Use of Artificial Intelligence (AI) in Picking Stocks and AI Companies?
Recently we have seen massive increases in the deployment and use of AI.
From face recognition at airport's passport control, Netflix giving personal recommendations based on learned user preferences and Amazon's table top talking shopping list app. thingy!
(not to mention, the upcoming driver-less cars and mini electric aircraft/flying cars).
There is no doubt it will continue to grow exponentially and seamlessly play a part in many aspects of our lives.
Personally, I could also see it assisting in stock selection. Set your criteria e.g. income/growth, your risk appetite and investment time frame and see what recommendations it comes back with for you to down select and choose from, with push alerts along the way whilst holding. Automatic recommendation of further stocks which become fitting to the identified pattern as it learns what you like to invest your hard-earned in. We can all screen for various criteria at the moment, but isn't AI the natural progression?
After all, it started with Graham and Doddsville folk pouring over Moody's reams of company statistics, to find value and opportunity; to disks of data being sent, via subscription, to investors to load into their desktop applications; to today's online data and toolsets. The evolution continues.
On another track, how do we recognise good AI companies to invest in? How do we separate the good companies from the pure hype? how do we avoid the bubble stocks (like Tilray in legal pot). And, to further complicate, how do we know its not a great company underneath, but has been accidentally hijacked by a large band of hopeful and over-enthusiastic investors (private and institutional).
I'm thinking of writing a (say) monthly or quarterly article about this, sharing my own thoughts and and maybe a bit of AI company research. Just for kicks as it's a topic I'm interested in. Maybe its a topic of interest and could promote some exchange of ideas (or maybe not). Just thought I would put it out there?
Disclaimer
This is not financial advice. Our content is intended to be used and must be used for information and education purposes only. Please read our disclaimer and terms and conditions to understand our obligations.
25 comments
Hi Taff6
Yea, I’m sure there are UK opportunities. I just meant generally Japan seem to be more interested in the future adoption of HFCs for transport.
In this sector ( or subsector or whatever it is), I’m especially interested in electrical power storage with a light weight. Next step after Lythium Polymere ( which is quite volatile). If these could advance at the same speed as the advances in AI there would be some real tech convergence and rapid adoption.
Gromley
LOL, Your Welcome, great postings BTW, be interested in your updates
https://www.ishares.com/uk/ind...
https://www.ishares.com/uk/ind...
Click on the ETF of choice, e.g iShares Automation & Robotics UCITS ETF
click on top Download button top right hand side of screen for further details
These are the top 30 in the iShares Automation & Robotics UCITS ETF
May be a good place to start your quantum computer tech research!
I've also read a little about the automotive sector of late EV, nevertheless are people backing the right technology or should we be looking at Hydrogen, Hydrogen/EV hybrid. Does anyone have any views here?
Yes. I have a view. Investors should completely ignore this thread & focus their investment into more predictable sectors with less hype.
Others may disagree, That is how markets are created.
Hi JonesJ,
I was really wondering if I should even reply.
I agree with you that all opinions are valid, but if you have read the thread you would see the posts cover many topics.
If you are just referring to AI companies, are you suggesting that people should not have invested in Nvidia, for example?
If your not interested, that's fine, but to suggest investors should ignore it is a bit strange? Surely Investors can make their own minds up what they choose to read.
I 'm not really expecting a reply, because of course, as an investor, you're not actually reading this anyway ;-)
Hi Taff,
Japan seem quite into Hydrogen fuel cells for the future.
Clean, as the only emission being water, however, currently main issue is you have to produce the pure hydrogen in the first place. This uses energy, and probably emissions at the beginning.
mmarkkj777
Thanks for your reply will look into Japan, however feel there may be some opportunities closer to home. Interesting choice for post BTW.
https://en.m.wikipedia.org/wiki/Quantum_computing
Thanks Taff - yes I have a feeling I might need one of those quantum computers; or possibly minus one if I can't quite work out where it is!
Hi Carey,
Yes you are right, according to Deloitte all top UK businesses (85%) are planning to invest in AI over the next 3 years.
