Keeping Up with the Quants

1. Analytical Skills are Essential in a Data-Rich World

If we can’t turn that data into better decision making through quantitative analysis, we are both wasting data and probably creating suboptimal performance.

Data-driven decisions. In today's world, data is abundant, and the ability to analyze it quantitatively is crucial for making informed decisions in business, government, and society. Without analytical skills, organizations risk wasting valuable data and making suboptimal choices. The rise of analytics is evident in various domains, from sports (moneyball) to online gaming and movie recommendations (Netflix).

Beyond transactional data. Organizations accumulate vast amounts of data, and they need to make sense of it to improve internal decision-making. This includes exploring data on human resources transactions to answer questions like employee retirement projections or the relationship between vacation days and performance ratings. Analytics helps summarize data, find meaning, and identify patterns.

Competitive advantage. Internet-based organizations like Google, Facebook, and Amazon use big data from online transactions to create new product offerings and features for customers. Whether seeking better internal decisions or more value for customers, analytics is essential for extracting value from data and gaining a competitive edge.

2. Analytics Classifications: Descriptive, Predictive, and Prescriptive

By analytics, we mean the extensive use of data, statistical and quantitative analysis, explanatory and predictive models, and fact-based management to drive decisions and add value.

Three types of analytics. Analytics can be classified into descriptive, predictive, and prescriptive categories, each serving a distinct purpose. Descriptive analytics involves gathering, organizing, and summarizing data to describe characteristics of what is being studied, historically known as reporting. Predictive analytics uses past data to forecast future outcomes by identifying associations among variables. Prescriptive analytics suggests courses of action through experimental design and optimization.

Qualitative vs. Quantitative. Analytics can also be qualitative or quantitative, depending on the process and data type. Qualitative analysis gathers in-depth understanding of underlying reasons and motivations, using unstructured data from a small number of cases. Quantitative analytics involves systematic empirical investigation using statistical, mathematical, or computational techniques, collecting structured data from a large number of cases.

Various analytical techniques. Several types of analytics serve different purposes, including statistics (collection, analysis, interpretation, and presentation of data), forecasting (estimating future variables), data mining (extracting patterns in large datasets), text mining (deriving patterns from text), optimization (finding optimal solutions), and experimental design (eliciting cause-and-effect relationships). These techniques often overlap in their application.

3. Big Data's Impact and Potential Across Industries

Big data and analytics based on it promise to change virtually every industry and business function over the next decade.

Ubiquitous computing. The rise of big data is enabled by ubiquitous computing and data-gathering devices, with sensors and microprocessors becoming increasingly prevalent. Virtually every mechanical or electronic device can leave a trail describing its performance, location, or state. This data, combined with information from the Internet and other media, creates vast data sources.

Industry transformation. Big data and analytics have the potential to transform virtually every industry and business function over the next decade. Organizations that get started early with big data can gain a significant competitive edge. Manufacturing, consumer marketing, and even self-driving cars are increasingly viewed as big data problems.

Analytical thinking. CEOs like Gary Loveman (Caesars Entertainment), Jeff Bezos (Amazon), and Reid Hoffman (LinkedIn) publicly advocate for analytical thinking and decision-making as a route to organizational success. All organizations in all industries will need to make sense of the onslaught of data, requiring both detailed analysts (quants) and decision-makers who can act on quantitative data and analysis.

4. Framing the Problem: The Foundation of Effective Analysis

A quantitative analysis starts with recognizing a problem or decision and beginning to solve it.

Problem recognition. A quantitative analysis begins with recognizing a problem or decision and framing it effectively. This involves defining the question the analytics will answer and the decision to be made based on the result. Sources for this step include curiosity, job experiences, decision needs, current issues, existing theories, and project offers.

Stakeholder analysis. Identifying and managing stakeholders is crucial for the success of any quantitative analysis project. This involves understanding their needs, assessing their interest and influence, managing their expectations, and providing regular feedback. Stakeholder analysis can identify primary decision-makers and how they are most likely to be persuaded by the results.

Focusing on decisions. Focusing on specific decisions that will be made as a result of the analysis is helpful in the problem-recognition stage. This makes participants realize the purpose of the analysis, identifies key stakeholders, and determines whether the analysis is worth doing. It also helps to identify the type of analytical story to be told.

5. Solving the Problem: Modeling, Data Collection, and Analysis

A model is a purposefully simplified representation of the phenomenon or problem.

Modeling and variable selection. Solving the problem involves deciding what variables are going to be in the model, collecting data that measure those variables, and then actually doing the data analysis. A model is a purposefully simplified representation of the phenomenon or problem, isolating the important, useful, and crucial features that make a difference. Hypotheses are educated guesses about what variables really matter in the model.

Data collection and measurement. The next stage is to collect and measure the selected variables. Measuring a variable is assigning a number to the variable; data is just a collection of these numbers. Data can be structured (easily captured in rows and columns) or unstructured (text, images, audio, video). Secondary data (collected by someone else) can save time, while primary data must be measured by the researcher.

Data analysis techniques. Data analysis entails finding consistent patterns, or the relationships among the variables that are embedded in the data. Various techniques, from basic analyses such as graphs, percentages, and means, to more elaborate statistical methods, can be used. The specific type of model depends on the number of variables, whether description or inference is desired, and the level of measurement available.

