In this post we present results of a research paper which shows that using Earnings Sentiment is a better predictor of monthly returns than Sell-side Analysts' Revisions.
Overview
Market:
Country/Region: North America, USA
Asset Class: Equities
Investment Universe: Russell 1000
Research & Application Context:
Alt. Data Factor: Media earnings sentiment
Method of Data Integration: Standalone
Period of Analysis: 01/01/2002-01/07/2018
Investment Direction: Long
Avg. Holding Period: month(s)
Avg. Portfolio Size: 200 stocks
Investment Process
Underlying investment logic
Instead of relying only on Wall Street's Sell-side Earnings Revisions, use Earnings Sentiment indices sourced by a Sentiment data provider from published news articles which give a larger and more frequently updated panel of earnings estimates.
Testing Model
Compare the efficacy of these two different estimates by building two monthly rebalanced portfolios composed of the top quintile (20%) of Russell 1000 stocks ranked by trailing 3-month Sell Side Revisions and trailing 3-month Earnings Sentiment and measure the returns.
Main Results

The research shows that for large cap US stocks in the period 2002-2018:
Earnings Sentiment measured over the last 3-months was more informative of the next month's market-adjusted return than trailing 3-month Sell-side Analysts' Earnings Revisions
Each month selecting the stocks in the highest Earnings Sentiment quintile outperformed the market by 1 percentage point/year versus 0 percentage points outperformance for stocks in the highest rank by Sell-side Earnings Revisions
There is only mild correlation between stocks in the highest monthly quintiles of Earnings Sentiment and Sell-side Analysts' Revisions
A long strategy based on selecting the top 20% of stocks by Earning Sentiment resulted in Yearly Return of 6.6% and Return per Holding Period of 0.54%
A long-short strategy based on selecting the top/bottom 20% of stocks by Earnings Sentiment resulted in Yearly Return of 1.5%
Remarks & Conclusion
In 2002-2018 Earnings Sentiment was a better predictor of monthly stock performance than Sell-side Earnings Revisions for large cap stocks.
Source: Empirical Research Partners, "Stock Selection: Research and Results", August 2018