Evidence-based investing research
Value Investing Strategy (Strategy Overview)
Allocations for November 2026 (Final)
Cash TLT LQD SPY
Momentum Investing Strategy (Strategy Overview)
Allocations for November 2026 (Final)
1st ETF 2nd ETF 3rd ETF

Monthly News Sentiment Predicts Stock Market Returns?

Steve LeCompte | | Posted in: Sentiment Indicators

Guide Market Indicators Sentiment

Key Insight

Uhl finds Thomson Reuters news sentiment outperformed LEI in predicting DJIA returns (negative sentiment strategy: 48% vs 8% buy-and-hold in 2010; monthly data 2003-2010; 12 out-of-sample trades).

Does news lead the stock market? In his September 2011 paper entitled “Reuters Sentiment and Stock Returns”, Matthias Uhl tests whether aggregate Thomson Reuters news sentiment (feeling, opinion or emotion evoked while reading a Reuters news article) predicts stock market returns at a monthly frequency. He aggregates monthly sentiment by summing individual articles coded as evoking positive (+1), neutral (0) or negative (−1) sentiment and assesses the relationship between aggregate sentiment and stock market returns via regressions. He considers the effects of positive and negative news separately, and compares the predictive power of news sentiment to that of the Conference Board’s Leading Economic Index (LEI) as a macroeconomic composite. Using Thomson Reuters news sentiment scores for over 3.6 million high-frequency news articles applicable to the U.S. stock market and monthly data for the Dow Jones Industrial Average (DJIA) and LEI spanning 2003 through 2010, he finds that:

  • Over the available sample period, Thomson Reuters news sentiment predicts future stock market returns more effectively than LEI at a monthly horizon.
  • Negative news sentiment is more strongly predictive of stock market returns than positive sentiment.
  • While most pronounced after one month, the power of news sentiment to predict stock market returns persists for several months.
  • A trading strategy that goes long (short) DJIA each month during 2010 when a rolling inception-to-date forecast projects an increase (decrease) in the index the next month beats a buy-and-hold return for 2010. Specifically:
    • The buy-and-hold return for DJIA during 2010 is 8% (monthly Sharpe ratio 0.48).
    • A timing strategy based on negative news sentiment produces a gross return of 48% (83% monthly win rate and gross Sharpe ratio 3.6).
    • A timing strategy based on positive news sentiment produces a gross return of 16% (50% monthly win rate and gross Sharpe ratio 1.26).

In summary, evidence suggests that investors may be able to exploit Thomson Reuters aggregate news sentiment for timing the broad U.S. stock market at a monthly horizon.

Cautions regarding findings include:

  • The reported 2010 DJIA buy-and-hold return (8%) does not match a simple calculation based on 2009 and 2010 index closes (11% before dividends).
  • In-sample testing in the study includes different combinations of variables and optimization of lag intervals, thereby introducing data snooping bias. Findings for the optimal scenario therefore likely overstate reasonable expectations.
  • Out-of-sample testing of multiple trading strategies also introduces data snooping bias, which tends to be particularly misleading for a small sample (12 trades). An independent test involving ten iterations of a simple model that randomly goes long or short DJIA during each of the 12 months of 2010 produces three iterations with gross returns much higher than the annual return of 11% (with one iteration over 30%).
  • Analyses do not account for a lag between measured month and availability of data to a trader. For example, the Conference Board releases LEI for a given month about three weeks after the end of the month. The Conference Board often revises previous-month measurements, making the lag for as-revised data well over a month (and thus not tradable during the month after the measured month). There may be no material lag for Thomson Reuters news sentiment measurements.
  • Trading strategy returns are gross, not net. Including trading frictions, the costs of shorting as signaled (including payment of dividends) and the amortized cost of using the Thomson Reuters service would reduce reported returns.
  • Analyses are linear. There may be interesting (and stronger) non-linear aspects to the relationship between news sentiment and stock market returns.
See “Short-term News Premium for Individual Stocks” for a summary of research on the application of the Thomson Reuters news sentiment to individual stocks at a daily frequency.

Market Indicators guide

More on sentiment

  • Testing a Proxy for the Goldman Sachs Panic Index Premium Claude-estimated Goldman Sachs Panic Index values (pixel-reconstructed from a chart) showed high readings (above 8) followed by 1.2% average next-week S&P 500 returns vs…
  • AI-simulated CFO Sentiment Premium GPT-5.4-simulated CFO optimism correlates with actual Duke-Federal Reserve CFO Survey responses (LLM mean 56.6 vs actual 60.1; regression coefficient over six standard errors from…
  • CFO U.S. Economic Sentiment and Stock Market Returns Premium CFO average U.S. economic sentiment shows little ability to predict S&P 500 returns; correlations with next-four-quarter returns are very weak and negative, while change…
  • How Are Sentiment-driven ETFs Doing? Premium An equal-weighted average of three sentiment ETFs (BUZZ, MOOD, NEWZ; monthly rebalanced) underperformed VTI (1.0% vs 10.7% CAGR; MaxDD -51% vs -25%; sample periods…
  • Goldman Sachs Panic Index Proxy Premium A VIX-percentile Panic Index Proxy (trailing 2-year rank, 0–100) shows weak lead-lag association with daily S&P 500 returns, with very high and very low…

All 309 articles in the Market Indicators guide

Guru Grades

1,206 predictions by 42 market gurus since 2022: 41% right. Always saying stocks will go higher: 62%.

As of the Nov 18, 2026 close

  1. 1 Ed Yardeni A 71%
  2. Always bullish B 62%
  3. 2 Jim Paulsen B 61%
  4. 3 Sam Stovall B 58%
  5. 4 Adam Turnquist B 58%
  6. 5 Ryan Detrick B 56%

Full leaderboard ›