A data scientist is not a quant.

While a masters in DataScience makes sense, I am not sure about a doctorate. Caltech has a doctorate program in Mathematical Computing that my son is attending. Other top universities offer doctorates in CS or Statistics where the work is interdisciplinary.

Data Science is very wide field for the different specialist’ qualification in the same grade as Applied Math.

Given the different views on this topic I am curious if you (collective you) believe that data science should develop into distinct disciplines or consist of a cross disciplinary education? It seems there is disagreement on what data science is today. Why can't mathematicians, statisticians, physicists and other disciplines sit comfortably under the umbrella of data science? Herbert Simon, Frank Knight, Dan Kahneman, Leonard Savage and many others did not focus on one discipline. What would you call their training? If data is the material of study then the science of understanding the is data relative to the subject. Would this be true or is there some other logical way to define the emergence of new ways of thinking about data.

Assuming the degree is tailored toward application in finance, I am curious about the depth of the study and how applicable it can be. Hope it includes AI as part of the study.

Agreed Yongshan as well as the cognitive sciences to understand how bias creeps into data analysis including ethical decisions when using data to sway opinion.

Data Science Encompasses Mathematics and embedded probability statistics and of course data mining which provides the algorithmic functionality for precision of duplication from the black box to production, but many verticals that intersect disciplines are still theoretical in nature and somewhat undeveloped. Having said this I agree with your collective in that we have to capture the full potential and practical use of Data Science. Let's explore these possibilities together .

I agree that it is imperative to know data science and have expert control of the numbers but as the market always shows it's not the math that makes money. It's a undefined combination of factors that simply aren't a formula. Besides the quant funds returns have been average this year. Maybe they find some bits and pieces that humans can't but are NNs and such tools a substitute or only a help? Good luck to all but as always, markets adapt and chew up and spit out those that think with the majority.

Much of the advancements in economic growth and innovation is directly attributable to analytical and scientific thought while much of the conflict and limitations we pose on society is due in part to human behavior. Igor points are well taken some applications are may serve only to facilitate human decision making while Angela's optimistic views may find new advances and solutions to complex problems that seem intractable today. There is room for both without value judgments of right or wrong.

I have proposed a cognitive risk framework that includes the human element needed in a new world environment that will undoubtedly include more machine operating in the benefit of our lives and contributing to more threats. I, like Angela, hope that we learn to advance together.

A.I. and machine learning technology are tools, data is the material on which to apply these tools and "data science" (who is not a data scientist these days?) provides the manual on how to apply the tools to the material. What is usually missing in these discussions are the theories that are tested by data scientists when applying their tools. Like you wouldn't build a house without a plan you shouldn't apply data science to a subject matter without a theory first. Here is where academics need to come in, but often don't. Without proper theories to subject matters the quick wins from data science will very soon be reaped and the hype be gone.

Wolfgang, your observations are spot on and I hope that my theories are the exception to your rule that academics don't apply or provide theories to data science. My theories are included in the Cognitive Risk Framework for Cybersecurity and Enterprise Risk Management but are applicable more broadly. My book ##CognitiveHack is an example of the application of my theories. My theories are not new but are based on the work of Frank Knight, Herbert Simon, Dan Kahneman, Paul Slovic and many Nobel Laureates in Economics and Decision Science. Your points are well taken and that is exactly why I developed the Cognitive Risk Framework and why it connects COSO ERM, ISO 31000, NIST and other frameworks. Risk frameworks are devoid of scientific rigor or the theories upon which risk, data science, and or scientific exploration is based. Last point, theories have value only when they are applied & useful in real life.

Igor and James' points are valid. Despite all the data coming in on a low latency pipes, and millisecond execution, the analytics ultimately rely upon understanding human bias others programmed in their strategy. Following herd reaction is not enough. To effectively mitigate the risk, especially in today's low volatility environment, it is necessary to have AI capability in your backtesting everyday. Our Quant has 2 PhD's and an engineering degree and still puts in 13 hour days, in his 60s. However, our slippage remains under 10%, and realized profits consistently above 90% positive trading .

In the trenches, is where the rubber meets the road. A great deal of what I read here is theoretical to a great degree, making me wonder if they are tested with hundreds of millions of at least notional capital traded. Strategies, and theories without market record are meaningless . Those who know do, rest teach.