Twitter mood predicts the stock market
By Johan Bollen, Huina Mao, Xiao-Jun Zeng. 2010
5,439 citations recorded by the source corpus.
Behavioral economics tells us that emotions can profoundly affect individual behavior and decision-making. Does this also apply to societies at large, i.e. can societies experience mood states that affect their collective decision making? By extension is the public mood correlated or even predictive of economic indicators? Here we investigate whether measurements of collective mood states derived from large-scale Twitter feeds are correlated to the value of the Dow Jones Industrial Average (DJIA) over time. We analyze the text content of daily Twitter feeds by two mood tracking tools, namely OpinionFinder that measures positive vs. negative mood and Google-Profile of Mood States (GPOMS) that measures mood in terms of 6 dimensions (Calm, Alert, Sure, Vital, Kind, and Happy). We cross-validate the resulting mood time series by comparing their ability to detect the public's response to the presidential election and Thanksgiving day in 2008. A Granger causality analysis and a Self-Organizing Fuzzy Neural Network are then used to investigate the hypothesis that public mood states, as measured by the OpinionFinder and GPOMS mood time series, are predictive of changes in DJIA closing values. Our results indicate that the accuracy of DJIA predictions can be significantly improved by the inclusion of specific public mood dimensions but not others. We find an accuracy of 87.6% in predicting the daily up and down changes in the closing values of the DJIA and a reduction of the Mean Average Percentage Error by more than 6%. Index Terms—stock market prediction — twitter — mood analysis.
Subfields: Artificial Intelligence, Economics and Econometrics, Management Science and Operations Research, Sentiment & Opinion Analysis, Time-Series & Forecasting
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Mohit Bansal
- Erik Cambria
- Percy Liang
- Jianfeng Gao
- Noah A. Smith
- Luke S. Zettlemoyer
- Yejin Choi
- Yoshua Bengio
In Economics and Econometrics
- Qing Li
- Erik Cambria
- Frank Xing
- Roy E. Welsch
- Mark Dras
- Leonardo dos Santos Pinheiro
- Olivier Kraaijeveld
- J. Smedt
In Management Science and Operations Research
- Weiwei Jiang
- Yumo Xu
- Shay B. Cohen
- Erik Cambria
- Frank Xing
- Roy E. Welsch
- Omer Berat Sezer
- A. Ozbayoglu
In Sentiment & Opinion Analysis
- Mohamed M. Mostafa
- Erik Cambria
- Sven Buechel
- Saif M. Mohammad
- Emre Kıcıman
- Diyi Yang
- Daniel Gayo-Avello
- Udo Hahn
In Time-Series & Forecasting
- Omer Berat Sezer
- A. Ozbayoglu
- Mark Dras
- Leonardo dos Santos Pinheiro
- M. Ugur Gudelek
- Yumo Xu
- Shay B. Cohen
- Fatemeh Hemmatian
Citations in the corpus
- Sentiment Predictability for Stocks (2017, neutral): “[BMZ11], cited 2800 times, found that there was statistical significance between the Granger Causality of certain aggregate ”mood” scores of 340,000 daily Twitter posts and the Dow Jones index approximately 3-4 days later.”
- Improving stock market prediction via heterogeneous information fusion (2017, methodology): “…(NLP) techniques, sentiment-driven stock prediction techniques are also proposed by extracting indicators of public mood from social media [9, 10, 11], where the positive mood for a stock will probably indicate a rise trend in the price and negative mood will more likely mean a decrease…”
- Deep Learning Stock Volatility with Google Domestic Trends (2015, methodology): “[4] incorporated public mood data in directional prediction of the Dow Jones Industrial Average.”
- Using Structured Events to Predict Stock Price Movement: An Empirical Investigation (2014, background): “Bollen and Zeng (2011) study whether the large-scale collective emotion on Twitter is correlated with the volatility of Dow Jones Indus-trial Average (DJIA).”
