Is It Possible To Predict Stock Prices?

Past share price performances predict the future, right? 

Ding, ding, ding – not so fast.

Any financial advertisement you see with investment advice will declare that past performance has nothing to do with the future success of any investment – yet this quickly spoken disclaimer usually follows an excited spokesperson saying how great a certain investment was in the past.

So, what gives?

If recent events are any indication, the past is quite possibly the worst indicator of future events. So, how Is it possible to predict stock prices or the market in general?

Can You Predict the Stock Market?

Share prices often surface patterns. Technical analysts will commonly spotlight cup-and-handles, bear flags, rising wedges, and so forth.

And no computer has ever been built to pattern match as well as the human brain. Which is why investors tend to overestimate their ability to spot patterns and profit from them. In short, the human brain is prone to errors in judgment.

Very few get really rich when it comes to stock forecasting because the market doesn’t exactly move in line with human expectations.

As with any complex mathematical problem, the future’s variables are far too numerous to be counted, much less predicted. Various models and formulas have been exercised – all of them have failed.

For instance, imagine trying to predict something much simpler than the stock market – consider the game of chess. The number of potential moves is 10 to the 120th power. That’s “1” followed by a whopping 120 zeroes – far greater than the human brain can comprehend, much less actually process.

Our brains aren’t useless – in fact, they’re great at educated guesses. But anytime emotions are involved – and when are emotions not involved when it comes to your money? – judgment errors become our greatest vulnerability – our kryptonite. Nowhere can we see this better than when we attempt to foresee the future.

As Nassim Taleb says in his book, The Black Swan, humans ascribe way too much predictability – teetering on psychic ability – to things and aspects of life in general that are truly not predictable.

The real chafe of the market is how forecasting models thrive on our bias. We have a tendency to believe, if a model has correctly predicted something in the past, that it’ll predict accurately in the future, too.

That’s like believing the carnival trickster who says he can tell you which side the coin will land on just because he’s previously predicted the landing correctly.

Okay – but what about those failsafe market predictions? You know the ones:

• The market always sees dips but they never last more than 10 years
• Housing prices never go down throughout the nation at the same time
• A-rated securities always have higher default rates compared to AAA-rated securities

Sure, some prediction models have proven themselves. Some people are better at forecasting than others. But, as a whole, humans simply stink at predicting the future.

Can Machine Learning Predict Stock Prices?

There are reports circulating suggesting that AI, or artificial intelligence, can decipher the financial market “code” using machine learning.

Machine learning has had its share of successful applications in such items as vision (driverless cars) and language (speech-to-text, robotics, etc.) – so it’s no surprise there are claims it can be just as successful in the stock trade. The thing about markets, though, is there’s really not a “code” to decipher.

However – there does exist a constant quest for the Holy Grail of Prediction. That systematic edge. This edge is the ability of the machine to recognize when to take a risk – and how much.

If you’re considering forking over some dough for this kind of program, you should be prepared to ask some tough questions.

• What gives this program that elusive edge?
• Is the edge sustainable in the long run?

The sobering truth of it is this:

Performance and capacity are inversely related in AI-based trading – the machines performing the trades are subject to this same relationship. The market isn’t stationary. It changes every day, multiple times per day. These changes are driven by outside events in politics, societies, and world economies. Natural disasters even affect the markets.

The data points we’re able to glean are limited by how often we want to “peek” and how far into the future we want to see.

Competition Is Stiff

You could compare the stock market to the Coliseum. Each gladiator saw the other gladiators as adversaries. In the stock market, the gladiators are not only the traders but also the stocks themselves and all those events mentioned above.

This adversarial nature is a double-edged sword.

Any edge discovered or insight uncovered is quickly copied and competed out of existence. So, maybe the ultimate discovery isn’t the Holy Grail after all. Maybe the ultimate goal is to discover a process, or series of processes, that identify conditional changes and opportunities – that can then adapt appropriately. But this makes predictions that much more difficult.

Really, the best prediction “software” is our own brain. Some actually consider our brain one of the most powerful computers ever created – or an enigmatic mystery too difficult to explain.

But the brain isn’t a mystery after all – it’s actually not a computer either. But it is a fantastic machine for predicting. That’s why we’re so good at things like chess and baseball.

Of course, we can’t exactly compute a math problem as fast as a calculator, but we are able to, for instance, realize where the baseball is most likely to land even while it’s still in mid-air. And we don’t even have to calculate things like wind speed, velocity, trajectory, or things of that nature.

That’s what a computer does. Think about that for a second – can you imagine the instant math you’d have to perform to know where that ball’s going to land? But it’d be a fair wager to say you’d (probably) be able to catch it.

Can Money Supply Predict Stock Prices?

There does seem to be a correlation. As the ratio of Total Market Cap to GDP rises above its median level, perceived market risk also rises.

Look no further than the March 23, 2020 announcement by The Federal Reserve to intervene when the stock market was collapsing following the global pandemic.

Thereafter, trillions were added to the Federal Reserve Balance Sheet and the stock market rallied thousands of points.

Can Neural Networks Predict Stock Prices?

As much as you would love to hear there’s finally been a fool-proof method invented for predicting stock prices – neural networks aren’t it, either.

And neural networks don’t really attempt to predict the price. Rather, it attempts an estimation of an upcoming value in a series of values.

This is because prices aren’t encoded with how they might change. For instance, if a stock price continues growing, what’s to say there won’t be some type of external upheaval, like a natural disaster or political turmoil, that causes a plummet later that day?

Prices are affected by these external events that a neural network has no knowledge of.

It may be possible to use neural networks for predicting stock prices in the very short term, say every five minutes, because the potential for external events occurring in these short time spans is lessened.

Remember, always invest responsibly. And if someone tells you they finally discovered how to beat the stock market, share this article with them.

---

The author has no position in any of the stocks mentioned. Financhill has a disclosure policy. This post may contain affiliate links or links from our sponsors.