Sentiment Indicators

Can investors get a trading edge from CME FedWatch, which tracks probabilities of changes to the Federal Funds Rate (FFR) at future FOMC meetings based on the prices of 30-day Fed Funds futures contracts? In their January 2025 paper entitled “Watching the FedWatch”, flagged by a subscriber, Stefano Bonini, Shengyu Huang and Majeed Simaan compare FFR forecasts from a simple model based on CME FedWatch to conventional model forecasts based on Fed Funds futures. They conduct statistical backtests of forecast accuracies during May 1994 through March 2024 (232 scheduled FOMC meetings). They then compare economic values of the two forecasts via two trading strategies that, 30 days before each scheduled FOMC meeting from the end of 2009 through 2023:

1. If the forecast is for a rate cut or no change (a rate increase), takes a long (short) position in Fed Funds futures contracts set to expire in the month of the next FOMC meeting.
2. If the forecast is for a rate cut or no change (a rate increase), takes a long (short) position in iShares Core U.S. Aggregate Bond ETF, AGG. After release of the actual rate decision, if the forecast is wrong, they close the AGG position.

Using daily values of specified variables over the ranges stated above, they find that: Keep Reading

“Daily Global Investor Sentiment” discusses the risk-on/risk-off (RORO) index as a measure of global investor risk appetite, with the underlying dataset publicly available. Can investors exploit this dataset for short-term timing of investments in stocks (risk-on) and government bonds (risk-off)? To investigate, we relate future daily returns for SPDR S&P 500 ETF Trust (SPY) and iShares 20+ Year Treasury Bond ETF (TLT) to daily RORO index levels. After rationalizing RORO index measurement days and SPY/TLT trading days, we consider simple lead-lag regressions to measure linear effects. We then compute next-day SPY/TLT returns by ranked tenth (decile) of RORO index levels to assess non-linear effects. Using daily RORO index levels as available (downloaded on 1/29/25) and daily total (dividend-adjusted) returns for SPY and TLT during 5/9/2003 through 1/28/25, we find that: Keep Reading

Can a multifaceted measure of investor sentiment convincingly predict returns? In their November 2024 paper entitled “Risk-on/Risk-off: Measuring Shifts in Investor Sentiment”, flagged by a subscriber, Anusha Chari, Karlye Stedman and Christian Lundblad explore risk-on/risk-off (RORO) as the variation in global investor risk taking behavior. Their RORO index captures time-varying investor risk appetite as the first principle component of daily changes in proxies for four aspects of investor risk: (1) advanced economy credit risk; (2) advanced economy equity market volatility risk; (3) funding conditions (liquidity) risk; and, (4) currency/gold risk. The proxies are:

• Credit – change in the ICE BofA BBB Corporate Index Option-Adjusted Spreads for the U.S. and the Euro Area, plus the U.S. BAA corporate/10-year U.S. Treasury note yield spread.
• Equity volatility – additive inverse of total returns on the S&P 500, STOXX 600 and MSCI Advanced Economies indexes, plus associated changes in VIX and VSTOXX.
• Liquidity – average change in the G-spreads for 2-year, 5-year and 10-year U.S. Treasury notes, along with the change in the TED spread, the LIBOR-OIS spread, and the bid-ask spread on 3-month U.S. Treasury bills.
• Currency/gold – growth rate of the trade-weighted U.S. Dollar Index against currencies of other advanced economies and the change in gold price.

Using daily values of these proxies during mid-2003 through early 2024, they find that: Keep Reading

Do some stocks entail emotional relationships that alter investor perceptions of risk and return? Is the effect exploitable? In their November 2024 paper entitled “Investor Emotions and Asset Prices”, Shehub Bin Hasan, Alok Kumar and Richard Taffler develop and test a measure of the emotional state of the market and assess its implications for individual stocks. Specifically, they each month:

1. Use a bag-of-words approach encompassing 295 emotion words to construct a market-level emotion index as the ratio of emotion words to total number of words in newspaper articles about the S&P 500 Index.
2. Estimate for each stock an emotion beta by regressing monthly excess returns versus the market emotion index over the last 60 months.
3. Sort stocks into tenths (deciles) based on last-month emotion beta and compute monthly value-weighted returns of the decile portfolios.

