Market Strategy
by Talley Leger, Chief Market Strategist
September 11, 2026
Surviving CAPE Fear: When to Worry About Valuations
“You’re gonna learn about loss.” – Max Cady, Cape Fear (1991)
For those who haven’t seen the movie, that iconic line was delivered with chilling precision by Robert De Niro as the vengeful psychopath, Max Cady, in Martin Scorsese’s 1991 psychological thriller, Cape Fear.
Robert Shiller’s Cyclically Adjusted Price-to-Earnings (CAPE) ratio recently achieved a dizzying 40.3x, soaring well above its 10-year moving average of 32.6x. When staring at that towering multiple, the highest since late-1999 (44.2x), it’s understandable for fearful investors to get spooked by Cady’s terrifying cinematic promise (see the chart below).
Shiller’s CAPE Has Achieved Terrifying Heights
Sources: FRED, Shiller, S&P Global, WCG, 09/9/26. Notes: NBER = National Bureau of Economic Research. SD = Standard deviation.
Valuations Are Meaningless for Near-Term Returns (i.e., 1-5 Years)
Like most if not all valuation metrics, the CAPE is a terrible market-timing tool. Indeed, it’s practically meaningless for predicting near-term stock market performance:
- Take a Deep Breath: If your time horizon is anywhere from zero to five years hence, the R2 between the CAPE and S&P 500 forward returns is in a weak range of 0.03 to 0.26 (see the chart below).
- Upside Risk: Said differently, high valuations alone don’t trigger immediate selloffs. Multiples can stay elevated for years, and selling stocks too early can risk missing out on significant late-cycle compounding as the reluctant buyers are forced into the market for fear of missing out (FOMO).
Valuations Establish Structural Guardrails for Long-Term Returns (i.e., 16 Years)
Sources: FRED, Shiller, S&P Global, WCG, 09/9/26. Notes: S&P 500 16-year nominal total returns (expressed as compound annual growth rates or CAGRs) were regressed against the natural logarithm of the CAPE (i.e., a linear-log regression). Interval of estimation = 1/1947 - 1/1999.
When Do Valuations Matter?
While today’s valuations don’t dictate today’s or even tomorrow’s price action, they can help define the structural guardrails for prospective generational wealth creation:
- When we lengthen our holding period to 16 years, the statistical relationship tightens significantly.
- Based on a 52-year estimation interval (1947-1999), the R2 between the logged CAPE and S&P 500 16-year forward nominal total returns rockets to a stellar 0.86 (see the chart above)!
- Across diverse economic environments – including the low-inflationary 1950s, the stagflationary 1970s and the disinflationary 1980s/1990s – valuations explained an impressive 86% of the variation in 16-year forward returns on stocks (see the chart below)!
Overpaying for Stocks Likely Reduces Their Return in the Long Run (i.e., 16 Years)
Sources: FRED, Shiller, S&P Global, WCG, 09/9/26. Notes: Out of sample period = 1/1999 - 9/2026.
What Valuations Imply About Long-Term Returns
The CAPE has a strong inverse relationship with long-term returns on stocks. Intuitively, overpaying for equities reduces their gains over the long haul, just as underpaying for equities enhances their gains over the long haul. My 16-year linear-log regression model cuts through the short-term noise to provide a sober estimate of distant future outcomes:
- Equation: y = -9.86x + 37.78
- Input: The natural log of the current CAPE (40.3x) is roughly 3.7.
- Output: Plugging that value into the model produces an estimated 16-year forward nominal total return of just 1.3% annualized (see the chart below).
The CAPE Has a Strong Inverse Relationship with Long-Term Returns on Stocks
Sources: FRED, Shiller, S&P Global, WCG, 09/9/26. Notes: Out of sample period = 1/1999 - 9/2026. Direct investments can’t be made in indices. Past performance isn’t a guarantee of future performance.
