Alternative Data is a revolution. More important than AI! p.2/2Evaluating stocks
There are now several companies that estimate oil inventories and fuel production volumes. They analyze how much crude is in tanks (by measuring the shadow cast by the tank) and how much fuel is on the way (tankers, ship manifests) and in tanks on sidings.
With satellite imagery, analysts can also estimate aluminium inventories at key depots.
And soon, we will be able to go a step further - we will be able to predict inventories, their consumption and product demand very precisely. These predictions will be made possible by mapping supply chains. Of course, the biggest and the wealthiest companies with their equally impressive teams will be able to do this. Snippets of this information are already available (by subscription) from many companies specializing in commodity analysis.
Analysis of non-obvious sources of risk
Machine analysis of sector networks can be an exciting source of early information about sources of risk that we are unaware of.
The problems of one company or group, at first sight entirely unrelated to our positions, can have a negative impact on them. An analysis of past correlations in the sector may show this.
Our positions (which we may not be aware of) may be correlated with some events, keywords and topics that Machine Learning analysis systems may track.
Conversely, positive news about another industry can cause capital to flow out of sectors in which we have positions.
Capital seeks better opportunities and returns encumbered with less risk, so stock prices can fall even if everything is "fine" with our industry. These falls are caused by prospects competing with each other. The major funds are now building tools to track this sentiment and the capital flows that follow. Such signals provide a broader understanding of the market.
Collected opinions can be an interesting bias indicator (social media and news sentiment)
The online sentiment is the opinions expressed online (news stories, blogs, FB and Twitter posts).
Sentiment can be positive or negative and is best thought of as a continuum that we can visualize with a line.
Sentiment evolves; sometimes, it can precede market changes or changes in indicators.
Sentiment can be good at predicting changes before they occur. Such was the case, for example, with sentiment around oil in 2017. Positive opinions in April began to turn negative, followed by declines.
As expressed by analysts and professionals, sentiment is an interesting indicator that shouldn't necessarily be used directly but can be a signal that something is worth closer examination.
When a well-diversified portfolio turns out to be highly correlated
There is a particular class of events that strongly influence price movements. These are "black swan" events - where entire sectors and entire markets enter into deep declines. An example is a recent decline due to pandemics.
Similar events can also occur on a smaller scale. For example, we can have a very well-diversified portfolio that, under the influence of circumstances, suddenly behaves in a highly correlated way. These occurrences are usually the result of events that trigger strong panic moves.
Analyzing the environment of sectors, companies, or currencies can allow us to catch such events and take into account the unexpected behaviour of the portfolio.
The tone of the message and company documents used in risk assessment
One fund looks at the tone of the messages coming from the company. The tone can be positive, negative, or neutral and can contain interesting anomalies (an example: attempt to hide something).
It has been noted that readability and clarity of communications are correlated with better performance.
The tone is not a signal in itself but gives valuable context. For example, suppose the company's performance is acceptable. In that case, a positive tone can suggest that it will continue to be a bad tone - that something is wrong and traders need to be more careful.
For example, contextual data only becomes valuable with a tone. They can extend a position, and they can also reduce risk. It is worth taking an interest in which contextual data will help you in trading.
Analyzing the number and type of patent applications allows you to assess a company's future potential
Such analyses are not yet routine, but many funds are already conducting them. Analyzing the number of patent applications and their type can provide valuable information about a company's profit potential in the future.
The fact becomes evident if you think you have two very similar companies. One has a hundred patents, and the other has zero. Both operate in a field where innovation is critical. What impact might the patent structure have on the bottom line 3-5 years from now?
Data on the number of scientific publications in a company can have a similar impact, an interesting indicator of possible future profits.
Geopolitical risk analysis in sensitive regions
According to a prepared key, the system filters information from the media in local languages. It selects information that may indicate increasing tension in the region and possible crises.
The analyst does not even need to understand the language, such as Farsi used in Iran, to read a message on a scale (e.g., 1 to 10) and gather that something is wrong and the tensions are very high.
For those investing in oil, this can be of paramount importance. Consider the abrupt jump in price after the drone attack on Saudi Aramco's Abqaiq-Khurais refinery (Sept. 14, 2019)
This event was impossible to predict, but the information about rising tensions was available a few days earlier. We now know that such information can be used.
It does not mean the rising tensions should result in an exit from the position. However, this certainly requires a sensible action plan in case the black scenario comes true.
An analysis of company documents in combination with a study for the needs of regulators shows whether a company is stretching reality in the ESG area
Companies are beginning to see that the news of them implementing responsible business (ESG) policies positively impacts their image and usually boosts share price. But some companies only talk and do little.
One major fund analyzes company behaviour by comparing publicly reported data with documents sent to regulators. As a result, it avoids the stocks of those companies where there are significant discrepancies in these areas.
This fund is gearing up for a massive, systemic shift in its approach to investing and sees ESG as the future with the highest returns.
For thematic portfolios - a source of knowledge about the sentiment that can have an impact
Many funds build a thematic portfolio by investing in companies in a similar industry or sector. Such thematic portfolios are sensitive to sentiment related to a particular topic.
