Please enable JavaScript.
Coggle requires JavaScript to display documents.
previous research findings and analysis, How sentiment changes…
previous research findings and analysis
Contagion Effects in Financial Markets: Influence of Green
Finance and Energy Sectors During Global Crises
pronounced interdependence
These markets are financially and behaviorally linked, even though they are often treated as separate asset classes
Green bonds ,Renewable energy assets ,Fossil fuel markets do not move independently
example:
1- A sharp increase in oil price volatility can affect renewable energy stocks 2- Stress in fossil fuel markets can spill over into green bonds through portfolio rebalancing and liquidity channels
Shocks in one market (price drops, volatility spikes, uncertainty) are transmitted to the others
stronger connections during geopolitical events and crises
results: Investors rebalance across all energy-related and sustainability-related assets simultaneously. Assets that appear different in normal times become highly correlated under stress
In econometric terms: Spillovers intensify, Cross-market volatility transmission increases and Time-varying connectedness spikes
causes: Uncertainty rises sharply, Investors become more risk-averse,Portfolio diversification weakens
due to the contagion effect: A shock originating in one market spreads rapidly to others which increases volatility and reduce hedging effectivness of green bonds
investor sentiment, Liquidity constraints and flight-to-cash behavior present an indirect causes
resilience of green bonds
-Lower volatility compared to traditional bonds and risky assets -Limited sensitivity to shocks originating from equity and commodity markets
This means green bonds are less prone to panic-driven sell-offs during turbulent periods
means: Absorb market shocks, Maintain relatively stable prices and Avoid excessive volatility during periods of stress
green bonds as safe haven
This allows investors to shift capital into green bonds during crises without amplifying losses elsewhere in the portfolio
Green bonds remain weakly correlated with conventional equities and commodities
Their performance does not deteriorate sharply during turbulence
Dependence structure and dynamic connectedness between green bonds and financial markets: Fresh insights from time-frequency analysis before and during COVID-19 pandemic
Diversification across investment horizons
Green bonds provide diversification benefits at all horizons (short-, medium-, and long-term), meaning they help reduce overall portfolio risk regardless of the holding period.
Declining hedging effectiveness in the long run
Although diversification remains, the hedging power of green bonds weakens over longer horizons. Over time, correlations with other assets tend to increase, reducing their ability to fully offset losses.
Portfolio inclusion of green bonds
Combining green bonds with equities and energy markets improves diversification, because green bonds behave differently from these assets, especially in the short run.
Active investors
(short-term traders) benefit more because:Short-run hedging effectiveness is stronger ; Portfolio rebalancing allows them to exploit time-varying relationships
Passive investors
(long-term holders) gain less protection, as long-run co-movements reduce hedging effectiveness.
IS / portfolio / hedging
Investor sentiment captures the degree of optimism or pessimism in financial markets that is not fully justified by fundamentals
Asset demand
Risk perception
Correlations across assets
Channels through which sentiment affects portfolios
Expected return channel
Sentiment-driven demand inflates prices
Higher prices → lower expected returns
Alters return forecasts used in portfolio optimization
Correlation / connectedness channel
Sentiment shapes co-movements across assets
Crisis sentiment increases market-wide correlations
Sector-specific sentiment (e.g., green sentiment) can decouple assets (When green sentiment is strong,Green assets move more independently from traditional assets and Their correlations with stocks, fossil fuels, or conventional bonds decline
Risk and volatility channel
Sentiment affects perceived and realized volatility
High sentiment → lower volatility
Low sentiment → volatility clustering and tail risk
Liquidity channel
Optimistic sentiment improves liquidity
Pessimistic sentiment causes liquidity dry-ups
Liquidity shocks reduce hedging effectiveness
Regime-switching channel
Sentiment defines market regimes (bull vs. bear)
Portfolio and hedge parameters differ across regimes
Effects of investor sentiment on portfolio and hedging
Hedging effectiveness
Sentiment affects cross-asset correlations
