Tal Schwartz is a serial entrepreneur, quantitative researcher, technology innovator, and former Professor of Finance. Over the past two decades, Dr. Schwartz founded and led several organizations including Clicktale, Expand Beyond, and the Technion's Bronica Entrepreneurship and Innovation Center.

As the CEO and co-founder of Clicktale, Dr. Schwartz pioneered the field of Experience Analytics. He grew Clicktale to millions in revenues, hundreds of employees, and raised over $100m from investors. Dr. Schwartz invented products that combined machine learning and psychological research to automatically answer business-critical questions, thereby enabling the “Experience Era” at the world’s leading companies.

Earlier in his career, Dr. Schwartz was a quantitative researcher at some of the most well-known hedge funds in the world including Citadel Investment Group, Mesirow Financial, and Clarion Capital. At these and other financial institutions, Dr. Schwartz developed and implemented trading strategies for equities and fixed income securities, modeling the term structure of interest rates, equity valuations, and portfolio optimization.

As an academic, Dr. Schwartz was a Professor, Visiting Scientist and lecturer at leading universities including the California Institute of Technology, the Technion - Israel Institute of Technology, DePaul University in Chicago, and Tel-Aviv University. He presented his research at industry conferences and published in leading finance journals.

Today, Dr. Schwartz is combining his passion for finance and experience as an entrepreneur, quantitative researcher, and machine learning innovator to launch “Autonomous Financial AI” that aims to disrupt existing investment paradigms.

Interview questions:

Datagran: You are a Doctor in Finance and passionate about AI and ML. How has AI been used in finance?

Tal S: AI's usage has expanded over the years. The first users were hedge funds. Back in the ‘90s, Jim Simons of Renaissance started the first one that was AI-centric and since then many followed. You can read about his adventure here: https://www.amazon.com/Man-Who-Solved-Market-Revolution/dp/073521798X

Datagran: Can it help investors make excess returns or beat the market? How?

Tal S: Well, if you are rich enough to invest in hedge funds that utilize AI, that's one way to do it, but you have to be an accredited investor.

I am also building an offering for anyone that can use ML to run a portfolio of assets and re-balances every month using only ML. It's available through one RIA (RIA - Registered Investment Advisor) right now, but others will offer it as well.

Datagran: I remember an interview with Jim Simons, in very very very basic terms; he said: "What we do is find patterns and check their significance". Do you still think we can look to the market from that perspective? Do you think this perspective is still valid?

Tal S: Hi Necati, yes that still works because humans are still a large proportion of investors. And humans have many built-in biases and mistakes that we make. There is a whole field of economics, called Behavioral Economics which tries to quantify this, and there are 2 Noble Laureates - Bob Shiller and Dick Thaler, who I know personally. They showed that even if you know that biases exist, we still fall for them. That's why an ML approach works well buying when people are fearful, and selling when people are greedy. There is alpha (excessive return) available if you build your models in the right way and know what patterns are repeatable and exploitable

Datagran: Are you completely focused on technical analysis as an input or do you prefer to use fundamental analysis too? A mix?

Tal S: In my fund, I use a combination of two strategies: pure AI that you can call "technical" because it is doing pattern matching and AI-assisted which tries to exploit behavioral arbitrage opportunities.

Datagran: You founded a company called Clicktale. Raised 100+ million dollars and exited the company. How was AI used in Clicktale? What problem were you trying to solve?

Tal S: In Clicktale we were trying to help businesses understand and optimize their customers’ online experience. The problem we needed to solve is that we were capturing everything users were doing online, millions of users, and each one had lots of micro-behaviors like click, scrolls, mouse moves, and underling HTML for example...how do you make sense of all this data? Well, we used ML to cluster users into different groups, we found the paths in the site and on the page that worked best for each group, and where the friction points were. The result was an actionable recommendation on how websites and apps could make changes that would improve the user experience and conversion rates.

Datagran: How did you go about solving these problems? Was there a data science team?

Tal S: Yes, we had a dedicated Data Science team to do research and dedicated developers to build a customized solution. We were SaaS, so everything we built was available to all customers.

Datagran: How do you think ML will affect marketing? What is the opportunity there?

Tal S: It's a huge opportunity. Marketers still use their gut too much and make lots of mistakes. ML can help guide marketers in many areas from optimizing websites to writing content to running campaigns. We are still in the early stages of change and most are ahead of us.

Datagran: There is a huge challenge there I think because usually marketing people are not technical… Growth Hackers are but how do you think the gap can be closed?

Tal S: The best companies understand this is the future and invest heavily in integrating ML with marketing. I think what Datagran is doing will certainly help. 

Datagran: Do yo think interpretation is important for marketers? Or, are marketers OK when they see the results (even though they do not understand what is going on)?

Tal S: We need to make ML super simple for everyone to use and benefit from.

Datagran: And, also, the same applies to markets? is that OK to not be able to interpret the results/patterns when you manage a hedge fund?

Tal S: Interpretation is important to build trust in the algorithms because without trust you won't get adoption. In the market, it is a bit different, because we can run 10,000 trades and see that we have statistical significance… something a marketer has a hard time doing. I prefer patterns that I can explain and I understand, but sometimes I trust the machine algorithm even if I don't understand the "why".

Datagran: Tal thanks. To wrap up the AMA. Where do you think the future of AI is headed? How will it impact our lives?

Tal S: We are at the beginning of the journey. AI will affect every area of our lives in the most profound ways. The impact will be exponential as new hardware and software are released. This will be the greatest change we will see in our lifetime, even bigger than the Internet. We are already seeing superhuman AI for specific tasks like games, diagnostics, etc, and over time we will see many more applications, eventually taking over many tasks that humans do today.

Datagran: Right, I think many people started getting it after they saw demos running on GPT-3.

Tal S: To me, this future is an exciting and inevitable part of humanity’s progress.

Datagran: The conversation could take another path but do you think that will be our natural evolution? Human to AI?

Tal S: Yes, eventually merging the two. For now, it's an extension of us via a smartphone, later as an implant or contact lense. We are already a type of cyborg and that integration will become tighter over time. The more AI we can harness in real-time, the better/smarter more effective humans we will become. We will evolve at the speed of Moors Law, much much faster than biological evolution.

Datagran: Amazing Tal. Well, thanks so much for taking the time to do this. Hope everyone enjoyed the conversation and we are more than happy to have you on this community.

If you are interested in the topic, join our Slack channel here.