Sector Rotation Strategies Driven By News Sentiment Indices

Executive Summary:

Such indices prove to be a valuable input into sector rotation strategies, thereby allowing sentiment to help anticipate which industries or companies will be successful in the upcoming stages of an economic cycle.

• Results indicate that negative sentiment is a strong leading indicator of future underperformance, while positive sentiment is not considered as clear a leading indicator of future outperformance.

• The study shows how news sentiment helps reduce risk while improving cumulative returns when constructing industry rotation strategies. During the test period of February 2005 through September 2009, using news sentiment instead of one-month return momentum improved annualized returns from 18.49% to 29.63%, and the Information ratio from 0.95 to 1.80.

• Further results indicate that tracking the sentiment of Top and Bottom industries could add value in detecting periods of stronger positive or negative sentiment environments. Such strategy could provide greater confidence in trading the market more aggressively, thereby allowing sentiment to define the magnitude of one's investment exposure.

Introduction

Applying news analytics in a trading model allows for the possibility not only to react in real-time to scheduled and unscheduled news events in a fully or semi-automated fashion, but also to consider the prevailing sentiment trend on a given market. Such trends can be captured by looking at aggregated news sentiment on single companies, sectors, industries or even on broader equity indices. In this paper, the focus will be on the construction and evaluation of a set of industry sentiment indices. Such indices may prove to be a valuable input into a sector rotation strategy, allowing sentiment to help anticipate which industries or companies will be successful in the upcoming stages of an economic cycle.

Generally, sector rotation is an investment strategy involving the movement of money from one industry sector to another in an attempt to beat the market. Most of the time, Financial markets attempt to predict the state of the economy, anywhere from three to six months into the future. That means the market cycle is usually well ahead of the economic cycle. This is crucial to remember because as the economy is in the pits of a recession, the market begins to look ahead to a recovery. Tracking industry and market sentiment could help not only to detect industries that are likely to out- and underperform the market in the next period, but also to decide on what directional view to take on the market.

Furthermore, industry sentiment may help detect periods of stronger positive or negative sentiment environments. Such strategy could provide greater confidence in trading the market more aggressively, thereby allowing sentiment to define the magnitude of one's investment exposure.

In Section 2, I provide an overview of the methodology on how to construct industry sentiment indices. In Section 3, I describe the data being used as part of the analysis. In Section 4, I present the results of the study focusing on the performance spread between industries with high versus low sentiment values. Furthermore, I consider how an introduction of a market sentiment overlay will impact the results. Finally in Section 5, I present the conclusion of the study.

Aggregated Sentiment Indices

In general, news sentiment indices try to capture the prevailing sentiment trend for a particular equity index, sector, or industry. In order to capture such a trend, it seems reasonable to consider an aggregation of news sentiment over a given window with the purpose of capturing the general "mood" of the market. To construct such indices, I suggest calculating what I define as a sentiment ratio, which considers the ratio of positive to negative news events over well-defined moving aggregation windows. Focusing on ratios rather than net values of positive to negative news allows for news flow normalization, which seems necessary as I have previously shown that strong seasonal patterns characterize news data. Using the empirical sentiment ratio distribution, the ratios are mapped into a sentiment index ranging between 0 and 100 with values close to 50 representing neutral sentiment. In the study "Construction of Market Sentiment Indices Using News Sentiment", I showed how to construct sentiment indices on the Dow Jones Industrial Average (DJIA) and EuroStoxx 50.