After using Chat GPT for the first time several months ago, I was immediately blown away by what the implications of this amazing technology would mean for #commodity markets in general and the #coffee #market in particular.
Since discovering Chat GPT, my programming speed and debugging efficiency has easily increased 500%. Mundane copywriting has increased by a similar amount. I use the software pretty much daily for a variety of applications. In terms of impact, the most apt comparison I can think of would be the introduction of the internet.
Specifically I will cover what #ArtificialIntelligence does well and how it might apply to trading using examples of past technologies and strategies such as quantitative analysis and the invention of the internet.
My conclusion is mixed: on the one hand the technology will be the most transformational new technology in 30 years, but on the other, I don't believe it is a profit producing holy grail. I also provide a somewhat counter-intuitive recommendation.
What is AI?
AI, or artificial intelligence, is a term used to describe the development of computer systems that are capable of performing tasks that typically require human intelligence, such as understanding natural language, recognizing images, and making decisions based on data.
In the context of commodity trading, AI has the potential to revolutionize the way traders analyze information and interpret market data. Using AI traders can rapidly analyze vast amounts of data. It's like having a personal analyst who can sift through all the data instantly and deliver the information you request, without complaining about how late it is or asking for a raise.
Also of interest is that with AI, traders can process vast amounts of data to identify patterns and trends that may not be immediately apparent to humans. It can also be used to develop predictive models that can potentially forecast future market movements. Overall, the use of AI in commodity trading is likely to increase in the coming years, as traders look for new ways to gain a competitive edge in an increasingly complex and data-driven marketplace.
[Improve your coffee trading strategy with our high-quality market reports and side-splitting dad jokes. Sign up now!]
What AI does well:
AI provides several useful capabilities, some of which will be directly important for commodity traders.
Most revolutionary of all the capabilities, is its human readability. AI functions as a powerful search engine, but one that can speak in a conversational way. This allows traders to rapidly analyze vast amounts of data in a way that is easy and intuitive to them. This can help traders to quickly identify trading opportunities faster than ever before.
For example, AI can create programs that execute trades based on predefined rules and parameters of fundamental or technical analysis such as inventory levels, weather patterns or price trends. It could also write the trading algorithm to execute the trades and manage the risk.
This will (or might…) help to remove some of the emotional decision making that leads to bad trading.
Finally, AI is highly replicable, meaning that the same algorithms and strategies can be used by multiple traders and firms, potentially increasing market efficiency and liquidity.
However, there are some big risks, which we will cover in detail in the last section but I’ll highlight just a few here. 1) AI is not entirely unbiased, as it is programmed by humans who may have their own biases and assumptions. 2) While widespread development and adoption of best practices in risk management and trading is generally a good thing, it can also open the financial markets up to widespread vulnerabilities.
Analogs to Previous Technologies and Potential Risks in AI:
The internet and how it Changed Trading
The internet also had a significant impact on trading by increasing the speed of information and democratizing access to information.
Since the adoption of the internet in the 1990s and 2000s, traders have access to real-time (and often free) market data, weather reports, and supply and demand data, all of which helps inform their trading decisions.
[Ready for incredible market insights? Nah, you wouldn't be interested...]
Additionally, tools such as charting and risk management software have become easily readily available, making it easier than ever for traders to analyze and visualize market data.
While, the internet equalized the playing field between the amateur trader and the hedge fund, this also made it harder to make money. Knowing the weather forecast in Brazil coffee regions or having a moving average trend-following strategy are no longer secrets that only a few people hold, they are available to everyone.
In the same way, we can assume that AI will democratize information, trading strategies and risk management to the lay person, where someone with rudimentary programming knowledge will be able to develop sophisticated trading tools with the help of AI. This will in-turn push the professional traders to step up their game into increasingly cutting edge and experimental technology.
Quantitative Analysis Technical analysis has its roots in Dow theory from the late 19th and early 20th centuries, and later in Elliot Wave theory from the 1950s. However, it wasn't until the 1970s and the widespread adoption of personal computers that technical analysis really started to gain popularity.
The use of technical analysis proved highly profitable for some of the earliest users, such as the famous Turtle Traders trained by Rich Dennis in the 1980s. These traders used simple moving average trading systems across a variety of markets, generating large profits and catapulting numerous trainees into riches and stardom.
However, the use of these systems has a couple of challenges: Arbitraged Opportunity and Increased Volatility.
[You don't have to be a coffee snob to appreciate our premium market reports. But if you are, we won't judge - we're just here to give you the best market insights available.]
