AI In Asset Allocation: Smart Tech For Smart Money

I Know First Research Team LogoThis article was written by the I Know First Research Team.

The financial industry as we know it – or knew it, at least, – is going through a transformation. Wolves of Wall Street, clad in their bespoke suits worth the budget of a small country, are giving up some space to the cohort of jeans-wearing millennials. And this new crowd brings more than a change in fashion. The generation of digital natives has a taste for all things high-tech and the know-how and savvy to make the reality live up to their taste.

(Source: Pixabay.com)

Artificial Intelligence and machine learning, two of the hottest buzzwords of today, play a major role in this process. This technology lies beneath many of the disruptions shaking the world of finance. These include AI-based robo-advisors that human asset managers now have to compete with, the rise of AI-driven quantitative hedge funds and various other AI-based fintech initiatives.

But this transformation is more than finding new ways to do old things. As the industry is transformed, so are its dogmas, theoretical pillars and conventional wisdoms. Value investment, for example, has long been seen as one of the best, if not the best, strategies for picking stocks for your portfolio. And yet, AIs, with their ability to process huge troves of data in the blink of an eye, offer an alternative to this venerable strategy.

Roughly the same thing, it can be argued, is happening with another key concept – that of asset allocation.

Weighting Risks And Biscuits

At its core, the idea behind asset allocation is quite simple. Asset allocation is the process of allocating the investor’s capital towards specific types of assets (most typically, stocks, bonds and cash) in a way that maximizes the returns and minimizes the risk. Or, to be more precise, allocating it in a proportion that is supposed to either deliver the returns at an expected level or stay below a certain risk threshold. These two, risk and reward, are what is essentially being juggled here. They are, of course, intertwined, and the idea is that you cannot have one without the other.

(Source: Pixabay.com)

After developing a strategy and investing the funds along the lines of the aforementioned proportions, you re-balance the portfolio every now and then as the distribution of value among assets of different types shifts. There can be different views on when exactly that should happen, but the conventional wisdom says that if one of the asset types has drifted 5% or more from its designated value, it’s re-balancing o’clock.

Things can get even less straightforward when you take into account the client’s goals and time horizons. A young amateur private investor long to make profit on their money instead of storing it in a bank and a high-earner bracing for retirement will pick two very different models. The same goes for a couple setting up a fund to make sure their kids do not have to worry about student loans. All these people will be looking at different time horizons, different risk thresholds and different expected returns.

Nevertheless, despite all this, there are still some one-size-fits-all solutions ready for those who are not looking to get particularly sophisticated. The share of stocks in your portfolio, says another conventional wisdom, should be equal to 100 minus your age. The idea behind this is to get more conservative in your investment as you grow older, so as to defend the assets that is already there.

Fair enough, you would probably say, so far, so good, but where does AI come into play here? Not in figuring out how many years you have left until your 100th birthday, that is for sure – here, a simple calculator is suffice. But in other aspects of asset allocation, AI can be quite a helper.

Weight The Signals: AI In Asset Allocation

As we can tell from the previous section, in its essence, all about optimizing the risk to returns ratio in line with the clients’ needs and wants. For AIs, this opens a variety of ways to step in and be of assistance.

First, an AI trained on a database of existing clients would be able to gauge the new customers, check their profiles against the database of the existing ones and suggest a portfolio in line with what other investors of a similar profile are using. The idea behind this is the same as the design of the AI that services like Netflix use to suggest shows that their subscribers could be interested in. This would allow for a nuanced classification of clients and a more sophisticated approach to each of them.

Even more interesting, however, is the idea of using AIs past the point where you have established the proportion of funds to go into each asset type. In other words, at the stage where you are looking for concrete investment opportunities. Here, AIs can be used in the same way high-tech quant hedge funds use them: to scrape the unstructured data from all over the Web to try to gauge the outlook of a specific company or even investment universe. This method offers a solid decision-making enhancement tool to any investment manager.

As promising as both options are, neither seems to take a step further and change the fundamentals of how asset allocation is done. This, however, is still possible using other AIs – those that model and predict the market dynamics, like the one trained and run by the Israel-based company called I Know First.

(Source: Iknowfirst.com)

The algorithm’s output is presented as a heatmap with two indicators – signal and predictability. Signal stands for the overall performance of the asset against others on the forecast, while predictability demonstrates how well the algorithm has been able to predict the stock before. Thus, top investment opportunities for the can be identified by picking out the stocks with the highest values for both indicators. From there, since both indicators are numeric, their values can be transformed into weights that would define how much of the funds allocated for this asset type goes to this specific financial tool. The whole process, of course, can be fully automated. When the company ran a trial of the kind, the portfolio run by its algorithm demonstrated some very impressive returns.

Another promising method has been developed by I Know First for one of its clients, a major asset management company from Japan. Here, the company has identified a number of investments baskets, including stocks, ETFs, bonds currencies and emerging currencies. After looking at I Know First forecasts for the assets in these groups, they calculate the signal to predictability ratios for them and use those to allocate their funds in a way that promises the highest returns.

This logic lies at the foundation of a strategy known as tactical asset allocation, in which the investors shifts the percent of funds allocated to different asset sectors on the portfolio to make use of the most promising options. This can be done on the basis of sector/region rotation, where the funds are rebalanced on the basis of prioritizing the most promising sectors in the best-performing geographic regions. Here, the I Know First algorithm can provide the clients with bottom-up forecasting across sectors, aggregating the predictions for individual stocks to provide an overview of a whole group they belong to, market by market.  Identifying the assets in the best-performing region that are most likely to show the strongest movement in the same direction where the sector they are part of is predicted to move and capitalizing on those can make for a viable investment strategy, as discussed in more depth here.

This approach is also indicative of how AI-driven tools can assist those who prefer risk-on, risk-off trading. The idea behind this approach is essentially to keep your finger on the pulse of the market, investing aggressively in a low-risk environment and going more conservative when the risks are high. I Know First Ai helps investors do exactly that, allowing them not just to follow individual assets, but also get a more general feeling of where the market is heading, and whether volatility is to be expected.

With elements of chaos theory, deep learning and genetic algorithms incorporated in its design, the I Know First AI collects new quantitative market data every day. This data is checked against a huge historic database covering 15 years of trading to seek out the interplay between the value of different assets and any similar patterns. After each new prediction delivered, the algorithm goes through a learning cycle and reconfigures its models based on its past successes and failures. This ensures that the accuracy of its predictions goes up with every iteration.