I held Nvidia shares for 3 years prior to October, and I may buy back in when I think the market has completely settled (who was it that said, "when there is a gold rush, invest in the companies that make the picks and shovels").
I still think, though, that we have only seen the tip of the iceburg for AI. I think the peak of Bitcoin is gone (just a personal opinion, not a recommendation to not invest :-) ).
Companies like Blue Prism that build software 'robots' (programs), to do routine computer based jobs to automate business processes, are still in their infancy (Blue Prisim has really grown but its not profitable yet).
Investing in AI now is a bit like investing in "websites" 20 years ago. Its not really a sector its a tool inside all sectors. At this point almost all successful companies will already have AI in their business already and will be developing it further.
An interesting play I have looked at is GPU manufacturers like Nvidia and others. All AI requires processing power, if you assume AI usage will increase then manufacturers of number crunching chips will do well.
I've seen it written that NVidia's share price progress over the past 5 years is largely driven by the additional processing required by Bitcoin (and other coin) miners and AI.
You may well see Nvidia as a major holding in the funds mentioned above (I haven't checked those specific ones but I have seen it in similar funds).
I have a passing interest in AI and have taken 2 minor positions in an AI powered ETF and AI Investment Fund.
The ETF is AIEQ (AI Powered Equity) and it buys/sells US companies using learning techniques powered by a IBM Watson computer.
https://www.etf.com/AIEQ#overv...
The Fund is Smith and Williamson AI
https://smithandwilliamson.com/en/personal/services/investment-management-funds/artificial-intelligence-fund
Its early days yet and both suffered from the October correction, but I am prepared to hold and track what happens.
Thanks Ian,
I'll have a read of those links.
Interesting post mark and interesting (to me) that you posted it now as just this week I have been turning my mind back to a project to use computing power and particularly "text analytics" to interpret RNS announcements.
Some of what I have in mind there could be refereed to as Artificial Intelligence / Machine Learning.
Mostly though I think the term is used to widely and I would say that a lot of the thing you refer to above are more properly just advance algorithms rather than AI as such.
It seems to me that Stockopedia already has a plethora of the type of thing you are looking for - StockRanks, screens the various indicators on the Stock Reports.
I'm not really sure therefore what it is more that you are specifically looking for?
Hi Gromley,
Thanks for your post.
Second point first.
Yes Stocko has lots of things that already go along the path I'm referring too. Stocko is great and I'm still learning how to get the best out of it (and learning more about investing in the process). In my musings (and that's all they were really in the first part), I was just thinking that the research ( that we all do) lends itself to AI because its very much rules based research and analysis). The real distinction of AI above other clever programs is its ability to learn (from feedback, mistakes and successes) and modify/improve how it operates. It could then operate as a useful sidekick or assistant.
Like most things in life these days, cars, computers, phones, etc. etc. there is actually nothing wrong with the ones we have today, but they will still improve as each new generation comes out. Its only when we step back and look at what we were happy with a decade ago that we register the real changes and evolution. I think AI is a natural evolution in investing (Institutional and Private).
First Point,
On your text analytics for RNS etc. Robbie Burns already does something very similar using an existing bit of functionality. He selects certain phrases, like improved, increased, better than, etc. and his search results come back highlighted green. He searches for issues, problems, difficult period/markets/conditions, etc. and his results come back highlighted in red. So he's still doing his research, but refining the process to save himself lots of time reading through the whole Company reports and announcements, etc. he says he knows he misses bits, but feels the time saved can be better spent.
The other bit of my post really was just to gauge if there was any interest in reports about AI companies (capability providers, like Blue Prism etc. as well as early adopters) . Not hype or ramping, just ideas about who is doing what and if its a company worthy of further investigation. I have not done this type of report before and fancied having a go, but only if I felt is was useful.
By the way, while we are here, I was wondering how your strategy of Long/Short held up during the small(ish) correction we had in Nov? I was interested when I read your post on the topic. I went to cash, but not before losing some whilst my stops triggered, etc. And buying back in now has its costs. At least it has given my chance for a rethink on how I want to structure my portfolio.