Robbing Peter to Pay Paul
A 40.3x multiple doesn’t mean the sky’s falling this week. However, it does mean that future returns have essentially been pulled forward. While I doubt that a crash is imminent, bullish investors like us should be statistically prepared for expected nominal total returns on U.S. large-cap stocks to compress into the low single digits over the next decade and a half. By contrast, U.S. small-cap and international stocks are even more compelling based on their valuation merits alone.
Portfolio Strategy
by Jim Worden, CFA®, CMT®, CAIA®, Chief Investment Officer
September 11, 2026
What is AGI and What Does It Mean for Us?
Two weeks ago, I had the privilege of dropping off my son in college on the north shore of Oahu in Hawaii. Unlike other times our family visited the islands, this time we didn’t rent a car at the airport. We instead used Turo, the Airbnb-like car rental company. My Turo rental was a 2026 Tesla Model Y with Full Self-Driving (FSD) included (see image below).
Source: Turo
I had heard a lot about FSD, but I had never been “supervising” behind the wheel of a self-driving car before. It was exhilarating and unsettling at first, but I gradually started to trust it more. My experience led me to contemplate AI more and consider at what point we might reach Artificial General Intelligence (AGI). According to Stanford University’s Human-Centered Artificial Intelligence (HAI), AGI is an AI system with general, human-level (or beyond) ability to learn, reason, and apply knowledge across a wide range of tasks and domains.1 AGI systems would not be limited to performing certain tasks when prompted but could handle different tasks independently.
AGI has been somewhat controversial as there’s no unified or standardized test for what AGI is or isn’t and what is considered human-level intelligence or not. Some profess that AGI is already here. Others say that it will take decades. This will likely be left to the experts and historians to decide, and we will likely only know when AGI has been achieved after the fact.
In my opinion, I believe the best frontier models will likely achieve AGI by 2030. That’s just a few years away. I don’t think this will massively disrupt society or replace millions of jobs overnight. I actually think that many new jobs will be created that didn’t exist before while we see both gradual and more immediate disruption in other areas.
One point to make clear, however, is that AGI does not equal perfection. If we had to wait for every AI model to be perfect at every single task, that would take a very long time. But just because something isn’t perfect doesn’t make it useless. My self-driving car experience is a good example of this. The FSD mode was as cautious or more cautious when merging, changing lanes, and adjusting speed for rain or other obstructions than I was, but the parking wasn’t always perfect. The car did not always see the perfect parking spot initially nor did it park perfectly in between the white-painted parking lines consistently.
Similarly, the best AI models still sometimes hallucinate – making up facts that aren’t true. Other times, the models omit information that is relevant to the prompt that should have been included. Some models are better than others at bringing in general knowledge and applying it to different subjects or tasks. The best models are already starting to show signs of this. Some of the errors mentioned above can be minimized at the prompt level or with the use of a mixture-of-experts model, more active parameters, reasoning modes, or other settings. The models are still not perfect, but they are also good enough to not be considered useless.
So let’s assume for a minute that, for critical tasks that require the highest degree of accuracy – self-driving cars, vertical take-off and landing (VTOL), medical analysis, legal advice, accounting, engineering design and architecture, or physical safety – the models are 99.9% accurate for these tasks, but only deemed 80% accurate for things that are more hypothetical or theoretical. Would these models be good enough to be considered human-level to meet the definition of AGI? If not, where do we draw the line? Will there likely ever be a model that perfectly predicts the weather a month from now? Or predicts the economy or markets perfectly? Likely never. This is where there will likely remain some ambiguity and probably where no one will agree on the day, the month, or even the year that models achieve AGI.
But it doesn’t really matter exactly when or how. I believe what’s important for consumers and investors is to know that eventually we will get there. And it doesn’t need to be apocalyptic with Terminator-style robots replacing humans. There will always be some folks who project how dire things will be and, as is often the case, they will likely be proven wrong after the fact.
When we transitioned away from travel by horse and buggy to rail, to car, and then to plane, there was also a high degree of angst. It will likely be similar with AI and AGI. I think it’s also helpful to try to understand that AI and AGI will solve new problems that we haven’t yet uncovered. There will also likely be some bad actors using AI for nefarious purposes. We’re already seeing that with video, audio, and images that look real but are not. I believe AI and AGI will also help detect or discover these things.
