“Prepare for the unknown by studying how others in the past have coped with the unforeseeable and the unpredictable.”
~ George S. Patton
The chart above displays the S&P 500 average versus the level of media uncertainty around past (and present) U.S. presidential elections. The yellow line is the average news and social media uncertainty level. The uncertainty average includes all presidential elections back to 2000. The white dashed line is the average S&P 500 return around all presidential elections since 1928. In the longer lower plot, the red line shows our 2020 level of U.S. uncertainty coming into the Trump-Biden match-up.
In 2020, uncertainty is higher than any other election. The “uncertainty collapse” (decline in the yellow line) starting 2-3 weeks before the elections (1) correlates with a rally in the S&P 500 and (2) happens when the media (and others) believe they know the likely outcome.
“Yes”, you say, “that’s a cool chart, but it’s just cherry-picked” (overfit) to current circumstances.
And our response is, “yes, you’re right, it is.” But that doesn’t mean it’s not useful.
Behavioral research finds that uncertainty inhibits financial risk-taking behavior. Perhaps the decline in uncertainty depicted above could fuel a stock market rally (or at least it will no longer keep risk-takers out of equities).
Today's newsletter looks at some of our recent research like the above, predictive models, and gives a few election insights using our unique psychological data set. We post our original research on Twitter and LinkedIn - please follow us there for timely updates.
We also hear a fair degree of skepticism – can media tone actually predict anything in real life? We’ve traded on this data previously with positive results, but that's not entirely convincing.
Over a year ago our team finalized a predictive model for the top 3,000 U.S. stocks. That model was formally launched as the StarMine MarketPsych Media Sentiment model. Every day it ranks the top 4,000 U.S. stocks with a single score ranging from 1 to 100. That score is determined by a machine learning model that pattern matches current media scenarios for each stock to the past. If a stock shows media trends that historically drove other prices higher, then it will have a higher ranking among its peers, towards 100. If sentiments are similar to those that preceded past declines, then it is ranked closer to 1.
The 30-day performance of stocks ranked in the top decile (ranks 90 to 100) is plotted alongside the performance of stocks ranked in the bottom decile (1 to 10) in the chart above. There are more than 350 stocks in each decile portfolio, and the rankings are updated once per month. The spread between the two deciles is plotted in the blue line.
The post-launch performance is consistent with history, with a Sharpe ratio of 1 for the spread between the top and bottom deciles. More remarkably, the returns are uncorrelated with traditional factors like analyst ratings and price momentum. Each theme tracked has a characteristic influence on prices – some lead to underreaction (trends - those that are boring or routine) while others lead to overreaction (reversals – those that are dramatic or headline-capturing). There are many interrelated patterns.
“There are only patterns, patterns on top of patterns, patterns that affect other patterns. Patterns hidden by patterns. Patterns within patterns.
If you watch close, history does nothing but repeat itself.
What we call chaos is just patterns we haven't recognized. What we call random is just patterns we can't decipher. what we can't understand we call nonsense. What we can't read we call gibberish.”
~ Chuck Palahniuk, Survivor
We’ve been running a small proof-of-concept portfolio on the model since the beginning of January 2020. The first day of each month it buys the top 10 ranked stocks and shorts the bottom 10. We call it “the Balanced Fund” (it is absolute return / dollar-neutral and we give profits to mental health charities). That market-neutral portfolio of the extremes is up 40% this year.
And speaking of predictions, in May 2018 we published a newsletter on public Anger towards companies and the buying opportunity it creates. This was picked up in the FT, and in January 2019 we launched it as a small pilot portfolio in the U.S. It is a long-only strategy, buying stocks with extremely high levels of public anger, and it is beating the S&P 500 by 10+% since launch. We also give profits to charity from this strategy, with all proceeds going to anger management-related causes (RISE in 2020).
After writing the 2018 newsletter and launching the portfolio, we discovered the story is not so simple. When we look across thousands of stocks, high anger actually has a negative impact. What we published in 2018 was based on extreme, headline-grabbing, explosive anger, often correlated with a stock price selloff – investors overreact to such extreme anger and prices ultimately rebound. However, across thousands of stocks, simmering public anger is corrosive and leads to stock price underperformance.
Our team performed a cross-sectional study of relative Anger levels in the media and tracked monthly returns over time for the top 4,000 U.S. stocks. As you can see above in the image, the stocks provoking the highest percentage of media anger had the lowest returns over time.
One valid question coming out of this research is, “Is the high level of social Anger in the U.S. the past few years holding back the U.S. stock market? Will there be a surge when (and if) it finally abates? Or alternatively, does this anger place the U.S. on the path of long-term economic decline?”
Since our last newsletter I’ve relocated back to California, and our world-class quants are in the process of relocating to our Singapore office. Tot Ziens to Amsterdam! We had a fantastic time there, and we’ll be back.
For full disclosure, we’ve traded five portfolios in our history. Fortunately, each time was profitable and beat the intended benchmark except for one – in January 2020 this year we launched a long-only “Positive” Fund. We closed it at the end of April 2020 because it was underperforming the S&P 500 and inconvenient to execute. In hindsight the Positive Fund would have recovered if we had simply stuck with it. Sometimes persistence is key...
It’s nerve-wracking to publicly reveal predictive models, since many/most fail after a time. Part of the reason we work with emotional and thematic data is the likelihood of finding universal and generalizable patterns of risk-taking behavior. Value investing may not make sense when interest rates are low, momentum is eroded by overcrowding. But the investor buying and selling in response to good and bad news is likely to be consistent and influential on prices over time … as long as humans remain the primary decision makers in financial markets.
Please reach out to us at [email protected] to learn more about how our products can help you profit in various global markets.
DISCLAIMER Your use of this the content contained in this publication is at your own risk. Opinions expressed by MarketPsych, LLC ("MarketPsych") are based on sources believed to be reliable and are written in good faith, but no representation or warranty, expressed or implied, is made as to their accuracy or completeness. MarketPsych does not receive compensation of any kind from any companies that may be mentioned in MarketPsych. Any opinions expressed are subject to change without notice. MarketPsych is not responsible for errors or omissions, or responsible to keep any information up to date or to correct any past information. Any recommendations contained in this material are for educational and information purposes only and should not be construed as a solicitation to enter into an investment advisory relationship or to purchase any security or other financial instrument. MarketPsych does not provide customized or personally tailored advice or recommendations through this publication. Advice and trading signals displayed contained in this material are of a generic nature and are not tailored to the specific circumstances of any visitor or subscriber. You should use any information gathered from this publication only as a starting point for your own independent research and decision making process. Any specific investment recommendations or advice contained or referred to in this material may not be suitable for all investors. You should carefully consider whether any specific recommendation is suitable for you in light of your financial condition. You are advised to conduct your own due diligence and seek advice of a qualified professional when it comes to making investment decisions for your own account. The risk of loss in trading securities and other financial instruments can be substantial. No trading program can offer the potential for profit without a corresponding risk of loss. The ability to withstand losses and to adhere to a particular trading program in spite of trading losses are material points which can adversely affect investor performance. You may sustain a total loss of you investment. Past performance is not necessarily indicative of future results. MarketPsych is not responsible for the success or failure of your investment decisions relating to information or services presented herein. MarketPsych reserves the right to refuse its services to anyone, either current subscriber or potential subscriber, with or without any reason or for no reason. MarketPsych reserves the right at any time to modify this disclaimer and risk disclosure without notice.