Hasta La Vista… Traders? (The Future Of Artificial Intelligence In The Financial Industry)

One of our favorite movies at Macro Ops is Terminator 2: Judgement Day.

The early 90’s classic had Arnold at his best and changed the sci-fi action genre forever.

People have always had a particular fear of technology and machines, but Terminator really solidified that picture. How many times have you heard the “Skynet” argument when it comes to today’s technology?

Fortunately, I don’t spend much time worrying about robots taking over the world. I’ll leave that to Elon Musk.

But what does interest me is what artificial intelligence (AI) means for financial markets.

While some believe the new technology will make human investors obsolete, I think that fear is overblown.

Here’s the thing… technology is always progressing. And it always has. Every innovation has provided advantages to certain groups for certain periods of time, but the playing field always evens out.

Tammer Kamel over at Quandl recently wrote a great piece on the transience of alpha in markets. In it he covers innovations throughout the years that produced alpha for investors. He explains how none of these advantages persisted because of the “inevitability of diffusion”. Eventually others got access to the information or technology that produced the alpha and nullified it.

The pre-1930’s had robber baron-esque investors planting moles in the government left and right. This gave them a heads up on any large government purchases or sales coming down the pipeline.

Back then embedded players frequently used their connections to get a jump on the markets. But this advantage didn’t last. In this case it wasn’t that everyone else started getting insider information too, but rather the government stepped in with security laws. Either way the source of alpha was eaten up.

Fast forward to the 80’s and 90’s and you had the first wave of computers taking over the markets.

Derivatives were all the rage — the more complex the better. Convertible bonds became popular because of how easy it was to use a computer to arbitrage your way to risk-free profits. But of course once computers became run of the mill, this easy money disappeared too.

Today the Internet of Things (IoT) is the latest informational advantage used to create alpha sources. James from the Operator community recently pointed out a nice graphic in the Comm Center depicting the use of IoT sensors and their data in different industries like farming and energy.

Smart farming, for example, involves placing sensors in the soil to determine moisture and nutrition levels to help calculate potential crop yields.

Smart Farming

Not only will this information help farmers with their resource utilization, but imagine how much it could help futures traders?

Another emerging technology providing an informational edge is Computer Vision (CV). CV involves using specialized cameras to collect detailed imagery data to be analyzed by computers. The insights produced are far beyond anything possible with the human eye.

And yes, I know what you’re thinking. CV is absolutely how Arnold chose his legendary leather jacket in T2…

But this type of robo-vision is no longer science fiction. It’s becoming more and more common. You can find it in everything from cars to even baby monitors. Nanit for example is a super high-tech baby monitor that’s looks like a small lamp post hanging over a crib.

Baby Monitors

This thing is ridiculous. It does everything from measuring a baby’s height and weight, to its sleeping patterns, temperature distribution, and general well-being. It’s billed as a cure to the sleepless nights parents have to suffer through with their crying babies.

My parents had a better, more cost-effective solution back in the day. They’d just let me cry. Classic parenting right there…

But Nanit has larger goals in mind than just a better night’s sleep. They plan to aggregate all the data from the eventual broad network of Nanit monitors to help research early child development. That research can then be used to assist doctors in medical diagnosis. Doctors haven’t had access to this type of data before and the hope is that it could lead to some great new insights.

Now apart from cars and baby monitors, CV technology is being used in all types of tracking. There’s simple applications like Density, which tracks people’s movements in a workspace.

Density Trackers

And then there’s more advanced facial recognition features that are highly useful in things like anti-terrorism.

CV is a brand new technology that’s making far more data available than ever before. And this is where the informational advantage comes in.

Imagine a hedge fund with enough resources to use a satellite with CV to track the number of customers coming in and out of department stores. What could do they do with that data? Possibly extrapolate the foot traffic into something more? Maybe use to help predict future earnings? This application of CV and others like it have the potential to give these funds a huge leg up on other investors.

But like I said, this tech is already becoming more and more common. You even have it in baby monitors now!

Technology is inherently deflationary. The high-tech sensors and CV cameras used in the IoT movement are constantly getting cheaper to produce. This brings prices down and opens the tech up to more and more people. As they gain the same advantages from the cheaper tech, the alpha in the process disappears.

And so it goes, on and on. No matter the technology, alpha is transient, and the playing field always evens out.

But now we we’re on the cusp of highly functional AI. This is the game changer right? This is the technology that marks the end of human involvement in the market.

“This time it’s different.”

Nope, sorry. Don’t think so.

AI will likely go the way of all our past technology. It will be a tool for us to use, not something that takes over the market.

But I see why people think AI could replace us as market participants. When you read about the innovation in the space, you start to think “damn… I don’t have a chance”.

