Authors

  • Danata Mometova
    ВМА 76 R

DOI:

https://doi.org/10.71337/inlibrary.uz.ijai.136120

Keywords:

X

Abstract

This paper examines how misinformation, rumors, and fabricated news stories influence modern financial markets. The study builds on theories of information asymmetry, behavioral finance, and market microstructure to show that false information can trigger significant volatility, distort investor behavior, and erode trust in financial systems. Using recent global cases such as the Associated Press Twitter hack in 2013, the GameStop and AMC short squeeze episodes of 2021, the collapse of the Terra–Luna cryptocurrency ecosystem, and rumors that contributed to the Silicon Valley Bank crisis in 2023, the paper illustrates how digital platforms amplify misinformation. Special attention is also paid to emerging markets, including Uzbekistan, where Telegram channels and social media commentary often affect perceptions of currency stability and bank liquidity. The paper concludes with policy recommendations, emphasizing the need for stronger regulatory frameworks, improved market surveillance, and digital literacy programs for investors.

background image

INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE

ISSN: 2692-5206, Impact Factor: 12,23

American Academic publishers, volume 05, issue 09, 2025

Journal:

https://www.academicpublishers.org/journals/index.php/ijai

page 250

«FAKE NEWS AND FINANCIAL MARKETS: MECHANISMS, EVIDENCE, AND POLICY

IMPLICATIONS»

Mometova Danata

ВМА 76 R

Abstract

This paper examines how misinformation, rumors, and fabricated news stories influence modern
financial markets. The study builds on theories of information asymmetry, behavioral finance, and
market microstructure to show that false information can trigger significant volatility, distort
investor behavior, and erode trust in financial systems. Using recent global cases such as the
Associated Press Twitter hack in 2013, the GameStop and AMC short squeeze episodes of 2021,
the collapse of the Terra–Luna cryptocurrency ecosystem, and rumors that contributed to the
Silicon Valley Bank crisis in 2023, the paper illustrates how digital platforms amplify
misinformation. Special attention is also paid to emerging markets, including Uzbekistan, where
Telegram channels and social media commentary often affect perceptions of currency stability and
bank liquidity. The paper concludes with policy recommendations, emphasizing the need for
stronger regulatory frameworks, improved market surveillance, and digital literacy programs for
investors.

Introduction


In recent years, financial markets have become increasingly vulnerable to fake news. Traditionally,
information was filtered through newspapers, television, and specialized financial publications.
Today, however, platforms like Twitter (now X), Reddit, Telegram, and TikTok allow any
individual to publish statements that can instantly reach millions of investors. This shift has
lowered the barriers for manipulation. False or misleading information does not need to be
carefully designed by insiders; it can be as simple as a viral post or an anonymous blog entry.
The implications are profound. Market efficiency relies on the assumption that information
available to investors is accurate and equally distributed. Yet, misinformation undermines this
assumption, creating short-term dislocations in asset prices, influencing order flow, and shaping
investor sentiment. In volatile environments, even a single post can trigger billions of dollars in
market value losses.
Theoretical foundations: from information asymmetry to behavioral amplification

Efficient markets, qualified.

In the strict EMH, public misinformation should be quickly

arbitraged away (Fama, 1970). In reality,

asymmetric access and reaction speeds

create temporary

wedges: not everyone sees the correction at once, and some agents (news-algos, HFT) react far
faster than retail.

Behavioral channels.

Empirically, media tone and salience shift risk premia and trading

intensity (Tetlock, 2007). Heuristics—availability, anchoring, herd dynamics—convert
ambiguous headlines into coordinated order flow, especially under uncertainty.


background image

INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE

ISSN: 2692-5206, Impact Factor: 12,23

American Academic publishers, volume 05, issue 09, 2025

Journal:

https://www.academicpublishers.org/journals/index.php/ijai

page 251

Related literature

Media tone → returns & volume.

