If you’re one for an emotion-based rollercoaster then look no further than the stock market.
The ups and downs of the many forces at play on share prices have made it impossible to predict, at least for now.
Over the last few years, the decision making process about what to invest in and when has increasingly been taken on by artificial intelligence (AI).
One company has even taken decision-making entirely out of human hands, launching a hedge fund making all stock trades using AI without any human intervention.
Aidiya, a Hong Kong based AI company generated a 2% return on an undisclosed amount in a day. While a short sample, that’s well above the market average.
‘If we all die, it would keep trading,’ Aidyia’s chief scientist Ben Goertzel told Wired.
The algorithmic trading market was valued at $9.3m (£7.2m) in 2017 and is projected to grow by over 10% annually.
And algorithms have had success.
In 2008, when the financial meltdown happened, the Dow Jones index lost around 50% of its value in 18 months.
In that year, an AI model devised by researchers returned 681% on investment. Over 13 years from 1992 to 2015, that study led by Dr Christopher Krauss averaged a return of 73% a year compared to an annual market growth of 9%.
So why aren’t the machines being left to run the stock market?
‘Stock markets have been using automation and machine learning for at least a decade now,’ Devina Paul, founding partner of Galvanise Capital, tells Metro.co.uk.
‘Some kind of high skilled human intervention has been and will always be required.’
This is said to come down to the ‘chaos’ of the markets and that self-fulfilling prophecies alongside all sorts of unquantifiable factors make human emotion and sentiment – a key ingredient to stock fluctuations – impossible to predict.
There are two types of chaos: level one and level two.
Level one chaotic systems are those which don’t react to predictions – the weather for example – and level two are those which do react to predictions, like politics, public protests and, of course, the stock market.
Does that mean that machines will eventually get to the bottom of the weather forecast but never the FTSE index?
‘Algorithms have turned out to be particularly effective at such times of high volatility [as in 2008], when emotions are dominating the markets,’ study leader Dr Krauss, of Friedrich-Alexander-Universität Erlangen-Nürnberg in Germany, has said.
‘During the last years of our sample period, profitability decreased and even became negative at times.
‘We assume that this decline was driven by the rising influence of artificial intelligence in modern trading – enabled by increasing computing power as well as by the popularisation of machine learning.’
This is getting confusing now, machines are better at predicting humans than other machines?
It used to be so much simpler.
There were always three methods to analyse and predict the stock market: financial, technical and sentiment.
- Financial analysis evaluates past statements, reports and balances sheets and comparing it to prospects, the market and changes in government policy
- Technical analysis relies on the idea all factors which can influence the price are included in the current price of the stock and therefore no fundamental information analysis is required. They believe that the prices move in trends and the same historic patterns
- Sentiment analysis relies on taking advice from experts and going through newspapers to monitor the stocks they’d like to invest in., going through text and data points to understand any changes that would move the price
Now machines are ‘effectively crunch[ing] millions upon millions of data points in real time’.
We’ll get back to the machines predicting machines question later.
In 2018, Fund manager Guy De Blonay suggested that 80% of the stock market is controlled by AI, making decisions faster than any human trader could do.
Yet traders are still hesitant about using machine learning tools, a Greenwich Associates study found, and 75% of respondents still do not use AI when it comes to trading stocks.
The Tiger Bears And Bull Index
Two different studies both suggested that an artificial neural network (ANN) may be the able predict the markets with higher accuracy than traditional methods.
But the most accurate predictor of markets in recent times appears to be the career of golf superstar Tiger Woods. It was popularised 10 years ago by entrepreneur Jeff Stibel, when Tiger’s tournaments positions were tracked against the Dow. The correlation is startling.
Is it a coincidence that Trump is now citing record market highs at the same time Tiger is winning again?
Of course it is and the Tiger Bears and Bulls Index as it is known is obviously utter nonsense but it is doing as good a job as any stock trader.
So why can’t we predict it?
And it could be because we’re still relying on humans.
Human factors, emotions and sentiment has created a new market of social media sentiment collection companies, which use social media to understand and harness human emotions.
Companies such as Social Market Analytics, Stocktwits and Dataminr are using real time social post analysis to understand how consumers are feeling and where shifts are likely. Using sentiment analysis on social media was found to show cumulative stock returns of 76%.
A 2016 study placed importance on using sentiment analysis to understand fluctuations in the market, though highlighted that such tests are biased towards blue chip (established multinational) companies which move in tandem with the broader market as a whole.
And there’s still the big question posed in studies that the more AI used in predictions, the less that AI can predict.
Can that really be the case?
‘Predicting what they will do depends on what they predict you will do, which in turn depends on your prediction of what they are going to do… back where we started,’ Dr Ron Chrisley, director of the Centre for Cognitive Science at the University of Sussex, tells Metro.co.uk.
‘Thus, the behaviour of the system is an unstable, chaotic circle.
‘The AI is typically part of the [financial] system it is trying to predict. The predictions the AI makes are then used to act in that system.
‘The presence of predictors (machine or human) in a system, acting in that same system based on their predictions makes the system more difficult to predict.’
We’re back to that second chaotic system again.
Even after all that, it seems that artificial intelligence will still be left with the human condition of imperfection.
‘In 10 years’ time, machines will be able to predict the stock market over a longer timeframe but never at a 100% accuracy,’ Swiss entrepreneur and investor Alessandra Sollberger tells Metro.co.uk.
‘There will always be some data which isn’t captured; you’re assuming full accuracy in your data capturing process, and that kind of state is impossible to achieve, even for machines.’
But as with all predictions about the stock market, the value of them can always go up and go down.
At least until we have a machine to work it out for us 100% of the time.
The Future Of Everything
This piece is part of Metro.co.uk’s series The Future Of Everything.
From OBEs to CEOs, professors to futurologists, economists to social theorists, politicians to multi-award winning academics, we think we’ve got the future covered, away from the doom mongering or easy Minority Report references.
Every weekday, we’re explaining what’s likely (or not likely) to happen.
Talk to us using the hashtag #futureofeverything If you think you can predict the future better than we can or you think there’s something we should cover we might have missed, get in touch: [email protected] or [email protected]