Core Concepts
A novel three-layer stacked Machine Learning system to detect financial opportunities, i.e. positive user speculations and forecasts about stock market assets, in micro-blogging data with high precision.
Abstract
The paper proposes a three-layer stacked Machine Learning system to detect financial opportunities, defined as positive user speculations and forecasts about stock market assets, in micro-blogging data.
The first stage of the system distinguishes between neutral and non-neutral entries. The second stage separates positive and negative emotions. The final stage extracts opportunities from positive statements.
The system leverages sophisticated linguistic features such as n-grams, emotion and polarity dictionaries, and frequency counters of hashtags, numerical information and percentages. It also considers temporal features based on discursive analysis.
Experimental results on a manually annotated dataset of 6,000 tweets demonstrate that the proposed system achieves satisfactory and competitive performance in financial opportunity detection, with precision values up to 83%. This endorses the usability of the system to support investors' decision making.
The authors also define two tolerance metrics to assess the system's performance from an operator's perspective, focusing on minimizing the presentation of non-opportunity entries as opportunities.
Stats
+- 15.5%
+2.26%
-2.9%
-16.75%
Quotes
"the average yield of Google is 15.5%"
"the #IBEX35 is going to make a good daily close, although I would like it to break 8,600C"
"#FelizLunes last day of the year and #ibex35 is going up jokingly :)"
"#DowJones closes with a 0.36 percent loss"