Financial Bubble Detection Using GSADF and LSTM-RNN Model: Evidence from Emerging Markets
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Forecasting financial bubbles is a crucial task in financial economics due to the disruptive impact of asset price collapses on markets and economic stability. This study proposes a novel approach to bubble prediction by integrating the PSY (Phillips, Shi, and Yu) procedure for bubble detection with Long Short-Term Memory Recurrent Neural Networks (LSTM-RNN), a machine learning technique well-suited for modeling nonlinear time-series patterns. Using weekly data from the Vietnamese stock market covering the period from 2015 to 2025, we construct a binary dependent variable indicating the presence of bubble episodes based on the GSADF test. Key macro-financial variables, including returns, volatility, and geopolitical risk, are employed as predictors. The LSTM-RNN model is trained and validated using a time-split approach (2015–2019 for training, 2020–2022 for validation, and 2023–2025 for testing), ensuring robustness and preventing overfitting. Out-of-sample results demonstrate that the LSTM-RNN achieves a high accuracy of over 81% and significantly outperforms a random walk benchmark. Our findings highlight the critical role of macroeconomic uncertainty, especially geopolitical risk, in driving bubble dynamics. This research contributes to the literature by offering an early warning framework that combines econometric detection with advanced machine learning, supporting better decision-making for investors and financial regulators in emerging markets.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.003 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it