By Jayson Forrest - Managing Editor  - IMAP Perspectives

According to James, there are four key elements of AI:

1. Data;
2. Machine Learning;
3. Hardware/software; and
4. Domain knowledge.

“There has been a shift towards an abundance of data. Big data has grown exponentially over the last 10 years, particularly as a result of the internet. There is a massive focus on having quality data, because if you feed AI good data, you get good results,” he says.

Machine learning is also a key component of AI, which allows for data analysis that automates analytical model building. It is a branch of AI that is based on the idea that systems can learn from data, identify patterns and make decisions with minimal human intervention.

“I view machine learning as a subset of AI. It’s a set of mathematical techniques and algorithms that can be used to create predictions,” says Dr David Allen - Head of Short/Long Strategies at Plato Investment Management.

“They can range from the simple, like ridge regression - a method of estimating the coefficients of multiple-regression models in scenarios where independent variables are highly correlated - to the complicated, like deep learning - a part of a broader family of machine learning methods based on artificial neural networks with representation learning.”   

The two other key enablers of AI are technology - which is enabling better management of massive data sets, as well as open source coding and Cloud computing - and domain knowledge, which is about properly understanding the problem that you are seeking to solve and knowing what the intricacies are of the area that you’re working in.

Importantly, the conduit for these four components of AI is the “human element”.

“The human element is important, because AI doesn’t develop itself. It’s humans that develop AI and we input our beliefs into each of these four areas that enable AI. Humans make the decisions about how to manipulate the data, and make the decisions on what sort of machine learning techniques are used,” James says.

David agrees adding the best application of machine learning is when it’s applied with a human hand on the tiller. He uses the analogy of an aircraft’s ‘auto-pilot function’, which requires a pilot (the human element) to be in the cockpit when this technology application is switched on.

However, James warns that if portfolio managers use more complex machine learning techniques in the portfolio management process - like random forests, booster trees and neural networks - they often sacrifice the interpretability of the outcome.

“Whilst these more complex techniques tend to provide marginally better results in predicability, they can’t tell you what factors contributed to the outcome. So, when using these complex models, it does become a lot harder to interpret them,” says James. “That’s why we opt to go with regression models, which are simpler and much more interpretable.”

With the automation of advice and the growing trend amongst portfolio managers to use AI in the portfolio management process, does that mean AI will eventually replace the role of portfolio managers?

“At Longwave Capital, we are agnostic at looking at any technology that can help us improve the outcomes of the decision-making process,” says David Wanis - Founding Partner, CIO and Portfolio Manager at Longwave Capital. “So, if there are parts of that process that can be improved, like parsing the content of an earnings call using Natural Language Processing, then we’ll use it.”

Longwave Capital has kept its portfolio management technology relatively simple. It starts from the point of determining the economic hypothesis of why a particular stock signal might work, and how that signal is connected to the real world and what’s happening within it.

“And then there’s the issue around ‘black box models’ to consider. These models may generate stock signals that are interesting, but as an investor, you can never understand the basis on which a particular stock was chosen, or the position size chosen, or the diversification chosen,” he says.

“However, with the Longwave Capital model, we combine insight with systematic tools, which enables us to understand on what basis a systematic signal is being recommended and how we can combine it in the portfolio.”

So, from David’s perspective, he believes portfolio managers will continue to use technology, as more of it becomes available, to help make better decisions for clients. “In that respect, it doesn’t feel like we’re at risk of being replaced by AI anytime soon.”

It’s a view supported by Dr Jerome Lander - the Chief Investment Officer of WealthLander.

“It’s all about value,” he says. “People who are adding value to the advice and investment process, like portfolio managers, are not about to be replaced by AI. However, people working in other industries, like bank tellers, who couldn’t add value either probably have or soon will be replaced by machines.”

Jerome supports the view that disruption in financial services is relatively slow, but that doesn’t mean that when great technology comes along, even though you don’t understand in detail how it works, it can’t create value.

“The problem is, a lot of value creation is temporary in nature, so it becomes tricky to work out when the value creation stops. If you don’t understand how value is created, then it should be harder to invest in it. I often see fund managers who actually create value, but I have no idea when that value is going to stop,” Jerome says.

“There’s a lot of ways to manage money, whether that’s with AI or not. Investors focus too much on just the ways of managing money, rather than the effectiveness of the implementation and the alignment of interests. So, it’s important to not be too biased towards one way of managing money. There are many ways to manage money effectively. It all depends on the skills and the quality of the implementation of that money.”    

With the rise of smaller, more nimble companies disrupting many of the more embedded players in the market, could disruptive technology lead to the inevitable decline of these established companies?

“Nothing is ever guaranteed or automatic,” says David Wanis. “Probably 30-50 per cent of the companies in the small caps universe are different to what they were three years ago. There is always renewal and new companies coming through this sector. Compare that to the ASX 100, and that figure is about 5-10 per cent.

“Companies that innovate are centred to the idea of wanting to make a positive change. They create a team that is galvanised around innovation and the problem they are solving. It’s not their deliberate intention to kill off any companies or disrupt them. They are driven by creating a better way to meet unmet needs through technology or through developing a new business model.”

David confirms that Longwave Capital sees continuous renewal in the small caps space as creating both challenges and opportunities.

“Many of these small companies don’t make it, but for those that do, perhaps in 10 or 20 years time, they too, will end up in the ASX 100. In fact, these businesses become the new crop of large caps. And the compounding wealth you can generate from these companies, if you can identify and hold onto those shares, can be terrific.”

David adds: “We’re always on the lookout in this part of the market for these small cap opportunities. Longwave Capital maintains a positive view when seeking out these companies. It’s important for us to know what it is they want to achieve, and how they want to use innovation to create positive change.”   

Dr James Yang is a Junior Trader and Senior Data Scientist at Coolabah Capital Investments; Dr David Allen is Head of Short/Long Strategies at Plato Investment Management; David Wanis is Founding Partner, CIO and Portfolio Manager at Longwave Capital; and Dr Jerome Lander is the Chief Investment Officer at WealthLander.

They spoke in the IMAP Specialist Webinar Series on ‘Investing in Disruption’. The series was moderated by the Head of Research at SQM Research, Rob Da Silva.

.

Black box model: A black box financial model is a catch-all term used to describe a computer program designed to transform various data into useful investment strategies.

Booster trees: A booster tree is a process that combines a learning algorithm in a series to achieve a strong learner outcome from many sequentially connected weak learners. In case of gradient boosted decision trees algorithm, the weak learners are decision trees. Each tree attempts to minimise the errors of previous tree.

Deep learning: A a part of a broader family of machine learning methods based on artificial neural networks with representation learning.

Neural network: A neural network is a series of algorithms that endeavors to recognise underlying relationships in a set of data through a process that mimics the way the human brain operates. In this sense, neural networks refer to systems of neurons, either organic or artificial in nature.

Random forest: Random forest is a technique used in modelling predictions and behaviour analysis and is built on decision trees. It contains many decision trees representing a distinct instance of the classification of data input into the random forest. The random forest technique considers the instances individually, taking the one with the majority of votes as the selected prediction.

Regression: Regression is a statistical method that attempts to determine the strength and character of the relationship between one dependent variable and a series of other variables (known as independent variables). Regression helps managers value assets and understand the relationships between variables, such as commodity prices and the stocks of businesses dealing in those commodities.

Sources: Investopedia, CFI.

.