AI can be used to predict future performance of the stock market
Getty
Have you ever wondered why your stock portfolio isn’t performing as well as a basic “stock market” ETF—or worse, your own expectations? That despite having the “right” stocks in your Robinhood account, your overall portfolio is not where it should be? There’s a reason why this is happening and it’s simple: buying a basket of stocks and calling it day just isn’t enough.
Determining how much of a stock to buy isn’t as simple as a basic math equation. Why build your portfolio based on yesterday’s stock market? Advancements in Artificial Intelligence (AI) make it easier now to predict what tomorrow’s stock market will look like with a high level of probability. Armed with this information, portfolios built with AI methodologies will structurally perform better for individuals.
What exactly is a diversified portfolio?
The concept of diversification is used so ubiquitously when discussing investing, its true meaning is typically lost (or missed!). When we think about diversification, we are referring to risk in your portfolio. There are multiple risks to consider but investors typically focus on only one type: performance. Measuring performance risk is obviously a key metric for an investor—why on earth would you be investing if you thought you wouldn’t get positive returns! But what many don’t know is that overemphasizing a focus on returns could raise other risks in the portfolio.
Here’s an example. Let’s say you put 40% of your portfolio into stock ABC. ABC goes up a lot but then suddenly drops within the same day. As a result, we would call stock ABC a “volatile” stock. Everything’s great when ABC is up, but what happens on a day when ABC drops 10%? Since you put 40% of your portfolio into stock ABC, a 10% drop means that assuming there are no changes to the rest of your portfolio—your portfolio would be down 4%. In order to break even, you would need your portfolio to gain more than 4%. That, my friend, is the power of compounding, and it is working against you. Perhaps it would be better to own a smaller portfolio percentage of ABC, giving you exposure to one of your favorite stocks without putting your entire portfolio at risk. his is called the risk-reward relationship. We’ll touch on that more in a bit.
The right kind of diversification
Knowing not to buy too much of a highly volatile stock only scratches the surface.You need to next consider the role and impact correlations have on your portfolio. Let’s walk through another example.. Say you want to buy stock ABC and stock XYZ. You’ve noticed that every time ABC goes up, so does XYZ. And when ABC drops XYZ drops, too. What we just described is what we call a positive linear relationship—also known as correlation.
Correlation is measured on a scale of -1 to +1.This range is where the linear relationships of changes in securities—in this case stock ABC and XYZ—is normalized. If ABC and XYZ have a correlation of 0.75, that would mean there is a relatively strong positive correlation between the two stocks. In other words, you can expect that every time ABC goes up, XYZ will go up, and vice versa. Therefore, you would be introducing similar risk into your portfolio by buying both ABC and XYZ. To avoid concentrating risk into similar or related ideas, it would make more sense to invest in a stock that has a weaker correlation to stock ABC. Otherwise it is just as risky as buying all of stock ABC.
But what about Risk?
Remember that risk-reward relationship we mentioned earlier? This is where it comes in. While risk-reward can be measured many different ways, we’re going to focus on the basic relationship of return to volatility.
When considering risk-reward, ask yourself how much return are you getting for every unit of volatility (risk) you accept. If you have a ratio below 1, that means the payoff you receive in portfolio returns is not enough to offset the risk of accepting volatility in your portfolio.. In a perfect world, you would want the risk-reward ratio to be well above 1.0. Keep in mind that the amount of portfolio return you receive significantly outstrips the risk you are taking with your portfolio. So when you think of building your portfolio, always focus on this relationship to help maximize the best risk-adjusted payoffs in your portfolio.
Putting it all together
As a reminder, these are the 3 main things to keep in mind when building a diversified portfolio:
- Avoid filling too much of your portfolio with high volatility stocks
- A strong correlation between stocks could introduce additional risk
- Always consider the risk-reward ratio when accepting volatility in your portfolio
After building a diversified portfolio, the next step would be to apply these rules to your portfolio. Typically, your broker will offer you some type of software to build your portfolio but the majority of these features are rudimentary and don’t offer the right tools you need to find the best portfolio. Unfortunately, that leaves Microsoft MSFT Excel and software engineering solutions as the only viable options that remain. Our favorite software engineering solutions are in R and Python, but running optimization schemes can be done in many programming languages.
How does AI fit in—and why should anyone care?
This might come as a shocker, but all of the exercises we outlined above are more or less worthless in their predictiveness. What’s even more wild is that most institutional investors on Wall Street are using basic math solutions that we all know is false. The problem with what we outlined above is that it is backwards looking—it is based on past performance. You may have heard of the saying “past performance is not indicative of future performance.” In this case, looking at past price performance is not going to give us any insight into how stocks will perform going forward: it is an ex ante vs ex post debate. Enter our AI asset allocations models from center right.
AI asset allocations models—specifically AI clustering models—help make more robust “forward” predictions of what stock prices will do based on a host of factors (which we won’t touch on in this article). However, the end result is always more stable portfolios since the inputs going into the calculation are forward looking. Let’s take a look at one of our favorite AI asset allocation models, Hierarchical Risk Parity, to see how it all works.
How does the AI asset allocation model work?
First, we consider correlation and how different correlation pairs differ by measuring the absolute distance between different correlations. By doing this, the algorithm begins to make sense of how these differences cluster together. We repeat this step until we are left with one large cluster. Then, we retrace our earlier steps and allocate weights to the different stocks depending on variance within each cluster. To better understand non-linear relationships, these clusters are arranged by importance (hierarchy). And just like that, we have a portfolio allocation that best represents our best bet for maximizing our return, minimizing our volatility for future periods by considering non-linear relationships, not just basic linear relationships.
Source: Forbes – Money