Hi Mark
On the long short idea - its been going very well (although frustratingly I've been a little tardy getting to the 50:50 Long : Short split that I had envisaged, so haven't felt the full benefit - in fact I suspect I would have been better to be 100% short, but that won't always be the case). I will do a report on the performance of the "paper portfolios" in the new year. I had also considered blogging on the ongoing performance of my real life shortolio, but haven't quite made my mind up on that.
Anyway back to AI. Agreed - it essentially about the ability to learn.
I guess the various things I'm working on at the moment fall slightly short of that :
It's funny again that you mentioned Robbie Burns's approach to RNSes as that's something I read about for the first time only this week. I'm essentially hoping to go one stage further than that and rather than just colour code the 'good' and 'bad' words but go on to 'decide' whether on balance a statement is good or not.
There are so many nuances to teach (for example "a good start to the year" might not be so good if the company routinely reports "an excellent start to the year). It may all turn out to be a vanity project, but having programme managed a system roll out that was big on Text Analytics I have got the bug a little,
Anyway if I get anywhere with this, I'll no doubt come back and tell.
As for investment in AI companies, it's not something I had specifically considered really, it feels a bit like investing in companies with blockchain in their name, End users won't buy AI, but they may well buy products that rely on AI. I don't mean to poo po the idea but it's not something I think I'd want to take as a primary focus.
Hi Gromly. Glad the long-short went well. As you know I’m planning to adopt something similar at some point.
Good luck with the Phrase recognition work. Please keep us posted. I’m really interested to see how you do.
Regarding AI focussed companies to invest in, I agree also and it wasn’t a main focus of mine. Probably only 6 or 7% of my total portfolio, but I think the two, Nvidia and Blue Prism, accounted for around 17% of the growth over 2 years, especially Nvidia.
Btw, Blue Prisms list of clients is very impressive many big blue chips. If they could only start making a profit......
Hi Bwakem,
Well you are partly right!
It not that I can’t be bothered to do the research, it’s more a case of I think AI could help to take the donkey work out of the research to allow me not to miss potential opportunities by time constraints. Similar to why I use Stocko now. I don’t use Stocko because I can’t be bothered to research in the way I ( we all) used to. It’s just that then I was running a company and didn’t have the time, so I ended up not researching effectively. Now I’m semi retired, I have more time, but I think the time would be better spent making the decisions based on the collected data ( and assessing the qualitative angles), rather than wasting time digging for data that is better done by machines.
The main step change with AI is its ability to “learn”. To learn a persons research preferences, then go out and get the data and present options and potential warnings, is not far beyond where we are today.
The other aspect I’m interested in is companies who provide AI capabilities, and companies who are early adopters.
Although I have held Amazon ( recently sold) I didn’t buy in at the beginning so no 10000% gain for me.
One company that I was an early investor in was a company called Autonomy. They had some great ideas, but the management ( shall we say) had their difficulties. They sold to HP about 6 years ago. I did well in the late 90s early 2000s though.
So in a nutshell, not so I don’t have to do the research. Just to augment research, because it’s the decisions that are critical, the physical searching is time wasted, and probably would be more thoroughly done by machine. Isn’t that why we all use Stockopedia?
By the way, if you got a 57% win ratio that’s actually really good. Many successful traders make a living with that or less. It depends on the deal sizes and gaining more on the 57% than you lose on the 43%. Profit targets, stop losses, etc ( but I’m sure you must have looked at this).
It sounds like you want it to pick the stocks you would have picked if you could be bothered to do the research yourself! That sounds like an episode of Black Mirror (Netflix) where it was possible to copy your brain into a computer and let it do stuff for you. That's fine if you are a good stock picker.
I think more useful would be a program to identify good opportunities automatically. The trouble with that is of course that as soon as it it works and becomes known about, it will change peoples' behaviour and it'll break.
