How does artificial intelligence and machine learning work and what are some examples of how individual investors can use AI in their investing.
In this episode you’ll learn:
- What is artificial intelligence, machine learning and deep learning.
- How is AI being used by different industries.
- How are AI models built with supervised and unsupervised learning.
- What are the components of a quantitative trading model and why it is insufficient to have an AI based stock ranking service.
- What are examples of AI based investment services and AI ETFs available to individuals
- Why using AI to make investment decisions is so difficult.
Show Notes
Machine Learning in 8 Minutes – Farhad Mahlik – Medium
End To End Guide For Machine Learning Project – Farhad Mahlik – Medium
Neural network racing cars around a track – Gigante – YouTube
Some Investors Had Hunch Yields Were About to Fall – Wall Street Journal
Inside the Black Box: A Simple Guide To Quantitative and High Frequency Trading by Rishi K. Narang
EquBot – Artificial Intelligence ETFs
Episode Sponsors
Become a Better Investor With Our Investing Checklist
Become a Better Investor With Our Investing Checklist
Master successful investing with our Checklist and get expert weekly insights to help you build your wealth with confidence.
Episode Summary
Artificial intelligence has become influential in almost every aspect of life; work, home, travel, and even investing have seen the growing use of AI to aid human processes. In this episode, David explores how AI is trained to process and learn information and in what ways it can be helpful to investors and the ways in which it becomes a detriment as well. Listen to the entire episode for an insightful look at the complexities of AI in the investing world.
Understanding how artificial intelligence works
Artificial intelligence is patterned after the human brain. To train AI to process information, large amounts of data are fed the system in increasingly complicated patterns. AI learns the data by recognizing patterns and building upon those patterns until it can effectively forecast what will happen. Some examples of AI include risk management, finance, identifying fraud, retail, tracking credit, industry and organization connectedness, social media, and marketing.
David explains that there are two types of artificial intelligence. One is rules-based learning, which is heavily influenced by human intervention. You put in a rule, and there is one, fixed outcome for that rule. The result is predictable. The second type is called deep learning, which uses artificial neural systems to develop complex algorithms. Patterned after the human brain, deep learning is highly complex. Deep learning AI draws conclusions based upon the patterns, complexities, and connections that it processes—at a much higher speed and capacity than the human brain.
The complexity of AI training and the questions it raises
The complexity of AI is founded in how it is trained to think. First, the system needs data—but not just any data. It needs refined, quality data for quality output. The AI system then learns the data and forms patterns with it—connections and similarities. Then the system learns to make decisions based upon the patterns it picks up on. With supervised learning, there is an expected output that is generated by humans, and the AI system is taught how to get from point A to point B. In unsupervised learning, however, is where input is given, but the output is not defined as it is in supervised learning. One example that David cites is that of learning a language. If you needed to learn a language without even knowing the basics, you would start by gathering as much information as possible and drawing conclusions from patterns in that data. That’s how unsupervised learning works.
The quality of the AI performance is largely due to the quality of the data. Is it unbiased? Is there enough data? Training AI takes an exuberant amount of time, and it only becomes more complex as the system grows in understanding. One of the issues with unsupervised deep learning is that because of all the hidden layers of pattern-building and algorithms that are created in the learning process, it cannot relay how it comes to the conclusions it does. Humans cannot go back and track how it made a decision.
The role of artificial intelligence in investing
One of the questions that arises from the pattern-building neurologic framework of AI is whether or not it can be effectively used for investing. David points out that investing isn’t a linear framework of patterns and connections. It can change sporadically, with no prior warning. When small changes in input can drastically change the outcome of an AI analysis, can they be trusted with processing the complexity of the investment world? Predicting the future is difficult—for humans and for AI. Listen to the episode for examples of failure in predicting interest rates—both from human and AI analyses.
There are ways that AI is currently being used to help investors decide what, when, and where to invest. There is the alpha model, which ranks stocks depending on how high they are likely to go up depending on price inputs, trend behaviors, and technical aspects. The fundamental criteria used by the AI for evaluation of a stock is based upon the quality of the company, the historical growth of the stock, etc. AI can also take into account the transaction cost of an investment and execute the investment.
David shares his personal experience using an AI alpha model system. The catch he found was that he had to build a portfolio based upon the suggestions of the AI system—which was updated daily. It took a great amount of work on his part to keep up.
Approaching AI as an individual investor and not as a firm
Is artificial intelligence useful to the individual investor? David shares some ways he believes the individual investor could experiment with AI, but he is skeptical of its overall effectiveness when it comes to personal investing. Some AI platforms make suggestions as to what portfolios to choose, however, which can be helpful. Another way to experiment is to invest in an AI ETF, such as EquBot. They have multiple AI platforms that combine fundamental and quantitative analysis for optimum performance—and their ETFs are powered by their AI.
