Paper: arXiv 2408.13214

Authors: Hongcheng Ding, Xuanze Zhao, Ruiting Deng, Shamsul Nahar Abdullah, Deshinta Arrova Dewi

Abstract

Accurate forecasting of the EUR/USD exchange rate is crucial for investors, businesses, and policymakers. This paper proposes a novel framework, IUS, that integrates unstructured textual data from news and analysis with structured data on exchange rates and financial indicators to enhance exchange rate prediction. The IUS framework employs large language models for sentiment polarity scoring and exchange rate movement classification of texts. These textual features are combined with quantitative features and input into a Causality-Driven Feature Generator. An Optuna-optimized Bi-LSTM model is then used to forecast the EUR/USD exchange rate. Experiments demonstrate that the proposed method outperforms benchmark models, reducing MAE by 10.69% and RMSE by 9.56% compared to the best performing baseline. Results also show the benefits of data fusion, with the combination of unstructured and structured data yielding higher accuracy than structured data alone. Furthermore, feature selection using the top 12 important quantitative features combined with the textual features proves most effective. The proposed IUS framework and Optuna-Bi-LSTM model provide a powerful new approach for exchange rate forecasting through multi-source data integration.

Complexity vs Empirical Score

  • Math Complexity: 4.0/10
  • Empirical Rigor: 7.5/10
  • Quadrant: Street Traders — empirical and applied, lower math complexity

Why this score: The paper relies on standard deep learning architectures (Bi-LSTM) and LLM fine-tuning, with relatively simple mathematical formulations; however, it demonstrates strong empirical rigor through extensive benchmarking against baselines, hyperparameter optimization with Optuna, specific error reduction metrics (MAE, RMSE), and a clear data fusion pipeline.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Forecast EUR/USD Exchange Rate"] --> B["Data Collection<br>News & Financial Indicators"]
  B --> C["Text Processing<br>LLM Sentiment & Movement Classification"]
  B --> D["Feature Selection<br>Top 12 Quantitative Features"]
  C --> E["IUS Framework Fusion<br>Text + Quantitative Features"]
  D --> E
  E --> F["Causality-Driven<br>Feature Generator"]
  F --> G["Optuna-Bi-LSTM<br>Model Optimization & Forecasting"]
  G --> H["Outcomes<br>Reduced MAE by 10.69% & RMSE by 9.56%<br>Multi-source data fusion superior to structured-only"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes the IUS framework that integrates unstructured textual and structured quantitative data, and an Optuna‑optimized Bi‑LSTM model for EUR/USD exchange rate forecasting.

Source sentence
We propose a novel IUS framework that integrates unstructured and structured data to predict exchange rates, setting a new benchmark in the industry.

Assumes 1

  • Missing quantitative values are linearly interpolated between known points. data
    Source sentence
    We use linear interpolation to fill in these gaps: vi = va + (vb −va)(ti −ta) / (tb −ta)

Data

  • Textual dataset from investing.com and forexempire.com · investing.com, forexempire.com text-news public
    2016-02-06 to 2024-01-19 · 35,427 texts (filtered to 20,329 relevant texts)
    Source sentence
    For the textual dataset, we collect data from investing.com and forexempire.com, covering the period from February 6, 2016, to January 19, 2024. The dataset includes all accessible data on these platforms within the period, totaling 35,427 texts
  • Quantitative financial data (exchange rates and market indicators) market-prices unknown
    Source sentence
    All collected raw financial data are fed into an indicator generator, which aligns the data and fills in missing values, producing the final quantitative features.

Recognised sources: Twitter / X

Costs

Not applicable The paper reports no trading or execution result.

Finds 3

  • The Optuna‑Bi‑LSTM model reduces MAE by 10.69% and RMSE by 9.56% compared with the best baseline model. performance
    Source sentence
    It can be observed that our proposed method consistently outperforms other models in terms of both MAE and RMSE metrics, achieving an improvement of at least 10.69% in MAE and 9.56% in RMSE compared with the best other model.
  • Combining unstructured textual and structured quantitative data yields higher accuracy than using structured data alone. predictive-accuracy
    Source sentence
    In terms of data fusion, by combining unstructured and structured data, the model is able to enhance prediction accuracy beyond what is possible with structured data alone.
  • Using the top 12 important quantitative features together with 31 textual features is more effective than using all textual features. predictive-accuracy
    Source sentence
    Furthermore, using the top 12 important features selected by the RFE method combined with 31 textual features proves to be more effective compared to analyzing all textual features.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 25 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (122)