I come from a programming background and spent a few weeks working on various machine learning and genetic algorithms to find patterns in and predict price movements in the dow, on a day-to-day basis and per-minute. Unfortunately all I did was prove that the price movements are essentially random and therefore not predictable by AI. The best long-term win ratio I could predict was about 57%, which after the spread and potential slippage and the risk of draw-downs made it not worth trading.
I haven't given up though.
Hi JonesJ,
That’s a really good point and it rather covers it.
AI is a really good growth sector. Like oil, railroads, automotive, aviation, computers, internet.
How do we find the good ones? ( a lot of the old breweries are no longer with us either).
Amazon does sell stuff ( but so do millions of other businesses). The story is in how they did it.
Finding the good ones in early stages of a new and growing sector is the way to growth, but is not easy, as you say.
I guess it’s the usual: good idea, well financed, strong management team, IP or a moat, etc.
If its all to difficult and not worth the effort for you, then I guess you have answered your own question.
But just because you don’t want to doesn’t mean others shouldn’t bother, or am I missing your point?
It’s a growth sector that I’m interested in.
Seeing the near-magical feats that machine learning is capable of, the number of markets and the intense application of algorithms to investing in them, I'm surprised there's any money left on the table for humans. I mean, I'm not surprised, because markets are not closed systems with just one group trying to manipulate each. But on the other hand if volatility just drained right out one day because some AI temporarily cracked the problem, I wouldn't be shocked at all.
I'd like to know where AI is adding to efficient markets and where it might be adding to inefficiency and volatility.
A good point Snickers,
I’m sure early trading algorithms contributed to market volatility and dips. Maybe even a large factor. However, perhaps as they have matured they also identify the renewed value and buy in the dips, helping to moderate ( as they don’t have the panic behaviour). Personally I still think there will be a place for humans to control the allocation of their own assets.
There are obviously the Quants, but not everyone will want to invest this way. Has any quant outperformed Mark Minervini for example? I can certainly see it taking out some of the leg work though, making a more systematic, metrics based investment approach easy to accomplish.
Sorry JonesJ,
Made the last post before you made the edit, so not really applicable now.
I just thought that Stocko was mainly for self-investors. There are always people who have studied and done research in all fields. It doesn't necessarily make them better investors.
It's not everyone's cup of tea, I understand that. People are more effective investing in what they know, however, things change.
Warren Buffett spent half a lifetime (everyone else's full lifetime) saying he wouldn't invest in Airlines or Tech. He's now done both.
Cheers,
Mark.
Hi JonesJ,
Thanks for your post.
How would you have categorised companies like Amazon, Uber, Netfix, Google, Airbus,..... then?
Well, there is a good question, so thank you for it.
Amazon sells "stuff", so no specialist knowledge is required to figure out what they do, or even why their business is different to Woolworths.
Even then, I have missed the 1000 bagger Amazon opportunity, whilst also missing out on loads of other tech companies which have lost a fortune. [The unfashionable brewing companies added to my portfolio in 2000 did just fine]
Figuring out why AI is any different to plain software and who has a sustainable advantage looks difficult. Also, it's fashionable, so lots of other people will be looking into the field.
Joining in myself is entirely optional.
Then if we look back at history, about a century ago, there were hundreds of car makers. Only a few thrived and prospered. There were also many entrants into the new industry called aviation, but many airlines have failed and lost a lot of money. Incidentally, Warren Buffet has done well by avoiding airlines for about 80 years. Perhaps that business now looks better after consolidation in the US.
So making money in any new field such as AI field is not guaranteed. I think it's still necessary to exercise exceptional judgement.
The first question, would be is it worth the time to get to know the field properly, when we know there is a lot of hype in the sector ?
Or should something quieter & less fashionable be the priority ? Where I don't have to compete with some PhD in AI.
I'm open to being convinced.
"On another track, how do we recognise good AI companies to invest in? "
That's in the "too difficult" pile.
I don't see why I would even want to compete with people who may have studied and done research in the field, therefore gaining sector specific knowledge.