How To Become A Financial Data Scientist (Or A Data Scientist In Any Domain)

Written by Matt Dancho on May 23, 2019



Becoming a data scientist in Finance can be a lofty challenge… unless you know how to streamline the path. Instead of asking “what skills do I need?”, we need to ask a different question. We need to first learn to see what the organization values. Then apply the 80/20 Rule to get there.

If you are a financial professional seeking to learn data science, then this is what you’ve been waiting for.

The opportunity to understand what organizations in Finance (or any Domain) value from the data science practice. If you understand what they value, you then know what skills to learn to streamline your path from where you are now to being a productive member of a Financial organization (or any organization).

My Talk "How To Become A Financial Data Scientist" - YouTube

What I’m covering here is focused on finance, but the same strategies can be applied broadly to domains as wide-reaching as Marketing, Research & Development, Medicine, and more.

I’ll assume that if you are reading this, then you have some knowledge on Portfolio Theory and Risk Management, which are required knowledge for the finance domain. What I will show you is how you can take this domain knowledge and extend it to value generating activities for the business using data science tools.

But, first, the cold hard reality…

Learning Skills That You Don’t Use Is A Waste Of Time

Most financial people that want to break into Data Science make major mistakes that cost them a career in data science. It starts with their first step.

Modern Data Scientist

"Modern Data Scientist Infographic" - Everything That Is Wrong With A Learning Strategy

They see the “Modern Data Scientist” infographic, and immediately feel overwhelmed. Worse, they begin down the path of learning everything. Learning everything is unproductive.

There is no strategy to this graphic. No foundation, no purpose, no intent. Just a smattering of skills that supposedly create a data scientist.

Worse, students starting out believe that this is the ultimate goal - A Unicorn - A mythical creature.

I’m hear to tell you that (fortunately) this myth is not a reality.

We All Start At The Same Spot - Zero

When we start out, we are the most vulnerable to making missteps. I personally remember feeling overwhelmed and directionless. It’s at this moment that we are easily influenced to take the path of learning everything (and anything). Learning everything is a costly strategy issue, but an easy one to make. With so many people saying different things, yet none of the “experts” are stepping up to give you mentorship.

It’s scary - being alone on this journey. Add to it that every misstep costs us time, and it’s easy to see why many data scientists struggle (and many don’t succeed). Time is our enemy. The longer we take on this journey, the more competitive it gets and the more likely we are to fail.

We Grow By Building Skills That Add Value

One thing I learned along my own journey was how to sell my value. It was an incredibly important lesson that I learned through my experience consulting. When I begin an engagement, I never go in saying I know something. Rather, I ask how I can help and listen for opportunity. Skills alone don’t sell. It’s solutions, results, and value that sells.

Marketing 101 - Skills Don’t Sell; Solutions Sell, Results Sell, Value Sells

We need to change our beliefs. Most of us feel that to get a job in data science, you need to learn data science inside and out. Machine Learning, Deep Learning, Neural Networks, Graph Theory - the list goes on.

This is not correct. Further, this strategy will cost you time, and likely won’t result in your desired outcome - to begin a career as a Data Scientist in Finance.

When we adopt a new mindset, one of value over skills, we begin seeking skills that add value incrementally to the larger goal. This is a step in the right direction, one that many data scientists miss.

Remember the “Modern Data Scientist” infographic? Don’t learn all of these skills.

Stop Modern Data Scientist

Bad Strategy: Learn Skills That Every Data Scientist "Should" Have

Rather, learn how to create value by incrementally adding skills to your toolkit. Focus on addressing what financial organizations want. Then learn tools that will deliver it.

Create Value

Good Strategy: Learn How To Create Value

The next logical questions is…

What Do Financial Organizations Value?

To be effective in a Financial Organization, you need to generate value for the organization. You do this by:

  1. Reducing Cost
  2. Increasing Revenue
  3. Maximizing Profit

Solving problems that address KPI’s (Key Performance Indicators) is a great place to start. Anything to do with customers, quality, service, performance, and so forth.

How Do You (the Data Scientist) Generate Value?

By taking Applications into Production.

Every day we make decisions based on intuition. When we use data to improve decision-making, value is generated for the organization by reducing costs, increasing revenue, and/or maximizing profit.

Applications that embed data science can save organizations $15M+! Don’t believe me? Here’s an expert application that I built that can easily result in multi-million-dollar-per-year savings.

A Real-World Example - Assisting Asset Manager’s Tactical Investment Allocation

The only way to make a difference is by understanding the people you seek to help.

Let’s walk through a short example of this - A Stock Portfolio Optimization Application (DEMO HERE) that I demonstrated at the R/Finance 2019 Conference.

Stock Optimization Application - YouTube Video

App Demo Here

Problem Statement

Asset Managers select stocks based on their knowledge of the company, market, and intuition of what the future holds. However, allocating an investment among the basket is a time-consuming problem that is costly if the Asset Manager over-weights a risky stock. A bad bet can result in lost Clients, costing the the organization millions in fees that would have otherwise been collected.

Solution Statement

We can use data-driven analysis to optimize the allocation of investments among the basket of stocks. Modern portfolio theory (Capital Asset Pricing Model) suggests that using the Sharpe Ratio (a metric of reward-to-risk) can reduce the riskiness of a portfolio while preserving returns.

Implementation

We can automate the portfolio allocation process by randomly calculating portfolios, calculating the Sharpe Ratio, and returning the tactical allocation strategy of the best portfolio. This allows the Asset Manager to focus on his or her job of picking stocks, while the investment allocation decision becomes automated using modern portfolio theory.

