Generating Alpha: Machine Learning Helps Traders

September 2nd, 2021 6 minutes read

What Is Machine Learning and How Does It Help Generate Alpha?

Machine learning (ML) is a way to train a computer system to identify trends using a given data set and autonomously learn from it. The system learns by analyzing large data samples and identifying common relationships and patterns. Using these, the machine can determine trends and decide if they should buy, sell, or hold assets.  

Maximizing the risk-adjusted return on any investment means you need to be able to accurately “predict” entry and exit.  The return associated with this residual but fundamental component is often called alpha. This component is the uncorrelated risk by what’s left after beta times market return is discounted. 

Why Investors Are Adopting Machine Learning: 3 Key Benefits

Prices alone cannot convey everything about an asset. In fact, most of the time, prices offer high level context, so you can’t treat them as investment signals. Instead, asset values are dictated by factors like an investor's opinion. This poses the question: are humans the only the ones capable of analyzing an investor's opinion? 

Believe it or not, today’s machine learning algorithms have become so refined that they often don’t need so-called “intuition” to gauge “opinion.” In fact, in some cases, you might even say that they’re objectively better at examining emotions because of their pattern recognition ability.

Most traders rely on their expertise, experience, and intuition to generate alpha. Now, it’s time to include machine learning in the equation, so you can more accurately predict market trends. Here are some of the benefits that you can expect to reap when you include machine learning in your arsenal of tools:

Benefit 1: Identifying Signals and Shifting Indicators

One of the cardinal rules of being a trader is knowing when to enter and when to exit. You can only do this by learning how to identify signals or shift indicators. While some are easy to spot, others have variables that the most experienced trader will have trouble identifying. Machine learning algorithms are typically able to do this through sentiment analysis.

What Is Sentiment Analysis?

Sentiment analysis is when a machine examines data (e.g., people’s posts, news stories, other forms of media) to determine the public’s opinion on a topic. Essentially, machines try to pick up the “emotional subtext” through words. They will identify which companies are liked, criticized, or boycotted by the general public at any given moment. This is an invaluable tool in generating alpha, because you can determine which assets you must trade for the short-term and which ones you should hold.

Benefit 2: Using Alt-Data Insights (Many Traders Won’t Check)

Your machine will have no problem crunching tremendous amounts of data, which means you can feed it videos, video transcriptions, photos, audio files, social media posts, presentations, webpages, articles, blogs, business documents, and more. With multiple algorithms at its disposal, the machine then uses these sources to further refine its determination of market sentiment. Through the machine, you get the sophisticated nuances of what people are saying, and more importantly, what they mean by what they say. 

Benefit 3: Maximizing Gains by Managing Risks

Alternative data is only helpful if a signal exists, and this is where you can combine machine learning with risk management techniques for better results. Maximizing your gain is all about reducing the risks involved in the volatile world of asset trading. Knowing when an asset price is due for a correction means you can take action before it even happens. With the right algorithms, you can predict positive and negative relative value  in seconds, helping you trade assets with reduced risk. More than upward and downward trends, though, machines can also estimate portfolio volatility, or even how much asset prices will fluctuate based on the type of data you ultimately feed the machine.

Should Traders Be Afraid Of Being Replaced By These Machines?

The short answer is “No.” As great as machines are at what they do, they can never completely replace you. To even begin to predict outcomes, computers will have to keep on relying on humans to feed them with relevant data. Here are three reasons why machines still need people:

Machines Are Unable to Define Their Own Tasks

Machines still need people to give them a designated purpose, at least for the foreseeable future. As the owner and operator, you still hold the keys. This means that as evolved as some systems are, machine learning is still susceptible to certain problems. It can still suffer from inaccuracies, discriminatory outcomes, and even embedded biases from the data set that you provide them.

Machines Have No Feelings 

Machines don’t have feelings or concepts of morality. These are traits exclusive to humans, and also means that people will still have to regulate machines. But for traders, while machines don’t get emotional or distressed, they can be combined with human out-of-the-box insights. You will still be responsible for the sustainable and ethical use of machine learning. Machines, for instance, may tell you to invest in a certain company, but your morals and those of your stakeholders may tell you otherwise.

If your firm focuses on impact investing and prohibits you from trading certain assets, you will still need to review what the machine tells you. And, until someone invents a reasonable algorithm for human ethics, investors like you don’t have to worry about being out of work.

Machines Cannot Predict the Future Completely 

Supervised learning relies on historical data and assumes that what has happened in the past is representative of what will happen in the future. This is faulty logic in times of political regime shifts, public relations scandals, natural disasters, and even worldwide pandemics. Inevitably, there will be things that algorithms can never predict. It’s important to note though that nobody can accurately predict these kinds of things. These flaws are shared by almost any other investment tool and technique. Your human insight and analysis into the situation will be needed.

Should You Start Integrating Machine Learning in Your Investment Strategy? 

No one can deny that using machine learning to generate alpha has its merits. While it has its problems, so do other investing tools. But the field is still growing, and it’s growing fast. The benefits of machine learning could very well double or even triple in a few years’ time.

Right now, we have millions and millions of data points that we can use to analyze people’s behavior, and this number will only keep growing. At the moment, banks and institutions, such as fintech startups, are 10 to 100 times better at predicting consumer behavior and creditor behavior, among others, than every theory that ever has been devised by financial professors.

Start incorporating machine learning in your investment strategy and start generating higher alpha today. Artificial intelligence can boost your analytical and decision-making abilities by providing the right information at the right time. Machine learning will only make the process of gathering information so much easier, and will give you the tools you need to make better investment decisions.


To learn more about how BAM can help you with assessing financial risk using our no-code proprietary machine learning tools, visit We help risk and investment professionals, such as yourself, filter through the noise through concise and actionable signals. Our goal is to help you make better investment decisions. 


Geraldo Filgueiras, Founder and CEO of (, @gerafilg)

Related Blogs
Fresh and Unique Alpha
Identification and validation of Alpha is the most important element that distinguishes portfolio managers. When seeking for Alpha, one should be aware to see the difference between market expectations and their own expectations of Alpha. Thus, it is important to implement processes that will continuously challenge and validate a professional's strategies and hypothesis.

August 2nd, 2021

Backtesting in the Age of Financial Machine Learning
It is predicted that by the end of 2025, we will have 175 ZB of data. This data could be very beneficial in grasping the right trading signals, and Alpha, if filtered correctly. Thus, it is important to implement backtesting as a strategy of filtering Alpha. However, when doing so, one should take into consideration the traps of backtesting and other limitations.

October 2nd, 2021