Balance (or disbalance) of Expression — an indicator of the ratio of positive to negative expressions within a single brand mention peak. When these proportions are nearly equal, we speak of balance. When they differ noticeably, this constitutes a disbalance.

Balance is defined as a ratio ranging from 50/50 to 45/55 — where one polarity holds only a minor, barely perceptible advantage (absolute difference between positive and negative expressions equal to up to 10%).

Three types of disbalance are identified:

• Moderate (+/-) Disbalance: e.g., 40 vs 60% (absolute difference between positive and negative share ranging between 11% and 40%)
• Strong (+/-) Disbalance: e.g., 25 vs 75% (absolute difference between positive and negative share ranging between 41% and 70%)
• Very Strong (+/-) Disbalance: e.g., 10 vs 90% (absolute difference between positive and negative share greater than 70%)

In our two previous articles Social Media Activity and Consumer Trends and Brand Media Success, we examined in detail the relationship between brand mention peaks on social media and users’ search and purchasing behavior.

We analyzed over 2 million mentions of 39 global brands across a 24-month period (December 2023 – November 2025). In addition, we collected search query statistics for this same period.

The influence of media peaks on search and purchasing activity was not only established but also examined in terms of variation across different product categories, as well as how peaks of varying composition can trigger behavioral patterns of differing intensity.

The next direction of our research was to examine stock price dynamics resulting from sharp surges in brand mentions on social media. Before turning to the data itself, however, we must first understand how this market operates and what different groups of investors pay attention to.

The Basis for Stock Buying and Selling Decisions

Stock market participants can be categorized by their speed of reaction and the type of information they rely upon. In professional circles, this is referred to as the spectrum of market efficiency, or trader and investor types by planning horizon.

The fastest actors are “Algorithmic Traders” (also known as “Noise Traders”) — their reaction to any given media event is near-instantaneous.

One of the characteristic strategies of such traders is: “Buy the rumor, sell the news”. The logic is as follows: when a rumor or expectation of a positive event emerges, buy early; then observe as the price rises. When the news is confirmed, early players lock in profits and sell.

To identify such trading signals, algorithms scan the internet (news feeds, Twitter (X), Reddit, Telegram). Like our own research, they search for anomalies and spikes in discussion volume, and also calculate: sentiment (the ratio of positive, negative, and neutral mentions), emotional intensity, and the dynamics of positive-to-negative shifts — the so-called social volume acceleration: “yesterday: 100 mentions → today: 5,000”

Beyond quantitative signals, algorithms also rely on non-numerical triggers through event detection. They flag keywords such as: : “lawsuit”, “hack”, "searches", "boycott", “CEO resigned”, “product recall”, “factory exploded” and others.

However, the stock market also comprises other categories of traders, for example:

- “Day Traders,” who react to the onset of price movement, often taking cues from the behavior of the first type of players.

- “Medium-Term Speculators” (Swing Traders). They may wait a few days until the emotions of the first two categories subside, entering a position at a lower price in anticipation of a rebound, or conversely selling if they determine that the news is genuinely bad.

— “Fundamental Investors” — this type of participant is the most conservative. They will not engage in spontaneous or medium-term plays. Their strategy is to await the quarterly earnings report, which will put everything in its proper place.

The Core Research Hypothesis

Accounting for the full diversity of strategies and market participant types, we hypothesized that brand mention peaks on social media may produce a dual effect on stock price movements. The first effect concerns speculative trading, which may commence within the same month and persist for some time. The second effect is long-term and indirect: when a media peak positively influences transactional search queries and, consequently, product sales, there is a probability that the company’s quarterly earnings report will also prove positive — which in turn may contribute to the stabilization of previous significant price adjustments or even to stock price growth several months later.

For this reason, it was decided to extend the observation window to a six-month period following the mention peak.

It was also decided to introduce a variable called Balance of Expression. This variable enabled a methodologically rigorous workflow in which a specific sentiment direction can only be assigned when one sentiment sufficiently outweighs the other, ensuring that positive or negative assessments reflect genuine dominance rather than marginal prevalence. Ultimately, the variable answers the question of how divergent the share of expressive statements within a peak is (do users express one dominant opinion, or are their views ambiguous?) and is calculated as the absolute value of the difference between the share of positive expressions and the share of negative expressions.

Through an information loss test (examining the correlation between the factorized balance of expression types and their numerical values) and the overall distribution shape, it was determined that a granular segmentation focused on the lower end of the range was required.

Balance is defined as a ratio ranging from 50/50 to 45/55 — where one polarity holds only a minor, barely perceptible advantage (absolute difference between positive and negative expressions equal to up to 10%). Three types of disbalance are also identified:

Moderate (+/-) Disbalance: e.g., 40 vs 60% (absolute difference between positive and negative share ranging between 11% and 40%)

Strong (+/-) Disbalance: e.g., 25 vs 75% (absolute difference between positive and negative share ranging between 41% and 70%)

Very Strong (+/-) Disbalance: e.g., 10 vs 90% (absolute difference between positive and negative share greater than 70%)

Sample

We analyzed 37 publicly traded stocks, comprising over 2,400 records of high, low, opening, and closing price movements across a period of 24 months or more.