Tetlock (Journal of Finance) shows that pessimistic language

in daily news predicts short-run market declines and volatility upticks—establishing a quant link
between text tone and prices.

Volatility dynamics.

GARCH (Bollerslev, 1986) remains the workhorse to quantify

conditional volatility spikes after news shocks

Propagation on platforms.

Falsehood diffuses faster than truth on social networks

(Vosoughi, Roy & Aral,

Science

), a structural reason misinformation shocks are sharp and self-

reinforcing.

Social media & runs.

During the 2023 U.S. banking turmoil, research documents that

Twitter activity accelerated withdrawals at Silicon Valley Bank—illustrating macro-relevance of
platform dynamics.

Measurement: from raw text to “information pressure”

A practical pipeline for detecting and quantifying misinformation-risk around tickers:

Data & signals

Prices/volume/quotes:

consolidated feeds or exchange prints (minute-bar granularity

minimum).

News & social:

full-text vendor feeds (e.g., Bloomberg News Sentiment “BNS”), plus

social-media firehoses where permitted; include

timestamps and source provenance

.

Third-party

analytics:

Refinitiv

MarketPsych

provides

real-time

sentiment/emotion/scarcity metrics across assets—useful as exogenous signals in VAR.

Text features

Polarity/subjectivity (VADER/TextBlob as baselines) and domain-adapted transformers

(FinBERT) for finance-specific tone; flag

asserted facts

vs

opinion/forecast

; detect

low-

provenance claims

(no source, “insider says”, synthetic images). (See also recent LLM-in-finance

sentiment surveys.)

Composite “Information Pressure Index (IPI)”

Roll up (i) negative/uncertain tone, (ii) novelty, (iii) virality (retweet/repost velocity, cross-

platform synchrony), and (iv) provenance penalties (anonymous, newly created handles). Use
exponentially weighted aggregation and z-scores relative to ticker baselines.


background image

INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE

ISSN: 2692-5206, Impact Factor: 12,23

American Academic publishers, volume 05, issue 09, 2025

Journal:

https://www.academicpublishers.org/journals/index.php/ijai

page 252


Econometric toolkit (with templates)

GARCH(1,1)

: model the conditional variance around events to estimate the post-shock

volatility multiplier. (Bollerslev, 1986).

VAR

: treat prices, order imbalance, and IPI as an interacting system; impulse-response

functions (IRFs) reveal how a misinformation shock propagates and decays. (Sims-style VAR).

DiD

: compare “treated” tickers exposed to a misinformation burst vs. matched controls,

before/after the event, to estimate abnormal returns/volatility while controlling for market-wide
moves (Card & Krueger for DiD canon).


Event studies: documented U.S./U.K. cases

1.

AP-Twitter hack (2013)

— S&P futures lurched within seconds after a fake tweet about

White House explosions; prices normalized as the hack was identified, illustrating extreme
sensitivity of latency-aware strategies to breaking headlines.

2.

F-35/Lockheed Martin tweet (Dec. 12, 2016)

— A single tweet by the U.S. president-

elect that the F-35 program was “out of control” knocked billions off LMT’s market value intraday;
defense peers also wobbled. Bloomberg, The Guardian, and other major outlets documented the
immediate drawdown.

3.

“Short-and-distort” (Farmland Partners, 2018)

— An anonymous Seeking Alpha post

alleging questionable loans triggered a >20–40% slide; the company later sued and a jury found
defamation on key claims. Illustrates how

plausible-sounding but incorrect

allegations can reprice

small/mid caps with thin float.

4.

Crypto headlines and policy repetition (2021)

— Re-circulated headlines on China’s

longstanding crypto restrictions coincided with sharp BTC drawdowns; the

re-statement

effect

matters when audiences treat old news as new—something both newsroom practice and
algorithmic curation can exacerbate.


background image

INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE

ISSN: 2692-5206, Impact Factor: 12,23

American Academic publishers, volume 05, issue 09, 2025

Journal:

https://www.academicpublishers.org/journals/index.php/ijai

page 253

5.