In optimistic periods, correlations may fall → hedging improves :check: :
In pessimistic periods, correlations often rise → hedging weakens :small_red_triangle_down:
Sentiment-driven herding reduces diversification when protection is most needed
Portfolio weights (asset allocation)
Low (pessimistic) sentiment
Shifts portfolios toward defensive or safe-haven assets
investors seek safety rather than return, moving capital into assets perceived as stable
High (optimistic) sentiment
Increases demand for risky or favored assets
Lowers perceived risk → higher portfolio weights
Result
: Optimal portfolio weights become time-varying and sentiment-dependent
safe haven or not ? :!?:
:red_cross:green bonds are not safe haven assets for equity investors but rather show positive co-movement during periods of market stress
:check:most time-varying green bond correlations with financial assets were low and did not change significantly during COVID-19
:check:Green Bonds only act as a safe haven during the normal market condition
:check:
green bonds are found to be the only asset that serves as a safe haven against large stock
market fluctuations due to the COVID-19 pandemic
:check:green bonds act as reliable safe-haven assets in times of market turbulence
:red_cross:green bonds had no diversification or hedge benefits for investors in conventional bonds
green sentiment
LIMITS AND EMERGING GAPS
Limited integration with portfolio and hedging frameworks
Green sentiment is rarely incorporated directly into portfolio optimization, hedging effectiveness, or asset allocation models, especially in a dynamic or regime-dependent setting.
Short time horizons and data constraints
Many green sentiment indices cover relatively short samples, limiting their ability to capture long-term structural changes or multiple crisis episodes.
Insufficient attention to market regimes and crises
The role of green sentiment during crisis periods (e.g., COVID-19, geopolitical shocks) is underexplored, despite evidence that sentiment effects are asymmetric across tranquil and turbulent regimes
Limited cross-market connectedness analysis
Few papers examine how green sentiment influences spillovers and connectedness between green assets and traditional markets (energy, equities, bonds) in a network or systemic-risk framework
Litterature
Green energy sentiment
mesurement:
Sentiment/emotion of tweets containing “green energy” before vs. after Russia–Ukraine war
results
Conflict shifted green‑energy sentiment: more negative emotions but rising confidence/trust in the energy transition opportunity
Green finance sentiment
mesurement:
Twitter sentiment on “green finance” using VADER
results
Majority of tweets are positive (60.2%), suggesting broad support yet some perceived risks
Carbon sentiment
mesurement:
BERT/LSTM sentiment scores on carbon‑related news integrated with firm ESG data
results
: More positive carbon sentiment strengthens the positive effect of ESG on sustainable growth by reducing uncertainty and financing constraints
Green economy policy sentiment
mesurement:
NLP sentiment on survey text about carbon tax & renewables in Indonesia
results
79% of responses supportive; negativity linked to short‑term economic concerns
Investor green sentiment
results:
Higher green sentiment predicts out‑performance of more environmentally responsible firms and higher capex/cash holdings
mesurement:
Abnormal flows into environmental ETFs used to build a Green Sentiment Index
Public sentiment on green consumption (Weibo)
mesurement:
CNN–LSTM sentiment classification of posts about green consumption
results
Attitudes mostly positive; negative sentiment tied to high prices, time costs, and problems in sharing economy
Environmental/climate tweet sentiment
mesurement:
Classical ML models on Thai social‑media posts
results
Sentiment toward environmental sustainability initiatives is predominantly positive, indicating growing long‑term awareness
Sentiment toward EU Green Deal
mesurement:
Sentiment analysis of 582k tweets on EGD
results:
Mostly neutral; positive outweighs negative. Peaks in sentiment align with major policy events
green sentiment affects portfolio composition
Risk and volatility channel
High green sentiment reduces perceived risk of green assets → lower risk premia
Portfolio weights shift toward green assets when sentiment is strong
Expected returns channel
Positive green sentiment increases demand for green assets → higher prices, lower expected returns
Investors may rebalance toward traditional assets if green assets become “overpriced,” or toward green assets if sentiment signals long-term growth
Correlation and connectedness channel
During pessimistic sentiment, correlations may rise, reducing diversification benefits and prompting reallocation.