Decryption and cryptography will likely be another area where AI, AGI, and quantum computing could be used to make these things more secure. But, as I mentioned above, we should always keep in mind that AGI does not have to be perfect to be incredibly useful to us. This is where some may confuse or conflate AGI with ASI (Artificial Superintelligence). ASI is the futuristic version of AGI in which robots infused with the smartest and most complex AI models know more than any human and can act independently without being given instructions on what to act on or how to act.
ASI is likely many years away. It also doesn’t need to fit the doom and gloom framework of robots controlling humans. People will disagree as to how far away this is and whether or not we can actually get there. It’s also helpful to remind ourselves that we, the humans, have engineered the AI models and not the other way around. Humans are far from being perfect at any one thing, let alone every single thing that is out there, new or old. AI models and robots are still designed, trained, and governed by humans, and their capabilities and behavior are largely shaped by those choices. That should give us some comfort, even though complex systems may not always behave exactly as intended. The media will, as they always do, try to sensationalize a model that goes rogue. We can question the ethics of design, but often, what is not highlighted is that models are still operating within systems designed and governed by humans.
My wife and I recently watched the Broadway musical, Maybe Happy Ending. It’s a story of two helper robots who fall in love. What’s touching is that they are aware of their own obsolescence, yet their greatest desire is to still valiantly serve their owners and help however they can, doing exactly what they were designed to do. They comically display their awareness of not being perfect but being okay with doing the best that they can in their circumstances.
This type of augmented assistance is more likely, in my opinion, where we are going with AGI and physical AI, including robots, self-driving cars, and other autonomous systems. I will leave the pundits to keep debating ad nauseam the risks and rewards of ASI, knowing that this is still likely a long way off and that having the right regulatory and ethical framework matters more than the capability of the models.
For now, in the next few years we will likely see AI and AGI allow models and physical AI to be more useful in our everyday lives. For example, if autonomous or self-driving cars can reduce the number of automobile accidents by even 30% to 50%, or if AI/AGI models can detect cancer or other diseases earlier, either outcome would be a massive win for society, and we should welcome it with open arms.
Footnotes
- According to Bloomberg data, S&P 500 total returns from 1928 through 2025, looking at January through August data (21.01%) compared to January through December of each year (28.13%), not including 2026 data.
- According to Bloomberg data, S&P 500 total returns from 2009 through 2025, looking at January through August data (17.54%) compared to January through December of each year (26.60%), not including 2026 data.
- According to Bloomberg data, S&P 500 standard deviation of total returns by month from 1928 through 2025. September and October have average volatility of 20.28% (ranked 9th out of 12) and 20.8% (ranked 11th out of 12), respectively.
- According to Bloomberg data, S&P 500 total returns from 1928 through 2025, November returns rank 6th best (1.17%) and December returns rank first (1.93%).
- According to Bloomberg, VIX index average monthly percentage changes from 1997 through 2026 YTD were 9.13%, 8.37%, and 4.88%, respectively, for August, September, and October.
- According to Bloomberg data, the VIX Index average monthly percentage change from 1997 to 2026 YTD was -6.71% for November. The next closest negative month was March at -3.18%.
Definitions
S&P 500: A stock market index tracking the performance of 500 of the largest publicly traded companies in the United States. It serves as a primary benchmark for the overall health of the U.S. stock market.
Operating EPS: A company’s net profit from regular business operations divided by its outstanding shares, excluding one-time gains or losses. It shows how much profit a company generates from its core everyday business.
P/E: A financial metric calculated by dividing a company’s current stock price by its EPS. It shows how much investors are willing to pay for every dollar of the company’s profit.
NBER Recession: A significant decline in economic activity spread across the economy, lasting more than a few months, as officially designated by the National Bureau of Economic Research. It is determined by analyzing factors like gross domestic product, income and employment.
