Sentiment Investing – A Beginner’s Ultimate Guide

In this article / beginner-friendly guide, we’re going to explore what sentiment investing is. Specifically, we’ll look at:

• What is Sentiment Investing
• Types of Sentiment
• Sentiment Investing Strategies
• Implementing the Fervent 5 Step Sentiment Investing Strategy

Let’s get into it.

What is Sentiment Investing?

Firstly, what exactly is sentiment investing?

Broadly speaking, sentiment investing involves identifying and implementing investment strategies that leverage sentiment.

In the context of Finance and Investing, Sentiment itself can broadly be categorized into three main types, including:

• investor sentiment
• firm sentiment
• aggregate / macro sentiment

In other words, sentiment investing can involve leveraging sentiment at the investor, firm, or aggregate level.

The specific implementation of sentiment investing involves a 5 step process, including:

1. Determining the type of sentiment we’re interested in exploring (i.e., investor, firm, or aggregate sentiment)
2. Measuring sentiment
3. Identifying whether there’s a relationship between sentiment and investment securities (e.g., stocks)
4. Conditional on an affirmative point 3, testing whether an investment strategy can be profitable
5. Implementing the strategy (conditional on point 4 displaying feasible/promising results)

We’ll go over all steps in this article.

But first, let’s be clear on the different types or categories of sentiment within the context of finance and investing.

Types of Sentiment

Like we said earlier, we can broadly categorize sentiment at the investor, firm, or aggregate level.

Investor Sentiment

Investor sentiment attempts to quantify how investors, in general, feel about some factor within financial markets.

For example, one might attempt to quantity how investors feel about:

• the stock market as a whole
• an individual stock
• political climate
• perceived trust
• confidence in the economy

This isn’t an exhaustive list by any means, but hopefully, you get the key idea. It’s about quantifying how investors feel about some factors within financial markets.

We’ve written a detailed article on how to measure sentiment using a Document Term Matrix but keep in mind, that article’s written for a fairly technical audience.

In this article, we’ll try and keep things pretty broad.

---

Related Course: Investment Analysis with Natural Language Processing (NLP)

This Article features concepts that are covered extensively in our course on Investment Analysis with Natural Language Processing (NLP).

If you’re interested in learning how to leverage the power of text data for investment analysis while working with real world data, you should definitely check out the course.

---

Firm Sentiment

If we think about things at the company level, then sentiment can relate to:

• the firm’s portrayal of a certain sentiment, and
• the firm’s actual sentiment

Intuitively, a company that’s expecting hardships in the future may not be upfront about it to investors. Why start a panic, they might think.

Thus, even if a company is expecting hardships in the future, meaning the management feels anxious, for example, they might still portray themselves as feeling confident or excited about the future.

At the firm level of sentiment then, there’s the firm’s actual sentiment and the one that it portrays.

It’s considerably harder to measure a firm’s actual sentiment vis-a-vis the sentiment it portrays.

And this is less to do with the ability to measure it, and more to do with data availability. More on that further down, when we talk about measuring sentiment.

For now, let’s think about aggregate or macro-level sentiment.

Aggregate Sentiment

Aggregate sentiment reflects the level of sentiment in an economy as a whole.

For example, this might attempt to quantify how the population as a whole feels about, say, economic prospects in the country as a whole.

One might also argue that aggregate sentiment includes things like:

• consumer sentiment
• business confidence
• aggregate trust

Typically, these types of sentiment are quantified and then made publicly available in the form of an ‘index’.

Regardless of what one might mean by “aggregate” (e.g., how a population feels vs. how companies within a country feel), or how one measures it, the fundamental idea is that aggregate sentiment is attempting to quantify sentiment at a macro-level.

In other words, it’s not about a small group’s emotions (e.g., investors) but the collective whole instead.

Okay, now that you understand what sentiment investing is, and what the 3 broad types of sentiment are, let’s think about sentiment investing strategies.

Sentiment Investing Strategies

The ultimate goal of most investments is to earn a return or profit. Sentiment investing is no different.

The goal is still about earning a return or profit.

It’s just that the root to get there, or the “means to profit” if you will, is by using or leveraging sentiment in some way.

Sentiment investing strategies include:

• buying more shares when investor sentiment is rising (i.e., when there’s “bullish sentiment”)
• taking short positions when investor sentiment is plummeting (when there’s “bearish sentiment”)
• increasing (decreasing) positions in industries where business confidence is rising (falling)
• investing in stocks that display strong confidence vis-a-vis those that display weak confidence
• buying bonds when anxiety in the economy is increasing

This is just about scratching the surface and isn’t an exhaustive list of examples by any means.