The web application available for demo here and described in the YouTube video does just that - helps an Asset Manager make better investment decisions that will consistently improve financial performance and thus retain clients.

Stock Optimization Application

Tactical Asset Allocation - Portfolio Weights Optimized Using Sharpe Ratio - App Improves Decisions

What Do You Need To Learn?

The road to go from where you are now to a data scientist in a Financial Organization can be accomplished in weeks, not years. But, you need to have a plan to strategically learn the right skills.

This where I can help. I’ve been there. I’m willing to step up. I’m willing to guide. But, make no mistake, it will take serious commitment on your part.

Here’s what I recommend that you learn and why. It’s called the “Data Science Workflow”.

The Data Science Workflow

Building applications like the Stock Portfolio Optimization App is what we call “Production”. This is the end stage of your efforts. But what you don’t see is the hard work that you (the data scientist) put in beforehand. That hard work is actually a process called the “Data Science Workflow”, and it looks something like this.

Data Science Workflow

The Data Science Workflow

The “Data Science Workflow” is the series of tasks required to go from business problem to business value. It’s a time consuming process that requires:

  • Business Problem Understanding - Working with process stakeholders (e.g. Asset Managers) to understand their unique business challenges

  • Communication of Business Value - ROI analysis of any solutions, communication with executive leadership to convey the value

In the middle is a complex series of actions that involve Data Science Tools (everything involved in going from machine learning to reporting and deploying web applications).

Data Science Tools Exposed

Here is the same graphic with a set of tools that can be used as part of the “Data Science Workflow”. The tools integrate throughout the problem solving and solution building process. This is how we add value to the organization.

Data Science Workflow - With Tools Exposed

The Data Science Workflow - With Tools Exposed (R-Track)

Value Comes From Tool Integration

Value comes from using a specific set of tools that incrementally add value along the “Data Science Workflow”. This allows us start with a business problem and end with a predictive web application that delivers business value that is tracked with reports and measured for ROI (Return on Investment).

Let’s break this down:

  • Specific Set of Tools: Focusing on this specific set of tools cuts the time to learn data science dramatically. This is the 80/20 Rule in Full Effect!

  • Incrementally Add Value: The tools combine into an integrated approach to solving problems. Therefore, we can’t just read a book on each tool independently. We need to learn the tools together to harness their power.

  • Delivers Business Value: The Application is the Value-Generator. Without it, the data science team adds little value to the organization.

  • Measured for ROI: Any business improvement should be tracked, reported on, and measured for return on investment. Changes in KPI’s converted to financial value.

This is why I teach project-based learning using the set of tools and processes that I use every day. I’ve developed a learning program that incorporates each of the tools to get you effective and fast. The best part, I’ll help you learn it through Business Science University.

Here is just one example of a student’s success story. Michael, a student of mine, is working through the first of two courses learning Data Science For Business. He’s 3-Weeks into my program and has recieved approval of his forecasting project from the Executive Team.

Course Feedback

Business Science University helps you deliver Business Results

The best part is that Michael’s story is not unique. We’ve helped hundreds of students add value to their organizations through Business Science University.

How Do You Become A Financial Data Scientist?

If you’ve read this article and decided that you’re more excited than ever to start this journey, then Business Science is here to help. I will commit to you if you are ready to commit to learning. If so, the tools can be learned in weeks, not years. Here’s the plan that integrates the 80/20 Rule, Project-Based Learning, Business Value Generation.

Learning Plan - Business Science Courses

Learning Plan - Business Science Courses

Project-Based Data Science For Business Courses

The 101, 102 and 201 Courses teach you each of the major skills and tools required to solve business problems with data science and machine learning. These are Project-Based Courses. The tools integrate while you solve the business problems presented in the courses. Through the process of solving these business problems, you learn:

  • Machine Learning - Supervised Classification, Supervised Regression, Unsupervised Clustering, Dimensionality Reduction, Local Interpretable Model Explanation - H2O Automatic Machine Learning, parsnip (XGBoost, SVM, Random Forest, GLM), K-Means, UMAP, recipes, lime

  • Data Visualization - Interactive and Static - ggplot2 and plotly

  • Reporting - Interactive and PDF (3 Reports each) - rmarkdown

  • Programming and Iteration - Build functions and then iterate with purrr mapping functions

  • Data Manipulation and Cleaning - Use dplyr and tidyr to wrangle data

  • Time Series, Text, and Categorical Data - Work with lubridate, stringr, and forcats

  • Data Science in Production - Implement a complex series of UI and reactive programming using shiny and flexdashboard


Unlock the 101 + 102 + 201 Bundle

Conclusion

Learning Data Science in Finance can be a challenge. With infographics and “experts” leading us in a million different directions, it becomes a challenge to find a single path to becoming a financial data scientist. This is the problem I set out to solve two years ago when I began teaching Data Science Education.

Fast-forward to today. I’ve spent a lot of time designing and developing a system that works fast. Life is short, we need to go from where we are now to where we want to be as quickly as possible. Otherwise we lose out on the opportunity for a better career and a better life.

To help you get there fast, I have created this system that will work for you. It requires commitment. But, it will get you results in weeks, not years.

Start Learning Today!

Learning Plan - Business Science Courses

Learning Plan - Business Science Courses

I look forward to helping you learn data science for business. I will do everything in my power to help you succeed.

-Matt Dancho, Founder of Business Science and Data Science Instructor at Business Science University