  1. Exchange rate predictability (2013)
  2. Exchange rate prediction redux: New models, new data, new currencies (2019)
  3. Utilising Sentiment Analysis To Predict Market Movements In Currency Exchange Rates (2024)
  4. Impact of News Sentiment on Foreign Exchange Rate Prediction (2023)
  5. Financial time series forecasting with deep learning: A systematic literature review: 2005–2019 (2020)
  6. A survey of forex and stock price prediction using deep learning (2021)
  7. News will tell: Forecasting foreign exchange rates based on news story events in the economy calendar (2020)
  8. Commodity prices and exchange rates: evidence from commodity-dependent developed and emerging economies (2023)
and 114 more
  1. EUR/USD Exchange Rate Prediction Using Machine Learning (2022)
  2. Forecasting exchange rate volatility: is economic policy uncertainty better? (2024)
  3. Improving exchange rate forecasting via a new deep multimodal fusion model (2022)
  4. Oil price and exchange rate behaviour of the BRICS (2021)
  5. Explaining Exchange Rate Forecasts with Macroeconomic Fundamentals Using Interpretive Machine Learning — → 2303.16149
  6. Do sentiment indices always improve the prediction accuracy of exchange rates? (2022)
  7. The dynamic effect of macroeconomic news on the euro/US dollar exchange rate (2020)
  8. Text-based crude oil price forecasting: A deep learning approach (2019)
  9. Crude oil price forecasting incorporating news text (2022)
  10. An optimal deep learning-based LSTM for stock price prediction using twitter sentiment analysis (2022)
  11. Making text count: economic forecasting using newspaper text (2022)
  12. Machine learning-based USD/PKR exchange rate forecasting using sentiment analysis of Twitter data (2021)
  13. Exchange rate market trend prediction based on sentiment analysis (2023)
  14. Sentiment-Driven Exchange Rate Forecasting: Integrating Twitter Analysis with Economic Indicators (2024)
  15. Analysis and Forecast of USD/EUR Exchange Rate Based on ARIMA and GARCH Models (2023)
  16. RMB/USD Exchange Rate Forecasting and Analysis by SES and ARIMA Model (2024)
  17. Statistical learning and exchange rate forecasting (2020)
  18. Exchange rate forecasting with advanced machine learning methods (2021)
  19. Predicting exchange rate under uirp framework with support vector regression (2022)
  20. A new ensemble deep learning approach for exchange rates forecasting and trading (2020)
  21. A new LASSO-BiLSTM-based ensemble learning approach for exchange rate forecasting (2024)
  22. Foreign exchange currency rate prediction using a GRU-LSTM hybrid network (2021)
  23. Forex exchange rate forecasting using deep recurrent neural networks (2020)
  24. Multivariate temporal convolutional network: A deep neural networks approach for multivariate time series forecasting (2019)
  25. Are transformers effective for time series forecasting? (2023)
  26. Research on particle swarm optimization in LSTM neural networks for rainfall-runoff simulation (2022)
  27. An efficient optimization approach for designing machine learning models based on genetic algorithm (2021)
  28. Automatic tuning of hyperparameters using Bayesian optimization (2021)
  29. Dynamical hyperparameter optimization via deep reinforcement learning in tracking (2019)
  30. Model Performance Prediction for Hyperparameter Optimization of Deep Learning Models Using High Performance Computing and Quantum Annealing (2024)
  31. The whale optimization algorithm approach for deep neural networks (2021)
  32. Compos Mentis at SemEval2024 Task6: A Multi-Faceted Role-based Large Language Model Ensemble to Detect Hallucination (2024)
  33. Keyword-based Annotation of Visually-Rich Document Content for Trend and Risk Analysis using Large Language Models (2024)
  34. Knowledge Tagging System on Math Questions via LLMs with Flexible Demonstration Retriever (2024) — link
  35. RoBERTa: A Robustly Optimized BERT Pretraining Approach (2019) — link
  36. Tweet Insights: A Visualization Platform to Extract Temporal Insights from Twitter (2023) — link
  37. Dual student: Breaking the limits of the teacher in semi-supervised learning (2019)
  38. On mean absolute error for deep neural network based vector-to-vector regression (2020)
  39. Classification vs regression in overparameterized regimes: Does the loss function matter? (2021)
  40. Evaluation of neural architectures trained with square loss vs cross-entropy in classification tasks (2020) — link
  41. Cross-entropy loss functions: Theoretical analysis and applications (2023)
  42. Polyloss: A polynomial expansion perspective of classification loss functions (2022) — link
  43. Weighted Signed Networks Reveal Interactions between US Foreign Exchange Rates (2024)
  44. Asymmetric volatility connectedness on the forex market (2017)
  45. Measuring the connectedness of the global economy (2021)
  46. Contagion effects of US Dollar and Chinese Yuan in forward and spot foreign exchange markets (2017)
  47. Currency Exchange Rate Forecasting with Social Media Sentiment Analysis (2022)
  48. Conditional correlation between exchange rates and stock prices (2021)
  49. Forex rates prediction on EUR/USD with simple moving average technique and financial factors (2020)
  50. Optimizing the performance of the fractal adaptive moving average strategy: The case of EUR/USD (2016)
  51. Speculation and hedging in the currency futures markets: Are they informative to the spot exchange rates (2012)