For the individual investor, however, using AI is time-consuming. You have to stay on top of the ever-adapting AI methods and conclusions and be willing to change your model based upon the daily updates of the AI system. You have to constantly feed it new data so it continues to improve. David suggests changing your strategy from focusing on getting better at predicting and to focus instead on getting out of harm’s way. Diversify your portfolio. Understand and keep up with what is happening in the economy. Make educated decisions. But don’t try to constantly predict the future.
Episode Chronology
- [0:20] The two types of AI: rules-based & deep learning.
- [3:35] Some examples of artificial intelligence.
- [4:59] How machine learning systems work.
- [6:10] Supervised learning vs. unsupervised learning.
- [9:10] Training deep learning AI models.
- [12:46] Complications with how deep learning models come up with their answers.
- [14:31] The role of AI in predicting interest rates.
- [17:16] Understanding the different factors and components of AI models when it comes to investing.
- [19:43] David’s personal experience with an AI-based investing service.
- [23:29] The Equbot and investing in ETFs.
- [25:46] The difficulties of using AI as an individual investor.
Transcript
Welcome to Money For The Rest Of Us. This is a personal finance show on money, how it works, how to invest it, and how to live without worrying about it. I’m your host, David Stein. Today is episode 256. It’s titled, “Will Artificial Intelligence Change Investing?”
Artificial intelligence “AI.” The term gets thrown out a lot in many different domains—how it’s going to change the world, how it’s changing industry, potentially how it’s changing investing. There are stock research services that you can subscribe to that help you pick stocks based on artificial intelligence. There’s ETFs that do the same thing. We’ll take a look at both in this episode.
What is AI?
But first, what is artificial intelligence? I like this definition by Radu Raicea. It’s from Free Code Camp. He writes, “Artificial intelligence is the replication of human intelligence in computers.” One of the things humans are good at is recognizing complex patterns and using those patterns to make decisions. Artificial intelligence does that. One way it does that is with machine learning which Raicea describes as, “It’s the ability of a machine to learn using large data sets instead of hard-coded rules.” These data sets are used to train computers. This machine learning’s about predictive algorithms. The ability to forecast what’s going to happen based on data.
Now there’s two types of artificial intelligence. One is a rules-based system. You have an input, and then it produces an output, and humans are very involved in designing it, and these systems are easier to understand because they show, “How do you get from point A to point B?” And that was some of the original artificial intelligence.
The second type is much more revolutionary and, frankly, much more complicated. It’s called deep learning. It involves neural networks, artificial neural networks, essentially replicating aspects of our brain. We have multiple nodes, and they’re connected hidden layers. These systems aren’t necessarily able to explain how they got the answer or the relationship. They’re extremely complex, and they’re based on math, very much a black box.
I recently finished a book by David Weinberger called “Everyday Chaos.” He wrote, “Deep learning’s algorithms work because they capture, better than any human can, the complexity, fluidity, and even beauty of a universe in which everything affects everything else all at once. The scale and connectedness of machine learning results in their complexity. The connections among the huge number of pieces can sometimes lead to chains of events that end up widely far from where they started. Tiny differences can cause these systems to take unexpectedly sharp turns. We don’t use these technologies because they’re huge, connected, and complex. We use them because they work.”
What is AI used for?
Now, what are some examples of AI? Well, risk management. Companies use artificial intelligence to predict credit risk when making a loan to decide whether the counterparty will default or not. Now there’s definitely some rules in place for how they can go about doing that. Finance, identifying fraud, trading algorithms, something we’re going to talk about in today’s episode. Using AI to figure out which stocks to buy. Retail was one of the earliest users of artificial intelligence as they fed it these systems, these machine learning systems, a huge amount of data to figure out who’s going to buy what, and what are the connections. A consumer might buy this particular product, and they find that when they buy that, they often buy a different product. It might not have anything to do with it. Not necessarily a logical connection, but that’s what these AI systems can figure out these complex patterns. Technology using email filtering, healthcare, the ability of these deep learning systems to figure out whether somebody is likely to get a particular disease or not. Automobile industry, self-driving cars is very much founded in artificial intelligence.
As a Money For the Rest of Us Plus member, you are able to listen to the podcast in an ad-free format and have access to the written transcript for each week’s episode. For listeners with hearing or other impairments that would like access to transcripts please send an email to jd@moneyfortherestofus.com Learn More About Plus Membership »
