{{title}}
{{textarea}}
{{task}}
{{select}}
{{team}}
{{input}}
{{solution}}
{{result}}
ICDS Insights Markets & Investors
Ambiguity Costs More Than Bad News
Price turbulence after a media peak follows how unresolved the media field is. The direction of the news adds little to it.
VM
Vadim Matyushkin Behavioral Scientist & Sociologist · ICDS
A bad headline is expected to hurt a stock, and a good one is expected to calm it. Two ICDS studies, run on different sectors and different countries, point to another ordering. The size of price swings around a media peak follows how clear the signal is. The sign of the news matters less.
We measured this twice. In both corpora a peak that leaves the public conversation divided is followed by larger price swings than a peak that settles it, whichever way it settles.
About the data
Article 3: 37 listed equities, more than 2,400 price records (open, high, low, close) over 24 months or longer, each peak classified by Balance of Expression, the split between positive and negative evaluative statements. Article 8: 53 public companies behind FMCG brands in 26 countries, a six-month window around each peak, tonality measured in the month before the peak and in the peak month.
Mixed Fields Draw Speculation
In the fintech corpus, peaks with a moderate split (about 60/40, positive or negative side) showed the highest speculative volatility in the peak month. Peaks with very strong dominance of one side (about 90/10 and above) showed the lowest. We read this as traders positioning ahead of the quarterly report when the field gives no clear verdict.
The quarterly report matters here. In 76% of observed events, months three and four after the peak coincided with an earnings release. Price movement in the first two months after a peak was on average 1.73 times more variable than the reaction after the report.
The Size of the Shift Predicts More Than the Level
In the FMCG corpus one variable held up under every robustness check: the tonality jump, the change in tone from the month before the peak to the peak month. A sharp, complete positive shift came with lower volatility (partial ρ = −0.337, p = 0.014, geography-robust). A moderate shift that leaves negative residue came with higher volatility. The market receives a clean signal in the first case and a contradictory one in the second.
| Predictor of volatility | Result |
|---|---|
| Tonality jump | ρ = −0.337, p = 0.014 |
| Balance of Expression | ρ = −0.214, p = 0.124 (same direction, n.s.) |
| Share of negative mentions at the peak | ρ = +0.038, p = 0.787 |
| Earnings cycle (positive vs. negative) | mean volatility 0.153 vs. 0.156 |
Fundamentals Explain Returns, Not Turbulence
Companies with a mostly positive earnings cycle gained +4.0% on average over six months after the peak. Companies with a mostly negative cycle lost −1.8%. Volatility was the same in both groups. A company with strong earnings and an unresolved media field saw about as much turbulence as one with weak earnings and the same field.
One configuration stood out. Where a negative earnings cycle met a large positive tonality jump, the average six-month change was −2.1% (13 companies), against +3.4% for all other companies. Positive messaging over weak results did not protect the stock, and the later correction was larger.
What Follows for Practice
1
Track the shift and its residue.
Mention volume and average tone show the level. The change in tone and the amount of negativity left after the peak show whether the field is resolved.
2
Treat turbulence as a communication task.
Improving the numbers does not lower it. Removing the ambiguity in the public narrative does.
3
Align communications with investor relations.
Positive brand messaging while negative earnings are pending raises the risk instead of lowering it.
What Is Established, and What Is Not
Established
The tonality jump predicts volatility on 53 companies and holds after controls for geography, peak type and earnings cycle. The 60/40 versus 90/10 pattern is shown on 37 equities in Article 3.
Established as a direction
Balance of Expression points the same way in the FMCG corpus but does not reach significance. Most FMCG brands are sub-brands of larger listed groups, so one brand’s peak is a small part of the parent company’s media field. In the brands that equal the listed company the pattern is stronger, and the sample is too small to test it.
Method limits
All relationships are correlational. The trader motive behind the mixed-field result is our interpretation.
Methodology note. Balance of Expression classes: balance 50/50 to 45/55, moderate disbalance about 60/40, strong about 75/25, very strong about 90/10. Tonality jump: change in the integrated tonality index from t-1 to t0. Volatility tests: Spearman partial correlation with geography controls, Mann-Whitney comparisons. Institute of Communication and Data Science. Working draft for internal discussion.
VM
About the author
Behavioral Scientist & Sociologist · ICDS
Vadim Matyushkin is a psychologist and sociologist with close to twenty years of research into digital behavior, trust, and information dynamics, and a researcher behind the Institute of Communication and Data Science (ICDS). His work has supported organizations including Coca-Cola, PepsiCo, Mars, Danone, Nestlé, and Bayer in moving from self-reported survey data toward direct behavioral evidence.
ICDS — Institute of Communication and Data Science is an independent research institute focused on understanding how trust, behavior, and reputation are formed in digital environments. ICDS operates as an intellectual institution — not an agency, not a platform, and not an educational provider.