The fundamental idea, however, is to base the investment decision on some form of sentiment.

In our Investment Analysis with Natural Language Processing (NLP) course, we teach you how to test and validate a sentiment investing hypothesis.

Specifically, we explore whether you could make a profitable investment strategy by investing in firms that display stronger net positive tone.

If you’re genuinely serious about learning sentiment investing rigorously, then do take a look at the course.

There’s no prior financial knowledge required, but you do need to know the basics of coding in Python.

Okay, now that you have some idea of what sentiment investing strategies look like, let’s dive deeper into the process for implementing such a strategy.

Implementing the Fervent 5 Step Sentiment Investing Strategy

Recall that earlier in this article, we said that implementing sentiment investing involves a 5 step process, including:

1. Determining the type of sentiment we’re interested in exploring (i.e., investor, firm, or aggregate sentiment)
2. Measuring sentiment
3. Identifying whether there’s a relationship between sentiment and investment securities (e.g., stocks)
4. Conditional on an affirmative point 3, testing whether an investment strategy can be profitable
5. Implementing the strategy (conditional on point 4 displaying feasible/promising results)

Let’s now consider each of these steps in more detail.

1. Determine the Type of Sentiment

Like we mentioned earlier, we can categorize sentiment into three broad types, including:

• investor sentiment
• firm sentiment
• aggregate / macro sentiment

The choice of which sentiment you want to leverage is very much a personal preference.

One can make a good case for each type, and the data will support each one, too.

Importantly, the three different types aren’t mutually exclusive. In other words, it’s not like you have to choose one or another.

You could combine different sentiment types into a single strategy, for example by:

• investing in firms that portray strong confidence during times of economic crises (assuming their portrayal is accurate/realistic)
• buying shares when investor sentiment is low but aggregate business confidence is high (assuming you’re in it for the long haul)
• shorting stocks that display overconfidence during times of economic crises (remember, though – shorting is incredibly dangerous)

In a nutshell, then, the choice of which type of sentiment you work with is very much a personal preference.

It’s also a function of what data’s available, and what you can get access to. This is something we’ll talk more about in the next step.

2. Measure Sentiment

Measuring sentiment can be as complicated or as simple as we like.

As we highlight in our sister article on the approaches of Sentiment Analysis in Finance, broadly speaking, we can measure sentiment using:

• lexicon/dictionary-based approach
• machine-learning approach

Both approaches have their own advantages and disadvantages. And since we’re trying to keep this particular article fairly broad and non-technical, we’re not going to dive too deep into the specifics of how to measure sentiment here.

That being said, if you’re interested in learning more, we recommend reading:

• An overview of Sentiment Analysis in Finance
• Estimating Sentiment for Sentiment Analysis
• NLP Applications in Finance

Now, recall that when we described firm sentiment towards the start of this article, we said that sentiment of the company can relate to:

• the firm’s portrayal of a certain sentiment, and
• the firm’s actual sentiment

We also said that it’s considerably harder to measure a firm’s actual sentiment vis-a-vis the sentiment it portrays.

And this is less to do with the ability to measure it, and more to do with data availability.

Actual vs. Portrayed Sentiment

In the context of companies, text data includes (for example):

• annual reports (e.g., 10-Ks or glossy PDF reports)
• conference calls / earnings calls
• interviews with senior management
• social media posts by companies

As with our previous lists, this is not an exhaustive/definitive list by any means.

Notice, though, that all of these forms of company-level text data are external. They’re all publicly available.

They’re not internal documents, internal emails, for example.

Thus, whatever we obtain or extract from these data sources, is almost certainly going to represent a company’s portrayal of sentiment more so than its actual sentiment.

Suppose you had access to private conversations with senior management to get “internal data”. It’s not like you can just use that in your investment strategies! Insider trading is illegal.

In a nutshell, it’s important to note that although there are different ways of measuring sentiment, and many sources of data, you want to be clear on what the data represents.

And on that note, it’s worth also looking at using something like a sentiment indicator or a sentiment index.

---

Related Course: Investment Analysis with Natural Language Processing (NLP)

This Article features concepts that are covered extensively in our course on Investment Analysis with Natural Language Processing (NLP).

If you’re interested in learning how to leverage the power of text data for investment analysis while working with real world data, you should definitely check out the course.

---

Using a Sentiment Indicator / Sentiment Index

Instead of going through the trouble of measuring sentiment yourself, you could use a sentiment indicator or a sentiment index.

Essentially, someone else has already gone through the trouble of estimating sentiment, then converted it or “packaged it” into an index, and made it available for everyone else.