By studying web search trends around keywords, one can get an idea of changes, increases or decreases in popularity.
Also, the popularity of other searches can be a clue - capital interested in companies in one industry may flow away to something that offers better opportunities, despite positive sentiment around "our" topic.
Company employment data
The arrival or departure of key people can be a reasonably obvious signal about a company's situation. Growing companies hire, troubled companies lay off.
Knowing whom a company is looking for and whom it is laying off is interesting contextual information. It can even be a stand-alone signal in exceptional cases.
Summary
Alternative data helps you learn what makes up a signal, assisting traders in understanding which signals are better and weaker and why.
Signals are of higher precision, which in turn reduces the risk. Contextual information coming from the environment also reduces risk. Analysing the sentiment in the environment allows you to improve your signal returns: extend your position or exit earlier.
ADs themselves can also be a source of high-quality signals, a prelude to taking a closer interest in a company (currency, commodity) or abandoning an idea.
Anomaly-spotting tools allow us to quickly spot suspicious companies that are 'cooking the books' or misrepresenting their compliance with new policies or laws.
AD is on the rise, but only a small percentage is exploitable.
ADs are the foundation for the next revolution that is fast approaching. Artificial Intelligence will take full advantage of them, to an extent and with a speed that will leave today's traders and investors far behind. But all is not lost. The markets will change, but also new opportunities and new regularities will emerge. We will devote many articles to them in the future.
A few words on the evolution of alternative data in trading
What we see now is the beginning of a revolution. The amount of data will only increase. The scope of their use in trading will grow.
"Alternative" data will improve signal quality, extend or shorten a position, provide contextual information, reduce position and portfolio risk.
The revolution we are witnessing today is as important, if not more important, than the AI revolution. If someone does not take part in this race - they will systematically lose positions in the market.
If someone does not join the race to use alternative data - they will systematically lose market positions.
Think of your trading (single instrument or entire portfolio) as a decision-making process that takes you through from analysis to signal, position management and post-trade analysis. Identify the main steps in this process.
Think about what data would help you at each step of your decision-making process? What will improve your understanding of the situation around the instrument? What will better show the context?
What will improve the signal? What will help filter out weaker companies, currencies? What will show the elements of the situation that matter when you are in a position? What data can prolong the position? What news can shorten a position?
Post-trade analysis: what else worked, what did you not take into account? What data would still be worth having? Is it financially viable? Can you have data for half a million for a few thousand, perhaps?
Try to support your decision-making process at multiple important points using different types of data. This type of thinking will prevail in the future (in my opinion), and doing so will give you an advantage. Traditional portfolio managers focus too much on one type of data and one type of tool.
Think about what data will show the agility of a company at multiple points? I am referring to production, management, recruiting new people. What factors will allow evaluating the innovativeness of the company? What data will show interest in the company's products? For example, what are the trends among major customer groups? Do you know what the main groups that buy the company's products are?
What are the main groups of customers who buy the company's stock? What are the trends in their sentiment? Where do you get such data? Are you able to estimate them somehow? One effective way of estimating is to build yourself a network of contacts (conferences, networking, social gatherings) among the key customers of the companies and ask their opinion from time to time. It still works today and will be irreplaceable for a long time to come.
An exciting piece of information is the development plans of some large customer who orders many products from "our" company and is considering increasing orders. Think what can indicate such activity? What data can show it?
When thinking about sentiment analysis, think about sentiment trend analysis: whether it is increasing or decreasing, which is also valuable information. Try to find the sentiment trend and ask yourself what makes it rise or fall. Sentiment data will be used for that very purpose.
A separate topic is the new risk analysis, focusing on the sources within the company itself and in its environment. Again, alternative data will help us a lot in this. It is a topic on which tomes can be written today, and many will be written as new data becomes available.
When thinking about data, don't just think about one type; it can be confusing. You can estimate the same process in many ways, look for it and look for reliable yet cheap sources. Think about duplicating important data, look at a process through the prism of many different data; the best ones do that. This way, you reduce risk.
Over time, analyzing and forecasting the positions of major investor groups will become an additional interesting tool.
In short, the fight over the next few years will be for better instrument selection (e.g., companies). Over time it will shift to a battle for earlier entry and better signal utilization among the top companies. I will be following this evolution closely.
Evolution and dynamics are the keywords in the future. We are used to the fact that, for example, companies' data is given every quarter, and for this time, it is a static result.
Meanwhile, it is just a photograph of a more complex, dynamic process. It's like photographing children running - the photo itself is a picture of a very dynamic process. A lot is going on in and around the companies themselves. The world is speeding up; children are not only running but running faster and faster.
Today we are moving from static black and white photography to a colourful, dynamic film shot in 4k technology. In it, we have shots from the front, from the side and drones above. And once we get used to this new image, we never go back to the old one.
Our next article on alternative data will feature tips on getting the most out of AD and what strategy to adopt in the fund.
Give it a boost 🚀 and drop a comment so we know to publish more for you. Cheers!
Follow: www.tradingview.com