SVB (Mar. 2023)

— Social-media velocity accelerated deposit flight dynamics during the

bank’s failure episode; in markets, financials saw gap moves and elevated intraday vol as rumors
and commentary proliferated.

Stylized empirical results you can expect (replicable blueprint)

Volatility multiplier (GARCH):

conditional variance rises by ~40–60% (event-window

average) vs. pre-event baseline, decaying with half-life on the order of tens of minutes but
sometimes persisting days when narratives linger. (Method anchored in Bollerslev 1986).

Abnormal returns (DiD):

treated names underperform matched peers by ~1–2% over 30–

120 minutes when the misinformation is

negative

and plausibly firm-specific; dispersion widens

as retail share of volume rises. (DiD framework per Card-Krueger).

Dynamics (VAR/IRFs):

a 1σ shock to IPI raises order-imbalance and widens spreads;

price impact peaks quickly and mean-reverts as refutation circulates, consistent with Tetlock-style
tone effects and “limits to arbitrage”.
Microstructure mechanisms: why prices move “too much”

Parsing at the speed of code.

Headline-reading algos trade before humans verify, creating

initial overshoots.

Inventory & limits to arbitrage.

Dealers widen quotes; capital-constrained arbitrageurs

cannot instantly offset one-sided flow.

Believability cues.

Official-looking branding, “insider” framing, or technical jargon

increases perceived credibility and delays skepticism—precisely why some manipulative posts
mimic press releases or lab memos.

Echo & re-statement.

Old or partial stories recirculate as “new,” driving multiple mini-

shocks (documented frequently in crypto and single-stock news cycles


Governance, enforcement, and platform policy (U.S./U.K. focus)

U.S. SEC enforcement.

The SEC has charged and/or settled with celebrities and

influencers who touted crypto assets without proper disclosures (e.g., Kim Kardashian, 2022; a
group of celebrities in 2023 in the TRX/BTTC matter). Even “unlicensed” voices fall under
promotion rules if they’re making security/asset promotions.


background image

INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE

ISSN: 2692-5206, Impact Factor: 12,23

American Academic publishers, volume 05, issue 09, 2025

Journal:

https://www.academicpublishers.org/journals/index.php/ijai

page 254

U.K. FCA “finfluencer” regime.

Tightened social-media promotion rules require

authorized approval or exemption and ban “click-bait” style promotions that could mislead;
guidance explicitly targets the financial-promotions lifecycle on platforms.

Platform analytics.

Market data vendors productize news-sentiment (e.g., Bloomberg

BNS) and behavioral feeds (Refinitiv MarketPsych), enabling surveillance teams to set thresholds,
alerts, and anomaly detectors in real time.
Emerging Market Perspective: Uzbekistan


In Uzbekistan, social media plays a growing role in shaping financial perceptions. Telegram
channels and anonymous posts often speculate about currency exchange rates or bank liquidity.
For example, during periods of global economic uncertainty, rumors about shortages of U.S.
dollars in local banks have circulated widely online, prompting long queues at exchange offices.
While such rumors are often exaggerated, they influence behavior in real time.
The country’s stock exchange is relatively small, but investor psychology is still shaped by
narratives. In the future, as Uzbekistan’s capital markets deepen, the influence of fake news could
become even more pronounced. Understanding this risk is vital for regulators and policymakers.
Practitioner playbook: detecting and trading around misinformation

For risk & surveillance

Maintain a

source-credibility registry

(domain age, historical accuracy, disclosure

practices).

Alert on (i) unusually

negative/uncertain

tone, (ii)

first-time source

for a ticker, (iii)

cross-

platform synchronization

within short windows, and (iv)

image/video artifacts

(synthetic

detection).

Couple IPI spikes with

market-quality metrics

(spreads, depth, cancel-replace ratios) to

trigger circuit-breakers in algo routing.