Popular “sentiment” indices include:

• VIX (aka CBOE Volatility Index)
• Fear and Greed Index

There are others, but these largely rely on “technical analysis”.

The empirical research consistently shows that technical analysis doesn’t work for long-term economic gains in investing.

And we’re really not fans of technical analysis, either. So we’re not going to touch on the other “technical analysis” indices.

Importantly, although the VIX is often described as a “sentiment indicator”, it’s actually a volatility indicator.

The underlying assumption when describing it as a sentiment indicator is that investors’ anxiety, fear, and other negative sentiments increase as volatility increases.

It’s still important to note that, strictly speaking, the VIX is not a “sentiment” based indicator. Instead, it focuses on volatility.

3. Identify Relationships with Investment Securities

Once you have your estimate of sentiment, you can then link that to investment-related attributes.

The specific attributes can vary vastly including (but not limited to):

• profitability
• stock returns
• risk of stocks
• leverage/gearing
• cash flow management

Regardless of which specific attribute you choose to work with, it’s a case of seeing whether there’s any relationship between that attribute and sentiment.

Thus, you might want to see, for example:

• if firms that display “overconfidence” sentiment have greater stock risk
• whether stocks that display “underconfidence” have lower growth rates
• whether the future profitability of firms is related to its present optimism

If there’s no relationship whatsoever, then it’s unlikely you’ll be able to exploit anything to create a profitable investment strategy.

If on the other hand, there is some sort of relationship – even simple correlations – then there might be an opportunity to “earn alpha” (i.e., beat the market portfolio).

This is why we said point 4 is conditional on an affirmative point 3. Let’s now consider the fourth point.

4. Test the Hypothesis

Okay, we avoided the word “hypothesis” all this while because we didn’t want to scare off beginners.

In a nutshell, a hypothesis is just a formal expression of an idea or notion or “hunch” we might have.

Remember the examples of sentiment investing strategies we talked about earlier?

We said that sentiment investing strategies include:

• buying more shares when investor sentiment is rising
• taking short positions when investor sentiment is plummeting
• increasing (decreasing) positions in industries where business confidence is rising (falling)
• investing in stocks that display strong confidence vis-a-vis those that display weak confidence
• buying bonds when anxiety in the economy is increasing

Each one of these can be formally expressed as a hypothesis. Let’s take the fourth one as an example – we can hypothesise it as:

“Returns of firms with stronger confidence are statistically greater than returns of firms with weaker confidence”

See what we did there?

We’re just rephrasing the “idea” into a more formal, testable expression.

By that we mean, once you’ve got an estimate for “sentiment” (in this instance, for “confidence”), you can then calculate portfolio returns for two different portfolios:

• portfolio 1: comprises of stocks with “strong confidence”, and
• portfolio 2: comprises of stocks with “weak confidence”

If the hypothesis is valid, then we should see that portfolio 1 outperforms portfolio 2.

In other words, we should see that the returns of portfolio 1, on average, are (statistically) greater than the returns of portfolio 2.

This can be tested and validated statistically using a t-test, for example.

Now, suppose you test it and find that the hypothesis is indeed valid, you can then consider implementing the strategy in the “real world”.

5. Implement the strategy

After statistically testing and validating the hypothesis, you can consider implementing it in the real world.

Continuing on the example of the strategy revolving around the “confidence” of firms/stocks, sentiment investing strategies could include:

• long only: buying stocks that display strong confidence
• short only: shorting stocks that display weak confidence
• long short strategy: buying stocks that display strong confidence whilst shorting those that display weak confidence

Before you actually implement it, you’ll also want to consider other “real world” factors, including (but not limited to):

• transaction costs (of buying and/or shorting stocks)
• tax implications (depending on your tax status / tax bracket)
• impact on existing asset allocation strategies and investments (e.g., looking at the impact on your new portfolio beta)

Wrapping Up

Okay hopefully, all of this makes sense and you now know:

• what sentiment investing is
• the main types of sentiment are
• what sentiment investing strategies entail
• how you can implement the Fervent 5 Step Sentiment Investing Process

Like we said earlier… if you’re genuinely serious about sentiment investing, and want to explore how to implement it rigorously, check out our course linked below.

In fairness, if you know the basics of Python and want to deepen your investment analysis skills, you might want to also check out our Investment Analysis & Portfolio Management (with Python) course.

Or just make your life easier and get the All Access Pass for unlimited access to all our finance and investing courses 😉.

That’s it for this particular article. We hope you found it useful.

---

Related Course: Investment Analysis with Natural Language Processing (NLP)

Do you want to build a rigorous investment analysis system that leverages the power of text data with Python?

Explore the Course