Using Price Volatility to Infer Sentiment in Financial News
Summary
PrivySense proposes estimating sentiment in financial news through observed price volatility. This reverses the conventional setup described in the paper, where sentiment classifiers are trained on human-labeled text and then used to inform securities trading. The work also examines existing sentiment classifiers and asks how sentiment should be defined in a trading context.
The authors scrutinize the use of human annotations, arguing that labeling choices can introduce subjective bias into financial news sentiment data. The paper situates this concern within the wider workflow of supervised sentiment classification: models are trained on annotated corpora, evaluated on held-out text, and may then be assessed through trading strategies against benchmarks. However, the provided text does not detail the volatility-based estimation procedure, report classifier comparisons or trading results, or establish that volatility provides a reliable sentiment label. The abstract therefore outlines a proposed method and measurement critique, but leaves its empirical strength and practical limits unresolved.
Key ideas
- The paper proposes using price volatility to infer sentiment expressed in financial news.
- This approach reverses the common direction of using labeled news sentiment to study market outcomes.
- The work evaluates how sentiment classifiers and trading-context definitions of sentiment should be assessed.
- Human annotation can introduce subjective bias into sentiment labels.
- The provided description gives no method details or empirical performance results for the proposed estimator.
Tags
Full text
# 1801.00091
# PrivySense: $\underline{Pri}$ce $\underline{V}$olatilit$\underline{y}$ based $\underline{Sen}$timent$\underline{s}$ $\underline{E}$stimation from Financial News using Machine Learning
As machine learning ascends the peak of computer science zeitgeist, the usage and experimentation with sentiment analysis using various forms of textual data seems pervasive. The effect is especially pronounced in formulating securities trading strategies, due to a plethora of reasons including the relative ease of implementation and the abundance of academic research suggesting automated sentiment analysis can be productively used in trading strategies. The source data for such analyzers ranges a broad spectrum like social media feeds, micro-blogs, real-time news feeds, ex-post financial data etc. The abstract technique underlying these analyzers involve supervised learning of sentiment classification where the classifier is trained on annotated source corpus, and accuracy is measured by testing how well the classifiers generalizes on unseen test data from the corpus. Post training, and validation of fitted models, the classifiers are used to execute trading strategies, and the corresponding returns are compared with appropriate benchmark returns (for e.g., the S&P500 returns). In this paper, we introduce $\underline{a\ novel\ technique\ of\ using\ price\ volatilities\ to\ empirically\ determine\ the\ sentiment\ in\ news\ data}$, instead of the traditional reverse approach. We also perform meta sentiment analysis by evaluating the efficacy of existing sentiment classifiers and the precise definition of sentiment from securities trading context. We scrutinize the efficacy of using human-annotated sentiment classification and the tacit assumptions that introduces subjective bias in existing financial news sentiment classifiers.Shown in full with attribution under the source's licence. Licence: abstract CC0
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.