February 2023

ONE MORE THING...

  • Hedge funds *should* like bear markets.

  • Does shorting “work”?


Inflation Expectations

As expected, investors remain focused on the Fed’s boxing match with inflation.

Naturally, inflation has an expectations component - that’s why everyone was so concerned early in the inflation cycle that a too-slow Fed would let inflation set its roots, consumers would expect higher prices, suppliers would raise costs of raw materials, businesses would raise prices in anticipation of future inflationary wage and supply chain pressures, and the Fed’s work of taming inflation would become even more difficult than current economic conditions would suggest. Consequently, efficient forward-looking markets price in the Fed’s current and projected future approaches to tame inflation.

This also helps explain how “good news can be bad for markets” (and inversely how bad news can be good for markets). When otherwise good news like expanding economic activity would be cheered, the Fed sees expanding economic activity as inflationary and must raise rates even more than forecasted and currently priced into markets to suppress those economic activities, which of course is bad for markets.

To be clear, it’s not that good news is actually bad - it’s that good news that is better than expected is bad relative to what is currently priced in.

So when the U.S. Labor Department reports that in January nonfarm payrolls increased by 517,000 jobs and the unemployment rate declined to a 53 year low of 3.4%, policymakers like Secretary Janet Yellen respond by trying to shape sentiment:

"You don't have a recession when you have 500,000 jobs and the lowest unemployment rate in more than 50 years,"

and

"What I see is a path in which inflation is declining significantly and the economy is remaining strong."

It is the investment manager’s job, then, to appropriately discount such responses - and plenty of investment managers are doing just that. CNBC’s Mike Winters details how Investors and the Fed disagree about where interest rates are headed in 2023:

As part of the central bank’s December forecast, 17 out of 19 Fed officials projected rates above 5% through 2023, with two of them suggesting rates higher than 5.5%.

However, despite Powell’s comments Wednesday, investors don’t seem to believe that the Fed will stick with elevated rate hikes in 2023. As of early Wednesday afternoon, traders projected a 97.2% probability that interest rates will be 5% or less by next December, according to the CME’s FedWatch tool.

Broadly, this reflects a belief that either a recession is imminent and the Fed will need to start cutting rates in 2023 to stimulate the economy, or that inflation will go down quickly within the year, negating the need for elevated interest rates, says Liz Young, head of investment strategy at SoFi, an online bank.

Interestingly, the Fed’s approach of “talking tough” in setting expectations has the potential to be a highly effective tool in the Fed’s toolbelt. If everyone expects a recession, then rational participants will change their behavior, the economy will cool down, and the Fed could in fact engineer the proverbial “soft landing” while enacting fewer rate hikes. After all, Fed-induced recessions are fundamentally different from balance sheet recessions.

When recessions are expected and forecasted, it is not just consumers but also state governments that can prepare to “ride out the storm”. The Wall Street Journal’s David Harrison notes that unlike in many prior recessions, state governments have amassed record-high reserves that can be deployed to soften the effects of an impending recession:

States will hold an estimated $136.8 billion in rainy-day funds this fiscal year, according to the National Association of State Budget Officers, up from $134.5 billion a year earlier, when they represented 0.53% of gross domestic product, the highest in records going back to 1988. This year’s figure would represent roughly 12.4% of their total spending.

Moody’s Analytics estimates 39 states have the reserves necessary to offset all the revenue expected to be lost in a relatively mild recession. Four more are within striking distance.
...
A broader measure of state reserves, which includes all unspent funds, whether stored in specified rainy day funds or not, will amount to 24.7% of total spending this fiscal year, down from 31.7% in 2022, according to NASBO forecasts. By contrast, states held just 8.9% on average between 2000 and 2020.

In 2007, right before the financial crisis, states held less than 5% of expenditures in rainy-day funds. That was one reason they had to make deep cuts during and immediately after the 2007-2009 recession.