- That’s So Annoying!!!: A Lexical and Frame-Semantic Embedding Based Data Augmentation Approach to Automatic Categorization of Annoying Behaviors using #petpeeve Tweets (2015, methodology): “Bollen et al. (2011) use a sentiment analysis approach to predict the American stock market via Twitter.”
- Using social media to enhance citizen engagement with local government: Twitter or Facebook? (2018, methodology): “The descriptive analysis in this study is based on Bonsón and Ratkai (2013) and Bonsón et al. (2017), who proposed and used a set of metrics to identify the dimensions of engagement through social media, and on Agostino (2013), who measured engagement, taking into account the population of the munic”
- A geographic approach for combining social media and authoritative data towards identifying useful information for disaster management (2015, neutral): “2013) and predicting the stock market (Bollen et al. 2011).”
- Topic Modeling based Sentiment Analysis on Social Media for Stock Market Prediction (2015, methodology): “Most of the previous researches are limited on predicting for one stock (Bollen et al., 2011; Qian and Rasheed, 2007; Si et al., 2013), and the number of instances (transaction dates) in a test set is rather low such as 14 or 15 instances (Bollen et al., 2011; Vu et al., 2012).”
- SemEval-2017 Task 4: Sentiment Analysis in Twitter (2017, background): “2012), economics (Bollen et al., 2011; O’Connor et al., 2010), social science (Dodds et al.”
- Measuring the Information Content of Financial News (2016, background): “…into tree major categories: bag-of-words (Kogan et al., 2009; Lee et al., 2014; Qiu et al., 2014), sentiment-based (Das and Chen, 2007; Li, 2010; Bollen et al., 2011; Vu et al., 2012; Luss and d’Aspremont, 2015), and information-retrieval-based (Schumaker and Chen, 2009; Xie et al., 2013; Ding…”
- A review of unsupervised feature learning and deep learning for time-series modeling (2014, background): “If additional information from sources that affect the stock market can be measured and obtained, such as general public opinions from social media [14], trading volume [134], market specific domain knowledge, and political and economical factors, it can be combined together with the stock price dat”
- A Benchmark Comparison of State-of-the-Practice Sentiment Analysis Methods (2015, background): “The first of them, applied by OpinionFinder and Pattern.en, for instance, splits this task into two steps: (i) identifying sentences with no sentiment, also named as objective vs. neutral sentences and then (ii) detecting the polarity (positive or negative), only for the subjective sentences.”
- Can Facebook Predict Stock Market Activity (2013, neutral): “Daily trading volume is detrended using the moving average detrending as in Campbell et al. (1993) and Tetlock (2007). Daily economic activity is measured by the ADS index (Aruoba et al.”
- Automatic Crime Prediction Using Events Extracted from Twitter Posts (2012, neutral): “For example, Twitter posts have been used to predict box office results [1], election results [2], and stock market trends [5].”
- Online Human-Bot Interactions: Detection, Estimation, and Characterization (2017, neutral): “Sentiment extracted from social media conversations has been used to forecast offline events including financial market fluctuations (Bollen, Mao, and Zeng 2011), and is known to affect information spreading (Mitchell et al. 2013; Ferrara and Yang 2015).”
- Like It or Not (2016, background): “…has recently attracted the interest of many researchers who analyzed Twitter data for a variety of different tasks such as making predictions [Bollen et al. 2011], detecting users’ sentiment towards different topics [Go et al. 2009], detecting users’ emotions [Mohammad 2012], and detecting…”
- Limits of Electoral Predictions Using Twitter (2011, neutral): “…public health (e.g. (Ginsberg et al. 2009), (Lampos, Bie, and Cristianini 2010)), business (e.g. (Asur and Huberman 2010), (Shimshoni, Efron, and Matias 2009)), economics (e.g. (Bollen, Mao, and Zeng 2010), (Choi and Varian 2009)), and politics (e.g. (O'Connor et al. 2010), (Tumasjan et al. 2010)).”