Cracking the Code: How Urdu Tweets Can Predict Currency Fluctuations

Exchange rates are like puzzles, and financial experts have been trying to solve them for years. They usually rely on past data, but that's not the whole story. What if people's feelings and opinions on social media could help forecast these rates? A recent study explored this idea, focusing on Urdu-language tweets and their impact on the USD/PKR exchange rate.

The researchers collected a massive 172,000 tweets from January 2021 to January 2025 using trending hashtags. After filtering out non-Urdu tweets, they were left with 45,048 contextually relevant tweets. The team then used four different methods to analyze the sentiment of these tweets: Gemini 1.5 Flash, a customized GPT-3.5 Turbo, GPT-4o, and XGBoost trained on FastText embeddings.

These sentiment scores were matched with exchange rate data from the State Bank of Pakistan and fed into three models: Long Short-Term Memory (LSTM), Xtreme Gradient Boosting (XGBoost), and a hybrid LSTM+XGBoost model. The results were striking - the hybrid model using GPT-4o-based sentiment outperformed the others, with a Root Mean Squared Error (RMSE) of 0.0831 and a Mean Absolute Percentage Error (MAPE) of 0.03%.

This study shows that public opinion on Urdu social media can be a valuable tool for predicting exchange rates. By combining historical data with social media sentiment, the hybrid model performed better than the LSTM baseline trained on historical data alone. This approach could be a game-changer for forecasting currency fluctuations, especially in times of economic uncertainty.

The study's findings highlight the importance of considering public sentiment in exchange rate forecasting. By tapping into the collective opinions on social media, researchers can gain a deeper understanding of market trends and make more accurate predictions. This could have significant implications for investors, policymakers, and anyone interested in understanding the complexities of currency markets.

The researchers' work demonstrates that hybrid architectures are more effective than standalone models in exploiting public opinion on Urdu social media for exchange rate prediction. As social media continues to shape public discourse, its role in informing financial decisions will only grow. This study offers a glimpse into the potential of social media sentiment analysis in shaping the future of finance.