Via Wired:

The New York company Rebellion Research, founded by the grandson of baseball Hall of Famer Hank Greenberg, among others, relies upon a form of machine learning called Bayesian networks, using a handful of machines to predict market trends and pinpoint particular trades. Meanwhile, outfits such as Aidyia and Sentient are leaning on AI that runs across hundreds or even thousands of machines. This includes techniques such as evolutionary computation, which is inspired by genetics, and deep learning, a technology now used to recognize images, identify spoken words, and perform other tasks inside Internet companies like Google and Microsoft.

In the simplest terms, [evolutionary computation creates] a large and random collection of digital stock traders and [tests] their performance on historical stock data. After picking the best performers, it then uses their “genes” to create a new set of superior traders. And the process repeats. Eventually, the system homes in on a digital trader that can successfully operate on its own. “Over thousands of generations, trillions and trillions of ‘beings’ compete and thrive or die,” Blondeau says, “and eventually, you get a population of smart traders you can actually deploy.”

Jesus.

Evolutionary computation, genetic algos, deep learning, neural networks… how the hell do you compete?

Well here’s what I think. This technology, just like every other technology, will soon become run-of-the-mill. The diffusion process has already started.

Diffusion is why you see a higher percentage of quant/systematic hedge funds springing up. They’re adopting this new tech and going to work.

Global Hedge Fund Launches

Diffusion is why good ol’ Stevie Cohen investing $250 million in Quantopian — a fund that allocates capital to investors who use its open-source coding platform to create profitable trading algos. Oh and what does it say on Quantopian’s front page by the way?

Leveling Wall Street Playing Ground

Diffusion is why Google decided to release TensorFlow to the public — allowing anyone to use Google’s technology to experiment with machine learning.

With everyone eventually using some form of AI, the alpha it provides will be eaten away. AI will become as common as a computer. Computers were a huge advantage at first, but do you know of anyone in the markets these days without one? It would be like not having a toothbrush when trying to brush your teeth.

And just like a computer, AI will be a tool where human input is required.

The counter argument is that the supposed point of AI is to circumvent human involvement. Once an AI system is set up, it can take in data itself and create its own investment strategies. It will learn and improve all without any interference from people.

And if that’s the case, the question becomes what will happen to the nature of markets if they’re full of machines battling each other as opposed to people. Will there be a fundamental change to how markets function? Will the human emotion that kept markets the same for thousands of years finally be removed? As Ethan from the Operator community recently asked in the, “Could the emotional aspect of market booms and busts subside?”

My answer? No, I don’t think so.

First off, all technology requires human input. It can’t be avoided regardless of how hard the techies try. We’ve discussed this same topic with cryptocurrencies.

Even with just the AI duking it out in the markets, there will still be human biases embedded in them. And that’s because humans originally made them. Studies are already finding that biases accidently input by programmers when creating a machine learning program actually become amplified as the learning process progresses. Instead of the bias clearing itself out through multiple iterations, it becomes far more pronounced in comparison to any human.

If these machines still have human biases, then we’ll likely still have trends with booms and busts and all the rest of that good stuff.

And this is not to mention the fact that the money AI strategies trade with comes from none other than people themselves. These are the same people with the same emotions they’ve always had. Hope, fear, greed — they’re all still there.

When markets are looking good, money will flood into these AI systems creating a boom. When markets turn, money will stampede out, creating a bust.

Humans still provide the key input to these systems — capital. Their control of this resource all but secures the humanistic aspect of markets.

And here’s the thing, if all these separate AI’s are self-learning with the same data inputs, the technological arms race between them will cause a lot of copycat results. One fund’s AI will one-up another, but then the other will eventually catch up. And the cycle will go on and on, but for the most part, a lot of the AI strategies will be doing the same thing.

And again, there’s no alpha in being the same as everyone else in the market. As Ben Goertzel, head of the AI-centric hedge fund Aidyia explained regarding deep learning —  “If everyone is using something, it’s predictions will be priced into the market. You have to be doing something weird. Finance is a domain where you benefit not just from being smart, but from being smart in a different way from others.”

Is this another potential area where human input will become crucial? You bet.

Human input can function on a level above a particular AI strategy to differentiate it from others in the market. The AI can be used as a tool. And of course while this human interaction may be a benefit, it will also secure the booms and busts we know and love.

My last bone to pick with AI is how people tend to overestimate its power. And at the same time they also underestimate the complexity of markets.

Markets are one of the most complex systems known to man. They are for the most part unpredictable and random.