The sample included the following product categories: cosmetics — 5, accessories — 5, clothing — 4, services — 3, electronics — 3, food — 2, entertainment — 2, travel — 1. We also added 12 fintech companies. This sector is of particular interest for studying stock price dynamics given the heightened investor attention toward these companies’ data.

General Findings

Some of the findings obtained serve to confirm the very fact of media peak influence on price dynamics. In 80% of cases, the opening and closing price dynamics in the month of the peak exceeded the average dynamics over the full two-year observation cycle.

It was also found that, in general, negative news exerts a stronger immediate impact on price adjustments than positive news. However, all of this pertains to the immediate market reaction — specifically the buying and selling that affects price movements within the month of the media peak.

As stated earlier, we tracked price dynamics over a six-month period and found that 76% of the events we observed in the third and fourth months after peaks coincide with quarterly earnings reports. What became immediately apparent is that price dynamics in the period immediately following brand mention peaks (within the first two months) are on average far more volatile than the reaction following the reporting period. On average, speculative price adjustment is 1.73 times greater than the price dynamics observed after the quarterly earnings report.

In all charts presented below, price behavior following such reports is highlighted separately. This allows us to visually distinguish the price impact of speculative market participants from that of conservative ones.

The Relationship Between disbalance of expression and Stock Price Adjustment

The influence of mention peaks on stock prices varies depending on the type of disbalance:

It is important to note how the severity of price adjustment declines following the publication of the earnings report in all cases except the last.

Why is pre-report price adjustment lower the more pronounced the dominance of one sentiment polarity in brand mentions? In our view, this corresponds to one of the algorithmic front-running strategies, where stocks are bought on “rumors” or signals not yet apparent to others, but sold once the “news” is confirmed.

Moderate Disbalance - — regardless of whether positive or negative sentiment dominates — drives stock purchases in the month of the brand mention peak. Stocks are not being sold; rather, they are bought in anticipation of price growth. And this growth during period t0 is the highest observed. It is as though such a balance is perceived as ambiguous and mixed — which serves as a signal for traders to play long while the majority has not yet determined which direction the trend is heading: toward success or failure.

The Relationship Between Dominant Mention Sentiment and Stock Price Dynamics

We selected exclusively those companies for which earnings data was available, and first examined how stocks behave when the report showed a decline in profits. For all companies, the price drop following earnings publication is evident. However, what interests us more is this: we observe that early traders may sense poor financial performance, but will more often play short when negative sentiment does not yet dominate the overall body of mentions to the point of overwhelming everything else. The game plays out better on ambiguous information terrain. Moreover, following their initial sell-off, a buy at the bottom — and then another sell at the peak — likely follows. This makes the post-earnings correction all the more dramatic, with extremely low brand value metrics in the moment.

Let us now examine the pre- and post-peak behavior of traders regarding the stocks of companies that ultimately reported positive earnings. It appears that here too, early traders somehow knew or sensed how the company was performing. They buy stocks, with the exception of cases falling into Strong negative Disbalance.

What is most important for us to identify here is how apparent the efforts of market players are on ambiguous information terrain, and how frequently these efforts lead to high price adjustment — which in turn increases risk for the company.

To test the hypothesis on positive disbalance, let us examine the corresponding charts:

Here we observe a reduction in price adjustment when positive sentiment reaches extreme dominance over negative. Meanwhile, a more mixed picture provides fertile ground for speculative play. The same pattern is seen in the 55% to 45% ratio: in the first month, a modest but discernible upward price movement is observed, followed by subsequent waves and a rebound in the month of the earnings report.

Stock market adjustment unambiguously leads to an increase in perceived investor risk. And the higher the risk, the higher the return investors demand. As a consequence, it becomes more difficult for companies to attract capital, and prospective investors require a discount.

Even when a company is fundamentally sound, the market interprets price swings as signs of business trouble. During periods of high price adjustment, there is more “emotional” selling and price drawdowns deeper than they “should” be. High in severity price adjustment can drive away institutional investors — precisely those conservative investors with low risk-sensitivity thresholds.

If a company is prepared to exploit the price adjustment itself — issuing shares at the peak — then such a strategy may be a deliberate one. But in most cases, this kind of dynamic causes harm.

We found that price adjustment is not strongly driven by credible rumors of declining business performance, nor by sales and marketing activity in general. We did not observe a consistent or widespread influence of these factors. However, a seemingly minor factor — media environment ambiguity — turns out to be a high-risk factor for companies whose shares are publicly traded.

Even an unambiguously negative information landscape may not be compelling enough for market players to stage interventions and destabilize the market. The same applies to an unambiguously positive one (when positive sentiment accounts for 95% or more of all evaluative mentions). In summary, it can be stated with a high degree of probability that in order to protect a company from financial risks, it is necessary to combat negativity within the overall body of brand mentions more aggressively than may be conventionally accepted as sufficient for marketing purposes. Ideally, it should be absent altogether — so as not to provide grounds for stock market speculation and emotionally driven selling. And, by extension, to ensure the long-term financial stability of the business.