellogg. Visual-recognition technology, for example, was used to identify painting subjects and brushstrokes, as well as to identify film genres and styles by evaluating narrative summaries, cast members or other internet data. They examined the careers of 2,128 painters, 4,337 directors, and 20,040 scientists using this new method, resulting in millions of pieces of work.

As a result of their work with the neural network, researchers were able to understand how each person’s work changed over time in terms of both exploration and exploitation.

For a hot streak, the study looked at the number of citations of scholarly papers, art-auction prices, and IMDB film ratings as indicators of success in each discipline.

We were able to correlate hot streaks with the creative trajectory of each individual, and we were able to identify what would anticipate these streaks,” Wang explains.

To achieve a hot streak, one must first explore all creative possibilities, then focus on a single “lane” of work and exploit it to its fullest potential. Across all three domains tested, this was the case.

Director Peter Jackson made a number of movies in a variety of genres prior to making the Lord of the Rings fantasy series, which was his first big break in the industry. Before his “drip period” (1946–1950), Jackson Pollock experimented with a wide range of genres, including abstract expressionism, realism, and expressionism.

Findings show that a hot streak isn’t just a matter of exploration or exploitation alone. For artists, filmmakers, and scientists, hot streaks following the exact exploration-exploitation sequence were 20.5 percent, 13.8 percent, and 19.2 percent more frequent than after a random point in a given career, respectively. An increase in exploration and exploitation alone was not responsible.

If you only do one, you’re missing out on the entire effect, according to Chown. There must be a combination of exploration and exploitation: trying out different things, learning about many fields and methodologies before settling in to create a substantial body of work with significant effect.

As Wang puts it, “Our work shows that people experiment and likely gain new skills from work in different subfields, and then help find the best one to exploit.”

People are acquiring knowledge and experience from multiple sources before making a final decision, according to a study that indicated that exploitation hotspots don’t tend to be the most recent ones discovered.

The findings of this study have ramifications for the ways in which people and organizations might create their own hot streaks.

Creative people now know that the exploration-exploitation sequence is the most likely to lead to a hot streak in any effort. A powerful combination of creative activities, Wang claims. This might entail, for example, a scientist looking at several different areas of biology before picking on one that he or she is more passionate about.

Hot-streak insights can also be used by organizations. This can be done by analyzing the career histories of academics and researchers in order to identify those who are likely to have a short-term hot streak, and then providing them with funds or other tools to help them achieve this.

In the same way, businesses can use this knowledge of hot-streak dynamics to make decisions on intellectual property management.

For example, “a firm could look at how concentrated their patents in certain areas to understand their pattern of exploration and exploitation of innovation,” says Chown, and then make adjustments to strive toward a hot streak. In the same way, a pharmaceutical or biotech company may evaluate how it operates within and across products for different therapeutic areas—cancer, diabetes, etc.—as it attempts to explore and exploit.

Additionally, the findings may have consequences for how firms arrange exploration and exploitation, and how these phases are incorporated into business strategy.

Research and development teams may be separated from product development teams in some companies, for example. As a result, each team member could only focus on one or the other rather than both, putting them at risk of not having their own “exploration and exploitation sequence.”

Many questions remain unanswered by researchers in the field of hot streaks. There is a lot of value in finding out why people are interested in certain topics, Chown says.

Similar to why an individual moves from exploration to exploitation, we don’t yet know why. It’s still not clear how much exploration is necessary before moving into exploitation, adds Wang.

“We now finally know that hot streaks don’t just happen by magic,” Wang explains. Finding the correct space to exploit for higher success” is a person’s exploration of various locations.