Quantitative strategy? As in, an employment of investing tactics to improve equity portfolio management?

That is what you’re likely going to find on the first page of Google if you search ‘quantitative strategy.’ At least, as of August 2017; I don’t know when/where/how you’re reading this and whether or not any terms I will be saying will be relevant to you!

But, what is relevant now is data. Huge, ever-amassing, ambiguous big data.

Quantitative investment strategies are employed to help investment portfolios beat the market. Using a variety of historical data, predictive statistical models spit out investment opportunities that are more likely to provide a higher return, or at least not deviate from the return expected by the given level of risk; conversely, the models can provide indications on securities that will not be expected to perform as previously indicated. More simply, a statistical model instructs portfolios’ buy and sell activities in order to make more $$$ instead of a manager using solely gut hunches.

This isn’t an investment blog, though. This is for data science and reconciling what we know as humans in our environments with what data knows mathematically.

What I am looking to flesh out with quantitative strategy is the use of data science concepts to influence strategic decision making. Expanding past investment opportunities, I am interested in driving knowledge from datasets and influencing decisions in ways that are unbiased. In the means that quant investors use models to build their portfolios, I want to use data to inform others in order to make the best decisions they can. Without them having to ask. This unsolicited advice can absolutely vary.

Hey, Sales Representative! Here is the best lead available for you right now based on information you don’t care about and will not use during the sales spiel.

Aloha, Chief Executive Officer! I really think you should consider abandoning this segment of your business due to risk indicators that haven’t been clearly identified or discussed within current business management. 

That sounds a little pretentious. I think I can know better about a customer than the sales representative? I think I can make an assertion to sever an entire business division? No, of course not. But, there are so many insights available in data that are just not as noticeable in the day-to-day of the business. By diving into data and exploring and running models, answers can be provided when no questions were posed. Or, maybe, a very ambiguous question was posed, and these insights can drive a direction to obtain that answer.

Consider a sales representative in a Software as a Service (SaaS) business that gets paid commission based on converting a lead to a sale. SaaS customers generate the majority of revenue on a monthly or annual basis. The longer they stay, the more value they provide to the business. There may be a certain segment of customers that stay longer than others with identifiable traits that can be revealed during a lead delivery. Those traits are valuable to the business, but the sales representative could disregard because it won’t help close the deal. Rather than expecting an individual on the front end of a process to be able to identify what happens later on in the customer life, lead generation scoring models can be developed to deliver the highest value leads first or provide a higher commission on those lead conversions. Lead generation scoring models are pretty standard in marketing departments and not exclusive to the data science community. I just think they’re pretty neat and a great example of quantitative strategies steering a developing customer base.

In a bold suggestion to terminate a segment of a business, there could have been a product a segment of customers use in which those customers are departing with the business. Especially in instances where customers have a mix of products, reasons for customer departures can be inconclusive at the time of exit. With the ability to mine and explore data, one could potentially pinpoint weak spots in the product mixes. Maybe a certain sect of customers that use a particular conglomerate of toggles for a similar reason and all depart — potentially, a cheaper substitution came to the market or the outside need for the toggles have been completely eliminated. A data scientist wouldn’t necessarily be able to pinpoint specific reasons from the data, but the correlation could be found. If outside research on the market for that particular toggle set isn’t being executed, then it isn’t likely for this discovery to have been raised by business managers. A suggestion so extreme as “DUMP THE BUSINESS NOW” will probably not be the first brought up, but the data could open up an environment for that conversation when it previously wasn’t even a consideration.

I first heard the term quantitative strategy on a professional visit to a mortgage company out of Columbia, Missouri: Veteran’s United (VU). Starting out a new role in the data science world, I needed some guidance. The friendly folks over at the VU data science department provided some insights on data science within a corporate environment, as well as an introduction to the term quantitative strategy — the official title of their department. I got a glimpse of how far I had to catch up when they showed me their Python notebooks and animated statistical models based on live mortgage data. Even their forecasting in MS Excel had more thought behind it than any other budget I’ve seen in my finance days.

A little on me. My history in no way reflects the traditional data science career route. In fact, my professional background is accounting, finance, and a little bit of information technology (mostly business analytics and project management). Educationally, my undergraduate is in finance and accounting and I am in the last semester of an MBA. I guess that explains my tendency toward the human aspect of data driven decision making; all of those management classes still sit inside my core. There has always been an affinity in me for information — to collect it, process it, share it. Data analytics certainly provides a means for that. So does gossip, but the market for gossip blogs is just too crowded.

I landed a position at a post-startup company to establish a data science concentration within the firm. I am provided time for learning and resources but am expected to determine the tools and means for execution of the quantitative deliverables. It is a great company, I am learning a lot, and I am completely enjoying myself. But I retain more information if I am able to explain it more clearly to others. If an executive asks for a line of best fit in order to make future predictive decisions on some volatile data, then I want to, instead of delivering a trend line in response, be able to clearly explain why this solo analysis may not be in best interests for long term predictions given the low significance value of the line.

Quantitative strategy consists of using quantitative measures to formulate decisions and steer execution. That’s still really ambiguous and broad in what it entails. Scorecards? Dashboards? Statistical models to hem and haw over? Predictive analytics that incorporates directly into the production environment? Is there database management involved? More topics? Well. That’s what this blog is for. I’m gonna learn some data science.