And yet, sticking with the theme of this segment, is this good news of states’ rainy day funds actually bad news for markets? The Fed’s war with inflation may require even higher interest rates due to sizeable rainy day funds allowing consumers to sustain their economic activity longer without making the behavioral changes that the Fed is trying to induce. As David Kelly of J.P. Morgan Asset Management notes, there is evidence of this also:

“The personal saving rate in the fourth quarter was 2.9% compared to 7.3% a year earlier, as consumers increased borrowing to try to maintain pandemic-era spending levels. However, this is likely not sustainable and slower credit growth, along with slower job growth, should imply less spending growth in the year ahead.”

Effective management requires managing volatility in the short term to outperform over the long term. As policymakers navigate our economic ship through these choppy inflationary seas, I remain confident that the Fed’s toolkit works and inflation will Likely trend back to the Fed’s 2% long term target. That confidence holds regardless of whether the Fed engineers a severe recession to get there sooner, a mild recession to get there more slowly, or a soft landing that slowly avoids a technical recession altogether.

ChatGPT learns from financial advisors

Is Artificial Intelligence truly “intelligent”?  

No, AI is not truly intelligent.  Machine learning, which is one of several areas within Artificial Intelligence, is a form of “intelligence” that results from throwing massive troves of data at a computer then letting the computer find patterns and relationships in the data.  But that brute force processing relies on a very limited and narrow interpretation of “intelligence”.  

Machine learning was behind the breakthrough in self-driving cars.  Programming a computer to drive, including explicit instructions on all of the various circumstances, judgment calls, weather conditions, and countless other contingencies that human drivers routinely do was too complex to be directly programmed - perhaps programmers could eventually get close to replicating human driving, but the potential for exceeding humans’ driving proficiency remained out of reach.  The breakthrough came in flipping the problem - developers packed test cars with loads of monitors, sensors, and cameras, humans drove them around town while data recorded everything that was happening, then algorithms searched for relationships in the data that characterized what it meant to “drive a car”.  What conditions existed when human drivers slammed on the brakes, activated their turn signals, or safely changed lanes?  

If this data-intensive process sounds inefficient, it is inefficient - for humans.  But these days computational power is cheap, and this data-intensive process turned out to be far more efficient and effective than directly programming a car how to drive.  

Chess shares a similar insight.  I love Garry Kasparov’s TED Talk Don’t fear intelligent machines.  Work with them., which I also previously featured in Likelyhoods.  As the World Chess Champion from 1985-2000, Kasparov’s story includes the moment when computers (Deep Blue) eclipsed the best human (Kasparov) in chess mastery.  Deep Blue was able to calculate much faster and deeper to defeat Kasparov through brute force, but to call this “intelligence” inappropriately diminishes the complexity of human cognitive processing.  Perhaps artificial intelligence will one day match or exceed what the human brain is capable of, but until then this form of “intelligence” is most definitely “artificial”.  

Which brings me to ChatGPT.  Of course investors want to know whether ChatGPT is better than humans at investing!  The traditional active management industry does not need even more attention drawn to their systematic underperformance relative to passively managed indexing - but is ChatGPT capable of outperforming the S&P 500 Index, or at least performing better than human managers?  Bloomberg’s Sam Potter and Katherine Greifeld Asked ChatGPT to Make a Market-Beating ETF. Here's What Happened:

What happens when you ask the hottest AI tool in the world to design an ETF that can beat the US equity market? It tells you the same thing every frustrated stock manager does.

[ChatGPT’s answer]:

“It is not possible for me to design an ETF that will beat the US stock market because the stock market is unpredictable and past performance does not guarantee future results.  Additionally, the composition of an ETF should be based on a thorough analysis of the market and a strategy for achieving specific investment goals, which I do not have the information or expertise to provide.  

It is important to consult with a financial advisor before investing in any ETF or stock to understand the potential risks and rewards of the investment and ensure that it aligns with your investment goals and risk tolerance.”

This more-or-less canned output perhaps should not surprise us since, by design, machine learning algorithms will naturally replicate the training data sets on which they were built.  

Nicely, ChatGPT seems to be aware of its own limitations.  When asked to “explain whether artificial intelligence can pick stocks better than a human”, ChatGPT said: 

Artificial intelligence (AI) has the potential to analyze large amounts of data quickly and identify patterns that may not be immediately apparent to humans.  This can make it a valuable tool for identifying potential investment opportunities.

