Algorithmic trading owns results season
When results season hits and share prices swing 5 to 15% in the time it takes to read a headline, algorithmic trading is doing the reading for you – but not the deciding.
The world’s gone mad. We’re in the middle of the US results season. Microsoft (NASDAQ: MSFT) has gone up 15.5% on its results, and it did that within minutes of the results announcement coming out. At the same time, Meta (NASDAQ: META) dropped 7.9%, Amazon (NASDAQ: AMZN) was up 10.1%, Apple (NASDAQ: AAPL) down 6.2%, and Alphabet (NASDAQ: GOOGL) down 7.1%. These are some of the most researched companies in the world, moving 5 to 15% within minutes of their announcements. Who is doing that? Who’s putting the trades on? I tell you, the robots are taking over. Has the human lost control? Let’s have a look at what’s going on. Let’s have a look at why share prices move enormous amounts within seconds of announcements, and let’s work out whether this is a good thing, a bad thing, and whether we can exploit it.
How research used to be done
Right, I’m a bit old school. When I joined stockbroking in 1982, it was a firm called Buckmaster and Moore. In order to have credibility with clients, they tried to collect – can you believe this – the annual reports of some of the biggest companies in the UK. If you didn’t have a library with the annual reports of British Telecom, Vodafone, Racal, British Steel, BP, and Shell – if you didn’t have at least ten years’ worth of their glossy printed annual reports – it meant the company didn’t think you were a proper broker, because the way companies used to disseminate information in those days was to have their accounts printed and sent to the brokers they thought were worth sending their annual report to.
When we got the envelope, it would be handed to the analyst on that particular sector. They would open up the announcement and think about it for a while, ruminate, maybe jot down a few notes about whether they thought they were any good, and hand them to the secretarial pool, who would type them up on a typewriter. Then they would get taken down to the printing press, and the analyst’s notes would be printed out and handed to all the salesmen. And if you were lucky, after two or three days, the analyst would write up two, three, four, or five pages, and these pieces of paper would go down to the printing press, get stapled together with our heading on the piece of paper – which was pretty fancy – and sent out by the messenger department.
Every broker had a printing press and a messenger department, and they would deliver it to the fund manager. The fund manager would hand this to the analyst on that particular sector at his fund management firm. They would get ten envelopes from the ten biggest brokers, open them all, ruminate, and after a couple of days hand it to the fund manager, who would then decide whether they needed to buy or sell. And that’s how the stock market worked.
How the AI actually does the research
These days, within minutes of a huge quarterly results announcement from Alphabet coming out, the share price is down 7.1% – literally within seconds. And how does that happen? How is the research done that quickly? Have the robots taken over? Well, let me tell you what is happening, and it’s actually quite a good thing. The big fund managers, the big hedge funds – I’m not sure we’ve really started doing this in Australia efficiently yet, but it’s clearly happening on Wall Street and in the major financial centres – hedge funds, fund managers, brokers, and investment banks are getting the announcements and dumping them into pre-programmed AI programs which are then doing the research.
The good bit about this is that the AI engines are not actually making judgements. What they’re doing is operating the pre-programming that’s been done by the analysts or by the fund managers, and they have programmed the AI to look up particular bits. If we were to boil it down to something quite basic, what they could be doing is extract all the numbers they’ve been asked to extract from the announcement, put them into a discounted cash flow valuation, and add up what the company’s worth. And if they’ve started out with an intrinsic value before the results, after the results AI’s worked out that the intrinsic value is this, which could be below or above. Then what happens in some of the fund managers, brokers, and hedge funds is that the algorithmic trading program kicks in and starts selling or buying the stock and driving it down to the new intrinsic value or up to the new intrinsic value.
So this is why things are moving so fast. The humans are still very much involved and in control, because they are programming the AI engine to interpret the results in a particular way, depending on their current expectations and assumptions. Those all still have to be put in. All that’s happening is the research on the day is being done extremely quickly by AI, or by a program the humans have probably more carefully programmed to analyse the results – and that is a competitive edge. It doesn’t denigrate the value of the research at all. In fact, it’s exactly the same quality, if not better, because the AI doesn’t make mistakes.
Why the human is being phased out
This is an evolving part of the stock market. It is going to become more prevalent, it is going to become more refined, it is going to become better. To have a human in that chain slowly, ponderously trying to do what an AI can do vastly more accurately in seconds is probably a disadvantage. So let’s take the human out.
In fact, a good example of how inefficient a human is: our input devices are ears and eyes. You could put on a USB drive more information than I could possibly read in the rest of my life, and yet a computer can pick all that stuff up. An AI engine can pick all that stuff up and analyse it in seconds.
We are being phased out of the research process, because speed matters, and it’s probably a good thing, because these AI engines are going to do things faster and more accurately. But it doesn’t mean the thought hasn’t gone into it, because these AI engines have to be programmed, and they’ll evolve over time to become vastly more accurate and relevant. They are still doubtless being programmed by humans making mistakes and being inefficient, and it will get faster and better, and the algorithmic trading will evolve as well.
Trading safely through results season
But now we are left with the idea that share prices are, I’m afraid, going to move rapidly on announcements, because anyone who’s not employing an expensive programmed AI engine to do the research is going to be left at a disadvantage. I don’t mind that too much, but it has made results season rather risky. I describe results season as a bit like running around on a battlefield in an orange vest – you don’t know when you’re going to get blown up. If you’re in small-cap, mid-cap, even large-cap companies these days over results season, you’re taking unnecessary risk.
It might actually be better as a trader not to hold stocks during results season. But after they’ve announced their results, after the share price has moved initially, usually big moves start trends. So a big rise means buy, a big fall means sell. And after you’ve seen that initial move, whatever it is, you are now in possession of all the facts – the stock has been derisked. All the information is already out there. The research has been done within minutes. And now you can buy the stock on bigger drivers, like what their industry is doing, what the market is doing – and your chances of catching the trend are better than guessing the results reaction.
So, AI research is here to stay. Don’t worry, humans are still in control, humans are still programming these things. They will become more a part of our stock market life. You can still make money without being an algorithm, without having systematic trading software, without having trend-following software. There’s still an opportunity. The market is not efficient yet. You can tell from the share price movements on results – you can still get an edge, maybe not on the fundamental research, but on other factors like trend timing, and you can manage your risk through quality, earnings, fundamentals. The game’s the same. It’s just that announcements are being more efficiently, and more quickly, factored into the share price. And I’m afraid that’s here to stay. It’s not going to change. But it’s not a big risk. We’re still in the hands of the humans.
If you want a bit of help trying to make money out of the stock market, sign up for a free 14-day trial of Marcus Today. And if you’d prefer to leave it entirely to us, take a look at MT20.