Using 65,825 news articles about the S&P 500 Index from 21 national and local newspapers, monthly returns and firm/stock characteristics for a listed U.S. stocks and monthly returns for various stock factors during January 1990 through September 2022, they find that:

Keep Reading

Are broad measures of public sociopolitical sentiment relevant to investors? Do they predict stock returns as indicators of exuberance and fear? To investigate, we relate S&P 500 Index return and 12-month trailing S&P 500 price-operating earnings ratio (P/E) to the percentage of respondents saying “yes” to the recurring Gallup polling question: “In general, are you satisfied or dissatisfied with the way things are going in the United States at this time?” Since individual polls span several days, we use S&P 500 Index levels for about the middle of the polling interval. To calculate market P/E, we use current S&P 500 Index level and most recently available quarterly aggregate operating earnings for that time. Using Gallup polling results, S&P 500 Index levels and 12-month trailing S&P 500 operating earnings as available during July 1990 (when polling frequency becomes about monthly) through October 2024, we find that: Keep Reading

Does investor fear level predict U.S. Treasury instrument returns? In their September 2024 paper entitled “Fear in the ‘Fearless’ Treasury Market”, Tianyang Wang, Yuanzhi Wang, Qunzi Zhang and Guofu Zhou examine how investor fear relates to future returns on U.S. Treasuries. They define bond risk premiums by duration as annual returns in excess of the 1-year interest rate. To measure investor fear level, they employ Thomson Reuters MarketPsych Indices (TRMI), which apply natural language processing to assess investor fear, positivity, negativity, optimism, pessimism, trust, stress, surprise, credit risk and volatility from 42,000 news and 800 social media inputs. They construct a bond market Fear Index by combining these assessments and suppressing noise via a 12-month moving average. They decompose the Fear Index by duration (short-term vs. long-term), depth (intense vs. mild) and source (news vs. social media). They compare their bond Fear Index with other sentiment metrics and examine its import globally. Using end-of-month prices for 1-year to 5-year zero-coupon U.S. Treasury notes, contemporaneous TRMI outputs and data for other potential bond return predictors during January 1998 through December 2022, they find that:

Keep Reading

What variables best explain increases and decreases in Cyclically Adjusted Price-to-Earnings ratio (CAPE or P/E10)? In their August 2024 paper entitled “Analyzing Changing ‘Investor Exuberance’: The Determinants of S&P Composite Index Total Return CAPE Changes”, C. Krishnan, Jiemin Yang and Xiyao Tan apply the following three techniques to investigate which of 42 potentially explanatory variables relate most strongly to changes in CAPE:

1. Linear regression with principal component analysis.
2. Least Absolute Shrinkage and Selection Operator (LASSO) regression analysis, which shrinks some regression coefficients to zero, thereby identifying the most important independent variables.
3. Elastic net, which combine approaches of LASSO and Ridge regression to distill the most important independent variables.

Using monthly values for CAPE and the 42  potentially explanatory variables during February 2000 through December 2019, they find that: Keep Reading

A subscriber asked for corroboration of a claim that the “Best Companies To Work For” (BCTWF) substantially beat the overall stock market. To investigate, we:

• Compile the employee survey-based top 10 BCTWF winners for 2014 through 2023 (10 years, so 100 companies).
• Optimistically assume winner lists are available by the end of March each year (in fact, it appears to be early April).
• Filter out private companies, leaving 46 BCTWF with publicly traded stocks.
• Calculate annual returns for each of these 46 BCTWF stocks from the end of March in the year they win to the end of the next March.
• Each year, form equal-weighted (EW) BCTWF portfolios and calculate average annual April-through-March gross returns.
• Compare annual BCTWF EW strategy gross performance to that of Invesco QQQ Trust (QQQ) as a benchmark.

