 |
| U.S. mood throughout the day inferred from Twitter |
Is there a better example of Keynes' "beauty contest" metaphor than Derwent Capital beating the market with a Twitter-based algorithm ? The underlying in Derwent's trading strategy is the feeling of the crowd. Bloomberg's Jack Jordan reports that the trading model is developed by computer scientists Bollen, Mao and Zeng and will gauge the mood of the market, tracking especially instances of word related to a calm mood:
Their results showed that rises and falls in the number of instances of words related to a calm mood could be used to predict the same moves in the Dow’s closing price between two and six days later, with a fall in these “calm” words being followed by a fall in the index. The other moods did not have the same predictive quality.
The algo is based on Bollen, Mao and Zeng's paper published by the Journal of Computational Science in March 2011:
Our results indicate that the accuracy of DJIA (Dow Jones Industrial Average) 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%.
Nofsinger in 2005 published a similar paper in the Journal of Behavioral Finance . He notes that the "stock market is itself a direct measure or gauge of social mood". The causality is reversed in these perspective. Social mood is not just correlated with market, it moves it. An algo based on Twitter would then allow its user to be ahead of the curve. In a world of constant media coverage, this has huge implications for our understanding of financial systems.
No comments:
Post a Comment