However, whether AI can pick stocks better than a human depends on the quality and effectiveness of the AI algorithms and models used.  AI-powered investment strategies can be complex, and the performance of an AI-powered stock picker will depend on the underlying data, models, and assumptions used.  

It’s important to note that AI is not infallible and can make mistakes, especially if the data used to train the model is biased or incomplete.  Additionally, AI-powered investment strategies are not immune to market fluctuations and can be affected by market conditions just as human-driven strategies can.

Another important factor to consider is that AI-powered investment strategies are still in their infancy and it is not yet clear how they will perform over the long term.

In summary, AI has the potential to be a valuable tool for identifying investment opportunities, but it is not a magic solution that can always pick stocks better than a human.  It’s important to conduct thorough research, including consulting with a financial advisor, before making any investment decisions.

Of course this dynamic of technology threatening to replace humans is not new.  Not too long ago Robo Advisors broke into the investment management world promising lower fees for the same or better investment advice than human managers could provide.  My general sense is that Robo Advisors serve a role for some investors, but their fees are still sufficiently higher than passive management and many investors still have questions about the quality of robo advice relative to human managers so as to raise questions about their value.

Let’s close with the inevitable backtesting and overfitting that you know is being done as-we-speak with ChatGPT.  Not surprisingly, some backtests are beating the S&P 500 Index over a few months, but just as with traditional active managers against passive indexing, the Law of Large Numbers catches up with them and their performance regresses back to the lower-than-passive-indexing mean:

There are already artificial intelligence-guided investments all across Wall Street — including in the ETF arena — and some are beating the market right now. 

A current standout is the AI Powered Equity ETF (ticker AIEQ), a $102 million vehicle that has returned about 9.9% in 2023 through Wednesday, compared with 4.7% for the S&P 500 Total Return Index.

It entered 2023 with a mixed allocation. Major holdings currently include home furnishing firm RH, Las Vegas Sands Corp., sustainable power company Constellation Energy Corp. and JPMorgan Chase & Co.

Returns analysis shows that the ETF’s consumer discretionary holdings — including shares in the likes of Caesars Entertainment Inc., Kohl’s Corp. and the meme-stock favorite GameStop Corp. — have been the biggest driver of performance this year.

However, expand the time horizon and AIEQ’s market-beating prowess comes undone. Since its 2017 inception, the ETF has handed investors about 41%, according to data compiled by Bloomberg. The S&P 500 Total Return Index has delivered more than 72% in the same period. 

Running thousands of AI backtests then cherry picking the handful of tests that beat the S&P 500 Index is not a reliable process.  In fact it is no different than the countless day trading systems that find a narrow backtested performance edge, but then fail to sustain that supposed edge into the future.  Investors beware.  

I love technology and love the potential that ChatGPT is offering every industry, including investment management.  But please, let’s understand and remain diligent about its limitations because ChatGPT is not (yet) truly “intelligent”.


ONE MORE THING…

The information and opinions contained in this newsletter are for background and informational/educational purposes only.  The information herein is not personalized investment advice nor an investment recommendation on the part of Likely Capital Management, LLC (“Likely Capital”).  No portion of the commentary included herein is to be construed as an offer or a solicitation to effect any transaction in securities.  No representation, warranty, or undertaking, express or implied, is given as to the accuracy or completeness of the information or opinions contained herein, and no liability is accepted as to the accuracy or completeness of any such information or opinions.  

Past performance is not indicative of future performance.  There can be no assurance that any investment described herein will replicate its past performance or achieve its current objectives.

Copyright in this newsletter is owned by Likely Capital unless otherwise indicated.  The unauthorized use of any material herein may violate numerous statutes, regulations and laws, including, but not limited to, copyright or trademark laws.

Any third-party web sites (“Linked Sites”) or services linked to by this newsletter are not under our control, and therefore we take no responsibility for the Linked Site’s content. The inclusion of any Linked Site does not imply endorsement by Likely Capital of the Linked Site.  Use of any such Linked Site is at the user’s own risk.

Previous
Previous

March 2023

Next
Next

January 2023