As more traders adopt the same strategy, it becomes increasingly difficult to profit from them due to slippage. This is because as more people buy a stock or commodity, the price is pushed up, and as more people sell it, the price is pushed down.
This means that if many people put on the same trade at the same time, they will push the buying price high enough and the selling price low enough to remove any profitability.
We can expect the same outcome from AI driven trading models. As soon as profitable strategies are found, the opportunity will quickly be arbitraged from the market.
This can lead to an arms race where it pushes the incentive to put the trade on earlier and to take it off earlier. The earlier we put on the trade the less certain the information.
Hence the widespread adoption of these strategies pushes the trades back further in time so that the probability remains very uncertain. So even with AI, the trading strategies will constantly be pushed to the edges of probability.
This also includes fundamental analysis applications of algorithmic trading such as the Capital Asset Pricing Model (CAPM) or trading weather forecasts. The opportunities for certain, 100% guaranteed trades are quickly arbitraged out and the trades are pushed back in time towards the realm of uncertainty and probability.
The sudden influx of traders following the same strategy can cause rapid price movements uncorrelated to the fundamentals. This is one of the main criticisms of trend following strategies, is that speculative trend following funds can extenuate bull bear markets simply by virtue of the fact that there are more funds trading this style of trading.
In the next section, we will see how Quantitative Trading such as that done by the infamous High Frequency Traders (HFTs) also made markets more volatile.
Quantitative trading and analysis really rose concurrently with the rise of the personal computer. Quantitative analysis focuses on the use of statistics and mathematics to understand and manage risk.
[Improve your coffee trading strategy with our high-quality market reports and side-splitting dad jokes. Sign Up here.]
Large banks and corporations began to increasingly rely on statistical methods for managing risk such in the 1990s and 2000s such as VaR (Value at Risk) analysis. These tools helped banks to understand and manage their risk, but can also give a false sense of confidence and blind-spots to risks.
Quants had their biggest failure in the global financial crisis of 2008.
Banks used mathematical models to manage their risk, but when some of those models had flawed assumptions (that a subprime mortgage could be AAA rated), it led to strategies that took unacceptably large amounts of risk. As this practice was widespread throughout the financial sector, it led to a systemic collapse.
This phenomenon is known as risk compensation.
It was observed in deaths from car crashes before and after the seatbelt was invented. There was some evidence and speculation that car deaths actually stayed the same after the invention of the seat belt because people began driving more recklessly because they believed that they were safer.
As companies increasingly rely on AI to develop risk management plans, they may feel more secure and engage in more risky behavior.
AI driven and developed risk models may make banks, funds and traders feel safer and therefore take on higher levels of risk. If these models end up being flawed and widespread, it opens up the system to widespread failure.
Another recent example of system-driven volatility occurred on May 6th 2010 with the "Flash Crash".
A large sell order by a mutual fund triggered a succession of algorithmic trading that dropped the Dow Jones by nearly 1000 points or 9% in minutes before quickly recovering. I remember when this happened as it was only a few weeks into my first job as a coffee trader and it set the whole office abuzz.
[Why settle for free stuff when you can have it all? Sign up for our premium coffee research free trial and get the value you deserve!]
The big takeway is that AI is here and despite calls for a slow-down from Elon Musk, it seems unlikely that that will happen. We are all going to have to figure out how to incorporate this technology into our lives and businesses, or we won’t last long.
However, we also need to be aware of the risks.
In my view, the biggest risk is that widespread adoption of AI models may make for rapid volatility in the market and also black swan type vulnerabilities.
For example some key datapoint such as weather factor or forecast may change and instantly trigger thousands of AI models to start trading and bring prices up to dramatic levels, perhaps before humans even know what is going on.
Ultimately, the technology may make the markets on balance more safe by instilling best practices, just like hospitals and airplanes have best practices that prevent a lot of harm. Even if widespread adoption of these practices could lead to widespread vulnerability, it is still better to have best practices than not.
Here’s where my view might be a little bit more controversial. While we do need to invest in this technology, we also need to invest in people. Ultimately, it is humans that have to live in this world and deal with the consequences, so it is humans that need to understand how to interact with the world in a safe and productive way.
The best defense against AI vulnerabilities is to be educated. Using the AI to leverage your skillset is smart. Trusting AI when you don’t understand what its creating is not smart. We want our employees to understand AI and use it in an effective way, but we need employees, we need people to drive the technology and not let the technology drive us. [If you found this article valuable, please support our business with a free trial to our premium reports.]