6. Communicating Results: Storytelling with Data

How you communicate about analytics is critical to whether anything is actually done with them.

The importance of communication. How you communicate about analytics is critical to whether anything is actually done with them. If a decision-maker doesn’t understand what analyses have been done and what the results mean, he or she won’t be comfortable making a decision based on them. Communicating analytical results in an interesting, attention-getting way is particularly important.

Telling a story. The most successful analysts are those who can "tell a story with data." Good stories have a strong narrative, present findings in understandable terms, and conclude with actions to take and the predicted consequences of those actions. Visual displays of information are particularly well-suited to this type of story.

Modern methods. Modern methods of communicating results include visual analytics, dynamic and interactive displays, and even tangible outputs like 3-D models. Games and simulations can also be used to communicate how variables interact in complex relationships. The goal is to engage the audience and make the results memorable and actionable.

7. Creativity and Quantitative Analysis: A Powerful Combination

The most successful uses of analytics are highly creative…

Synergy of creativity and analytics. Creativity and analytics are not opposites but are often closely related. The most successful uses of analytics are highly creative, and creativity is an important component of successful analytical approaches to problems. Creativity alone, without any data or analytics, usually does not provide enough support for making the best decisions.

Creativity in the analytical process. Creativity is essential in problem recognition and framing, identifying relevant previous findings, selecting variables, and presenting results. However, it is less appropriate in the data analysis step, where adherence to established statistical methods is crucial.

Four stages of creative analytical thinking. The creative process follows four stages: preparation (groundwork), immersion (intense engagement), incubation (internalization), and insight (breakthrough). These stages map onto the six steps of analytical thinking, with preparation corresponding to problem recognition and review of previous findings, immersion corresponding to modeling, data collection, and data analysis, and insight leading to results presentation and action.

8. Cultivating a Quantitative Mindset: Attitudes and Habits

Your usual quantitative attitude forms your quantitative habits.

Quantitative attitude. To become a proficient quantitative analyst, it's essential to cultivate a quantitative attitude, which includes being open to learning about numbers and insisting on a high standard of evidence. This involves overcoming the fear of numbers and feeling comfortable when encountering them.

Quantitative habits. Developing quantitative habits is crucial for becoming a proficient quantitative analyst. These habits include demanding numbers, never trusting numbers, being particularly suspicious of causation arguments, and asking questions.

Quantitative knowledge. To develop quantitative knowledge, it's important to study the basics of statistics and research methods. This can be done through online courses, textbooks, or formal degree programs. The key is to practice quantitative analyses and apply them to real-world problems.

9. The Importance of Questioning and Skepticism in Data Interpretation

He uses statistics like a drunken man uses a lamp post, more for support than illumination.

Never trust numbers blindly. It's crucial to approach data with skepticism and a critical eye. Numbers can be misleading, outdated, or inaccurate, and they can be misinterpreted to advance hidden agendas. Always question the relevance, accuracy, and correct interpretation of numbers.

Causation vs. Correlation. Be particularly suspicious of causation arguments. Just because two variables are correlated does not mean that one causes the other. Consider whether people could have been randomly assigned to conditions for one of the factors. If not, the causal inference is not supported.

Ask probing questions. Develop the habit of asking questions to understand the problem and the process more clearly. This includes questioning the source of data, the representativeness of the sample, the assumptions behind the analysis, and the potential for alternative interpretations.

10. Building Effective Relationships Between Quants and Decision-Makers

Effective quantitative decisions are not about the math; they’re about the relationships.

Mutual accommodation. Effective quantitative decisions require a mutual accommodation between business decision-makers and quantitative analysts. This involves building mutual respect, understanding each other's skills, and speaking the same language. The goal is to make analytical decisions while preserving the role of the executive’s intuition.

Decision engineering. Intel's decision engineering group emphasizes the importance of relationships between analysts and decision-makers. This involves the analyst understanding the business problem, speaking the language of the business person, and engaging skeptics in the process.

Building the model. The next step in the relationship is for decision-makers and quants to collaborate to build the basic model. The key quant person drives these brainstorms to elicit inputs (data elements, sources of data, ideas for detecting and fixing bad data), outputs (what solution slices and dices are most desirable, what display methods are most intuitively satisfying to the intended business users), what are the key variables, and what are the key relationships between variables.

11. The Analytical Responsibilities of Business Decision Makers

It is not my job to have all the answers, but it is my job to ask lots of penetrating, disturbing, and occasionally almost offensive questions as part of the analytic process that leads to insight and refinement.

Learning math and statistics. Business decision-makers have a responsibility to learn something about math and statistics. This includes understanding measures of central tendency, probability, sampling, correlation, regression, experimental design, and visual analytics.

Questioning assumptions. Executives should understand and question the assumptions behind analytical models. This involves determining whether the world has changed in ways that call the model into question.

Pushing back. It's important to push back when you don't understand something. This includes requesting data and analysis, rather than anecdote or opinion. By asking questions and challenging assumptions, executives can ensure that analytical models are relevant, accurate, and well-understood.

Last updated: February 26, 2025