  52. Can oil prices forecast exchange rates? An empirical analysis of the relationship between commodity prices and exchange rates (2015)
  53. Commodity currencies (2003)
  54. A simultaneous trade model of the foreign exchange hot potato (1997)
  55. Oil price fluctuations and US dollar exchange rates (2010)
  56. The impact of oil shocks on exchange rates: A Markov-switching approach (2016)
  57. Commodity currencies and the real exchange rate (2004)
  58. How does the US natural gas market react to demand and supply shocks in the crude oil market? (2017)
  59. Gold and the Dollar (and the Euro, Pound, and Yen) (2011)
  60. The price of gold and the exchange rates: Once again (2008)
  61. Exchange rates and commodity prices: Measuring causality at multiple horizons (2016)
  62. Oil price, agricultural commodity prices, and the dollar: A panel cointegration and causality analysis (2012)
  63. Returns and volatility spillover between food prices and exchange rate in Nigeria (2020)
  64. The relationship between agricultural commodity prices, crude oil prices and US dollar exchange rates: A panel VAR approach and causality analysis (2015)
  65. Volatility spillover between oil and agricultural commodity markets (2013)
  66. Forecasting the US Dollar in the 21st Century (2023)
  67. Monetary policy and long-horizon uncovered interest parity (2004)
  68. Determining the EUR/USD exchange rate with US and German government bond yields in the post-crisis period (2015)
  69. Euro area sovereign yields and the power of unconventional monetary policy (2018)
  70. Cross-border interbank networks, banking risk and contagion (2015)
  71. Dollar funding and the lending behavior of global banks (2015)
  72. Short-run forecasting of the euro-dollar exchange rate with economic fundamentals (2012)
  73. Reforming LIBOR and other financial market benchmarks (2015)
  74. Deviations from covered interest rate parity (2018)
  75. Measuring fragmentation in the euro area unsecured overnight interbank money market (2018)
  76. ECB monetary policy and the euro exchange rate (2018)
  77. Dynamic linkages between exchange rates and stock prices: Evidence from East Asian markets (2007)
  78. Predict forex trend via convolutional neural networks (2019)
  79. A review and empirical analysis of neural networks based exchange rate prediction (2018)
  80. Dynamic relationship between stock prices and exchange rates for G-7 countries (2001)
  81. The comovement between exchange rates and stock prices in the Asian emerging markets (2012)
  82. Stock prices and exchange rate dynamics (2005)
  83. Dynamic relations between stock returns and exchange rate changes (2014)
  84. Dynamic linkage between real exchange rates and stock prices: Evidence from developed and emerging Asian markets (2014)
  85. The relationship between stock price index and exchange rate in Asian markets: A quantile regression approach (2012)
  86. Dynamic linkages between US and Eurodollar interest rates: new evidence from causality in quantiles (2021)
  87. A study of exchange rates movement and stock market volatility (2010)
  88. Stock and foreign exchange market linkages in emerging economies (2013)
  89. Carry trades and currency crashes (2008)
  90. Exchange rates and global volatility: implications for Asia-Pacific currencies (2007)
  91. Improving volatility prediction and option valuation using VIX information: A volatility spillover GARCH model (2019)
  92. Business-cycle anatomy (2020)
  93. Predicting returns with text data (2019)
  94. Latent dirichlet allocation (2003)
  95. Machine learning in finance: A topic modeling approach (2022)
  96. EVALUATION OF ANNUAL REPORTS OF COMPANIES WITH SENTIMENT ANALYSIS: AN APPLICATION IN BIST100 INDEX (2023)
  97. A guided latent Dirichlet allocation approach to investigate real-time latent topics of Twitter data during Hurricane Laura (2023)
  98. Dynamic topic models (2006)
  99. A practitioner’s guide to lag order selection for VAR impulse response analysis (2005)
  100. Vector autoregressions (2001)
  101. Application of Vector Autoregression (Var) on Modelling and Forecasting Average Monthly Rainfall and Temperature (2024)
  102. Multivariate Time Series Forecasting to Forecast Weight Dynamics (2023)
  103. Identifying long-run risks: A Bayesian mixed-frequency approach (2018)
  104. Using recursive feature elimination in random forest to account for correlated variables in high dimensional data (2018)
  105. Cost-sensitive feature selection using random forest: Selecting low-cost subsets of informative features (2016)
  106. Correlation and variable importance in random forests (2017)
  107. Adaptability of financial time series prediction based on BiLSTM (2022)
  108. Multi-source aggregated classification for stock price movement prediction (2023)
  109. The performance of LSTM and BiLSTM in forecasting time series (2019)
  110. Optuna: A next-generation hyperparameter optimization framework (2019)
  111. A Ship Trajectory Prediction Method Based on an Optuna–BILSTM Model (2024)
  112. Volatility forecasting for high-frequency financial data based on web search index and deep learning model (2021)
  113. Comparing predictive accuracy (2002)
  114. Out-of-sample exchange rate predictability with Taylor rule fundamentals (2009)

1 of 122 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.