Optimistic green sentiment creates segmented demand, making green assets less sensitive to shocks in traditional markets
Regime-dependent behavior
The effect of green sentiment differs in: - :eight_pointed_black_star:High vs. low sentiment regimes - :eight_pointed_black_star:Crisis vs. tranquil periods
This implies time-varying optimal portfolio weights
potfolio composition using LSTM
(LSTM) networks are mainly used to forecast returns or price movements, then plugged into classical or RL-based portfolio optimizers
litterature
Hybrid models (CNN‑LSTM, attention-LSTM, MCA-LSTM) frequently outperform plain LSTM in forecasting and portfolio metrics
Performance is data- and horizon‑dependent: in some assets or regimes, simpler methods like EWMA forecast equally well or better
Several studies report outperformance versus major indices (S&P 500, DJIA, FTSE 100) in backtests
Across multiple markets (US, UK, Vietnam, Indonesia), LSTM-based portfolios often beat linear models, ARIMA, SVM, RF and equal‑weight benchmarks in cumulative return and Sharpe ratio
better forecasts → better input to allocation models → potentially higher return or Sharpe ratio.
Roles of LSTM in Portfolio Construction
Direct allocation / weighting:
Some work skips explicit risk modeling and uses LSTM-forecasted expected returns to weight assets, especially over 1–2 year horizons
Risk-factor or volatility modeling:
LSTM/RNNs model latent factor volatilities or cointegration spreads that then drive MV or minimum-variance portfolios
Return prediction then Markowitz MV optimization:
Many studies use LSTM to forecast asset returns and feed them into mean–variance (MV) optimizers, often after preselecting high‑potential assets
Within RL policies:
LSTMs form the state encoder inside reinforcement learning agents that output portfolio weights in real time
How LSTM is used in portfolio composition
Volatility & risk forecasting
As result :
Portfolio weights are adjusted dynamically / Assets with higher predicted risk get lower weights
Many studies use LSTM to forecast:
Conditional volatility
Tail risk
Downside risk (VaR / CVaR)
Regime-aware portfolio allocation
LSTM can implicitly capture non-linear regimes:
Stable vs crisis
Low vs high volatility
IN crisis vs stability framework case Combine Markov Switching + LSTM
Return prediction → portfolio weights
Example: LSTM forecasts next-period returns of green bonds, S&P 500, oil, etc. then Use these forecasts in:
Mean–variance optimization
CVaR minimization
Risk-parity
Dynamic hedging strategies
LSTM predicts future returns (or volatility, or correlations)
These forecasts are then plugged into a portfolio optimization model
Hedging & safe-haven analysis
In green finance literature, LSTM is used to:
Evaluate hedging effectiveness
Build dynamic hedge ratios
Compare LSTM-based strategies vs traditional models (GARCH, DCC)
LIMITS
It does not directly explain why investors rebalance portfolios
LSTM does not provide economic interpretability and real‑world robustness is uncertain
Requires large datasets and careful tuning
Overfitting and regime shifts (e.g., COVID-19) can degrade performance; incremental or adaptive schemes are being explored
LSTM generally helps more as input to a well-chosen optimization/risk framework (MV, MSAD, Sharpe-max, RL) than as a stand‑alone trading rule
How sentiment changes correlations and spillovers.
DCC-GARCH (dynamic correlations)
TVP-VAR connectedness framework
Quantile VAR (QVAR)
Frequency-domain connectedness
How sentiment predicts or conditions asset returns
Predictive regressions (returns on sentiment)
Quantile regression (to capture asymmetric effects)
Markov-switching mean models (bull vs. bear sentiment regimes)
Impact of sentiment on volatility and downside risk
GARCH / EGARCH / TGARCH
GARCH-X (sentiment as an exogenous variable)
Downside risk models (VaR, CVaR conditional on sentiment)
Transmission of liquidity shocks under different sentiment states
VAR with liquidity and sentiment measures
Regime-switching liquidity models
Panel regressions with liquidity proxies
Structural shifts between high- and low-sentiment states
Markov-switching VAR
Threshold models (sentiment-based regimes)
Smooth transition models