For systematic portfolios

Treat information shocks as a

state variable

: condition intraday risk budgets and

participation rates on IPI; widen “do-not-trade” bands during unverifiable bursts; fade only when
(a) reputable refutation arrives or (b) order-flow imbalances normalize.

For IR/Compliance teams at issuers

Pre-authorize rapid response channels (verified accounts, press pages with cryptographic

timestamps).

Publish

machine-readable

corrections (RSS/Atom/structured JSON) so data vendors

propagate refutations as fast as the original rumor.


background image

INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE

ISSN: 2692-5206, Impact Factor: 12,23

American Academic publishers, volume 05, issue 09, 2025

Journal:

https://www.academicpublishers.org/journals/index.php/ijai

page 255


Replication guide (concise)

1.

Sample.

Identify misinformation events via (a) platform takedown logs, (b) vendor tags

(BNS “Rumor/Unverified”), (c) fact-checking archives.
2.

Windows.

[-60, +240] minutes around timestamp

t₀

.

3.

Outcomes.

Mid-quote returns, realized variance, bid-ask spread, order imbalance, depth.

4.

Models.

o

GARCH(1,1) on 1-min returns to estimate post-event variance uplift (Bollerslev).

VAR([prices, IPI, imbalance]) with IRFs (Sims).

o

DiD on abnormal returns vs. matched peers/ETFs (Card-Krueger).

5.

Robustness.

(i) Drop days with scheduled macro releases, (ii) placebo on

old

news

resurfacing, (iii) jackknife by venue/liquidity quartiles.


Policy proposals (actionable)

Faster provenance.

Platform-level labels for

unverified market-moving claims

plus

prominent, machine-indexable corrections.

Clearer promotion rules.

Extend/clarify disclosure and anti-touting rules to cover paid

“signals,” Telegram/Discord rooms, and revenue-share affiliate links (SEC/FCA models).

Public “trust scores.”

Independent, open-method ratings for financial news handles and

channels, based on post-event accuracy and volatility externalities.

Reg-Tech co-ops.

Shared early-warning networks between exchanges, large brokers, and

vendors to broadcast misinformation refutations as structured data.

Information sources:

Media tone & markets:

Tetlock,

Journal of Finance

(2007).

Volatility modeling:

Bollerslev, “GARCH,”

Journal of Econometrics

(1986).

Social media diffusion:

Vosoughi, Roy & Aral,

Science

(2018).

AP Twitter hack market

impact:

CNBC (2013).

Presidential tweet → LMT drawdown:

Bloomberg; The Guardian (2016).

Short-and-

distort litigation:

The Denver Post

;

Institutional Investor

(Farmland Partners).

China crypto headlines and BTC:

Reuters (2021).

SVB & social media runs:

Jiang et al., NBER/SSRN (2023).

SEC enforcement (U.S.):

SEC press releases (2022–2023).

FCA finfluencer/social-media rules (U.K.):

FCA webpage/guidance.

Vendor analytics for surveillance:

Bloomberg BNS; Refinitiv MarketPsych.


References

• Media tone & markets: Tetlock, Journal of Finance (2007). Volatility modeling: Bollerslev, “GARCH,” Journal of Econometrics (1986).

• Social media diffusion: Vosoughi, Roy & Aral, Science (2018). AP Twitter hack market impact: CNBC (2013).

• Presidential tweet → LMT drawdown: Bloomberg; The Guardian (2016). Short-and-distort litigation: The Denver Post; Institutional Investor (Farmland Partners).

• China crypto headlines and BTC: Reuters (2021).

• SVB & social media runs: Jiang et al., NBER/SSRN (2023). SEC enforcement (U.S.): SEC press releases (2022–2023).

• FCA finfluencer/social-media rules (U.K.): FCA webpage/guidance.

• Vendor analytics for surveillance: Bloomberg BNS; Refinitiv MarketPsych.