Sure, prediction of anything is possible when you have every single input and variable available to you. But when is that ever the case? Even with AI-juiced supercomputers with access to the ever expanding internet… this just doesn’t happen.

The number of inputs in a complex social construct like the market are effectively infinite, creating ridiculous amounts of complexity. This is especially true when those inputs come from oftentimes fully irrational human actors. And then add in the concept of reflexivity with perceptions affecting outcomes and vise versa?

We have layers on layers on layers of complexity here. Is is possible for AI to solve one of the greatest puzzles in the universe? Probably not. It’s insolvable.

There are already doubts about deep learning and its application to markets. As Stephen Roberts, a professor of machine learning at Oxford University, explains — “[deep learning could be good] for extracting hidden trends, information, and relationships, [but it’s] still too brittle with regard to handling of high uncertainty and noise, which are prevalent in finance.”

Uncertainty and noise? Yup, those are basically alternative names for markets.

And all this is not even mentioning the unknowns that come with AI in the markets.

Remember the Flash Crash? What about 08’s derivatives mess? LTCM?

All these blowups were instances of shit getting too complicated for people to handle. Now we have self-learning machines taking off and doing what no human can. How much complexity is embedded in that? Do we even understand the market risks?

For example, if all these AI systems begin trading in a similar way, are we going to get frequent massive crashes as all the money in the market shifts to one side of the boat? Does liquidity dry up and cause crises? Who knows!

Maybe AI eventually even gets banned from trading due to all the unforeseen problems the saturated AI market creates. As every good investor knows, unknowns can be very dangerous.

But really the most likely outcome of this AI revolution is that humans and AI work together.

AI is will be just like all the technology of the past.

But hey, I could be completely wrong. (Fallibility, by the way, is something no AI supercomputer will beat me in. Or maybe it will? Ha.)

Maybe machines do take over the financial markets and there’s no room for human traders anymore. Maybe they solve all the complexity. Maybe they figure out the mysteries of the universe while they’re at it too.

If that is the case, then maybe we’re not too far away from the robots just completely taking over in general… like Skynet.

I’m not too worried though. I’m sure Elon Musk will figure something out…

 

 

Subscribe To Our Newsletter

Brandon Beylo

Value Investor

Brandon has been a professional investor focusing on value for over 13 years, spending his time in small to micro-cap companies, spin-offs, SPACs, and deep value liquidation situations. Over time, he’s developed a deeper understanding for what deep-value investing actually means, and refined his philosophy to include any business trading at a wild discount to what he thinks its worth in 3-5 years.

Brandon has a tenacious passion for investing, broad-based learning, and business. He previously worked for several leading investment firms before joining the team at Macro Ops. He lives by the famous Munger mantra of trying to get a little smarter each day.

AK

Investing & Personal Finance

AK is the founder of Macro Ops and the host of Fallible.

He started out in corporate economics for a Fortune 50 company before moving to a long/short equity investment firm.

With Macro Ops focused primarily on institutional clients, AK moved to servicing new investors just starting their journey. He takes the professional research and education produced at Macro Ops and breaks it down for beginners. The goal is to help clients find the best solution for their investing needs through effective education.

Tyler Kling

Volatility & Options Trader

Former trade desk manager at $100+ million family office where he oversaw multiple traders and helped develop cutting edge quantitative strategies in the derivatives market.

He worked as a consultant to the family office’s in-house fund of funds in the areas of portfolio manager evaluation and capital allocation.

Certified in Quantitative Finance from the Fitch Learning Center in London, England where he studied under famous quants such as Paul Wilmott.

Alex Barrow

Macro Trader

Founder and head macro trader at Macro Ops. Alex joined the US Marine Corps on his 18th birthday just one month after the 9/11 terrorist attacks. He subsequently spent a decade in the military. Serving in various capacities from scout sniper to interrogator and counterintelligence specialist. Following his military service, he worked as a contract intelligence professional for a number of US agencies (from the DIA to FBI) with a focus on counterintelligence and terrorist financing. He also spent time consulting for a tech company that specialized in building analytic software for finance and intelligence analysis.

After leaving the field of intelligence he went to work at a global macro hedge fund. He’s been professionally involved in markets since 2005, has consulted with a number of the leading names in the hedge fund space, and now manages his own family office while running Macro Ops. He’s published over 300 white papers on complex financial and macroeconomic topics, writes regularly about investment/market trends, and frequently speaks at conferences on trading and investing.

Macro Ops is a market research firm geared toward professional and experienced retail traders and investors. Macro Ops’ research has been featured in Forbes, Marketwatch, Business Insider, and Real Vision as well as a number of other leading publications.

You can find out more about Alex on his LinkedIn account here and also find him on Twitter where he frequently shares his market research.