Managing Director Paul Quinsee, the global head of equities at J.P. Morgan Asset Management (JPMAM), believed that he knew what was needed from data analysts to turn them into the most successful members of his team.

Similar to the talent scouts found in Moneyball: The Art of Winning an Unfair Game, a bestselling book by Michael Lewis on the way data science changed baseball, Quinsee was responsible for overseeing fundamental research analysts play in the office for nearly 40 years.

When the head of the investment platform and former head of equity data science and trading, Kristian West, returned to him, Quinsee was surprised to find results of years of research notes compiled by predictive analysis. Though he may have been less hardened than the scouts in the novel when hearing what the data believed was necessary to win, the characteristics were not completely intuitive.

The machine learning model of West not only “read” stored notes compiled over many years, it included personality tests of the analysts, their views, how often they consulted with companies, their complexity, and the types of models utilized.

It was discovered by West’s team that the greatest forecasters wrote the shortest research notes. They consulted companies more often, with simpler models, and implemented extreme language and tone in their notes. Unlike the stereotype that claimed a common correlation between former Wall Street analysts and football players, the superforecasters did not participate in sports.


Despite the fact that many quants including but not limited to Renaissance Technologies and D.E. Shaw have significantly embraced advanced computing and data techniques, a surprising number of traditional name-brand firms are still in the beginning stages of crafting proprietary tools for portfolio managers that make decisions on fundamental factors like a new product’s financial potential, corporate management’s visionary qualities, or the organization’s ability to exist in a pandemic.

Furthermore, managers aren’t always up-to-date with the assessment and work needed to implement the capabilities artificial intelligence has to offer for data. In one particular example, the 70-person Investment Science Group of Wellington Management has focused their efforts on applying analytics to investment theories and investors which includes mitigating and uncovering the downside of their biases.

However, to be fair, it is still very expensive to equip any company with data science platforms and tools less artificial intelligence. While the industry is in competition with other technology companies for talented software developers, surveys demonstrate that the majority of asset managers believe it is critical for risk management and performance. Earlier in the year, Accenture, a consulting firm, discovered that asset managers possessing centralized and industrialized artificial intelligence implementation across their platforms were achieving substantial bumps in returns that were risk-adjusted.


The data science team at JPMAM was established more than 3 years ago. However, West redirected them to focus efforts on 4 key projects.

  • Governance, Environmental, and Social Information (unfortunately, still little data in this region)
  • Natural language and linguistics processing to create predictive insights
  • Provision of aggregated business and retail data from the bank J.P. Morgan
  • Application of data science to investments and research

This information can be observed from a dashboard on the technology’s platform Spectrum as well as the creation of cohorts and models like the one that may examine millennial spending habits.


A staggering $400 million was spent by JPMAM on this project despite the fact that the figure includes development for derivatives, equity trading, and broker relationship management teams.

In many ways, their machine-learning fund, known as the U.S. Applied Data Science Value fund, provides an overview of the framework. Picture a fund that utilizes a model to mimic what an investor might do. This model is responsible for “reading” external and internal research, observing trading data, and subsequently providing stock recommendations.

The underlying manner in which such applications adapt and learn when processing data is inexplicable to the average individual. Many of the investors, however, are still required to explain the decision-making behind investments to boards, donors, and trustees. JPMAM designed its machine-learning model to articulate at every stage the reason behind its decisions to make the explanation process far easier for regulators and customers.

It is critical to mention the way the company separated the product from the framework that powers it. Much of the work that West did in 2018 overhauled trading, but the applications were constructed in a vertically integrated manner such that developers may create an app that increased returns while another group would look for data sources that served to aid profitable ideas for fixed income. Sadly, both pieces of technology remained in their own worlds.

Starting in 2021, Mike Camacho was appointed as the new CEO of J.P. Morgan Wealth Management Solutions. He had West completely overhaul the infrastructure to include research, development, portfolio construction, and trading. Under a unified global order management system and machine-learning model, more than half of all trades from this year were automated.