We focus on gross average annual return, standard deviation of annual returns, gross annual Sharpe ratio, compound annual growth rate (CAGR) and maximum drawdown (MaxDD) based on annual data as key performance metrics. We use the yield on 1-year U.S. Treasury bills (T-bill) as of the end of each March to calculate Sharpe ratios. Using annual dividend-adjusted BCTWF and QQQ returns and annual 1-year T-bill yields from the end of March 2014 through the end of March 2024, we find that: Keep Reading

Can pattern recognition software, combined with news sentiment, identify profitable short-term stock trades? In their July 2024 paper entitled “Technical Patterns and News Sentiment in Stock Markets”, Markus Leippold, Qian Wang and Min Yang test the ability of pattern recognition software (convolutional neural network) to find profitable technical reversal patterns within U.S. and Chinese stock candlestick charts. They consider four pairs of technical patterns: double tops and bottoms; head and shoulders and inverted head and shoulders; broadening tops and bottoms; and, triangle tops and bottoms. They use Bollinger bands to find local maximums and minimums, with the standard deviation multiplier set at 1.1 based on parameter tuning. They augment pattern recognition with news sentiment from Refinitiv for U.S. stocks since 2003 and from Tonglian for Chinese stocks since 2014 during the 10 trading days around each pattern. They identify combined tops as double tops, head and shoulders, broadening tops or triangle tops coupled with negative news and combined bottoms as double bottoms, inverted head and shoulders, broadening bottoms or triangle bottoms coupled with positive news. They first consider each technical pattern as an independent event and measure abnormal returns for holding intervals of 1, 5, 10, 21 and 42 days after a signal. They then examine the performance of a portfolio of events for a 1-day holding interval. They use U.S. stock data from 1992 to 1999, enhanced via two data augmentation strategies, for pattern recognition software training and validation. They then apply the trained software to U.S. stock data from 2000 through 2021 and Chinese stock data from 2005 through 2021. Combined pattern-sentiment test periods are shorter based on availability of sentiment data. Using price series for all U.S. common stocks and Chinese A-shares and news sentiment data as described through 2021, they find that: Keep Reading

Does the combination of an artificial intelligence (AI)-generated financial news sentiment with a complex financial stress metric generate good stock market timing signals? In their April 2024 paper entitled “Mixing Financial Stress with GPT-4 News Sentiment Analysis for Optimal Risk-On/Risk-Off Decisions”, Baptiste Lefort, Eric Benhamou, Jean-Jacques Ohana, David Saltiel, Beatrice Guez and Thomas Jacquot devise and test a risk-on/risk-off strategy for stock market timing. The strategy combines:

• Stress Index (SI): based on VIX, TED spread, Credit Default Swap index and realized volatilities across major equity, bond and commodity markets, all normalized and then aggregated by category. Overall SI is the average of category results, rescaled to fall between 0 and 1.
• News sentiment signal: 10-day moving average of ChatGPT 4 assessments of the sentiment (1 for positive or 0 for negative) in Bloomberg daily market summaries.

They consider six strategies:

1. Benchmark (or Long Only) – buy and hold the index, with constant volatility scaling to match the final (retrospective) volatility of an active strategy.
2. VIX – weight the stock index according to VIX, with times of stress indicated by VIX above its 80th percentile.
3. SI – weight the stock index according to the value of SI as described above.
4. News – weight the stock index according to the ChatGPT 4 news sentiment signal.
5. SI News – weight the stock index according to the product of SI and News.
6. Dynamic SI News – because SI News either significantly outperforms or underperforms SI alone during subperiods, each month weight the stock index according to either SI alone or SI News, whichever has the better Sharpe ratio over the past 250 trading days at the end of the prior month.

They test the strategy on the S&P 500 Index, the NASDAQ 100 Index and an equal-weighted combination of these two indexes plus the Nikkei 225, Euro Stoxx 50 and Emerging Markets indexes. They assume trading frictions of 0.2% of value traded. Using daily values of all specified inputs during January 2005 through December 2023, they find that: Keep Reading