Moody’s Seasoned -Year Baa Corporate Bond Yield: The average interest rate paid on corporate bonds with a 20-year maturity that are rated Baa, which represents medium-grade investment bonds with moderate credit risk. It serves as a key benchmark for corporate borrowing costs.
Natural Logarithm: A mathematical function that determines the exponent to which the constant e (approximately 2.718) must be raised to equal a given number. It is widely used in finance and science to model continuous growth rates.
Standard Error: A statistical metric that measures how much a sample mean is expected to deviate from the true population mean. A lower standard error indicates that the sample data provides a more accurate estimate of the whole population.
R-Square (Coefficient of Determination): A statistical measure that indicates the percentage of variance in a dependent variable that can be explained by an independent variable in a regression model. It ranges from 0 to 1, where higher values show a stronger fit between the data and the model.
Gauss-Markov Theorem: It states that in a linear regression model where the errors have an expected value of zero, are uncorrelated, and have equal variance, the Ordinary Least Squares (OLS) estimator is the Best Linear Unbiased Estimator (BLUE). This means that among all linear, unbiased estimators, the OLS estimator achieves the lowest possible variance for the regression coefficients.
S&P 500 Index: A market-capitalization-weighted index of 500 leading U.S. companies and a widely used measure of U.S. large-cap equity market performance.
VIX Index: The Cboe Volatility Index, which measures the market’s expectation of 30-day forward-looking volatility using S&P 500 Index option prices.
Volatility: The degree of variation in the price or return of an investment over time. Higher volatility generally indicates larger price swings.
Standard deviation: A statistical measure of the dispersion of returns around their average and a commonly used measure of volatility.
Price return: The change in the price of an investment or index, excluding dividends or other distributions.
Total return: The change in price plus dividends or other distributions, assuming reinvestment.
Support: A price level at which buying interest may emerge and potentially slow or stop a decline.
Resistance: A price level at which selling interest may emerge and potentially slow or stop an advance.
Vega: A measure of an option’s sensitivity to a change in implied volatility.
Option writing: The sale of option contracts, which creates contractual obligations for the option seller if the option is exercised.
Swap agreement: A derivative contract in which counterparties agree to exchange cash flows based on specified terms or underlying market variables.
Convexity: The nonlinear relationship between changes in an underlying market factor and changes in the value or behavior of an investment or strategy.
GFC: Global Financial Crisis, generally referring to the 2007–2009 financial crisis.
Artificial intelligence (AI): Computer systems designed to perform tasks commonly associated with human intelligence, including learning, reasoning, pattern recognition, decision support, and content generation.
Artificial General Intelligence (AGI): As used in this article, AI with broad, general capabilities that can learn, reason, and apply knowledge across many different tasks and domains at roughly human-level capability or beyond, rather than being limited to a single narrow task.
Artificial Superintelligence (ASI): A hypothetical form of AI whose capabilities would substantially exceed human cognitive abilities across most or all relevant domains. There is no universally accepted test or timeline for ASI.
Frontier model: An AI model considered to be near the leading edge of current capabilities at the time it is evaluated. Which models are considered frontier models can change rapidly.
Hallucination: An AI-generated output that appears plausible or confident but is inaccurate, misleading, incomplete, or unsupported by the underlying facts.
Prompt: The instructions, question, context, or other input provided to an AI model to guide its response.
Mixture-of-experts (MoE) model: An AI architecture that contains multiple specialized components, or “experts,” and routes a given input through only some of them rather than using every model parameter for every task.
Active parameters: The subset of a model’s parameters that are used during a particular inference step. In some architectures, including many mixture-of-experts models, only part of the model is active for a given input.
Reasoning mode: A model setting or operating approach that allocates additional computation or intermediate processing to a task in an effort to improve the quality of the response.
Physical AI: AI used in systems that perceive, interact with, or act in the physical world, such as robots, autonomous vehicles, and other automated machines.
Autonomous system: A system capable of carrying out some tasks or decisions with limited human intervention while operating within defined objectives, controls, or constraints.
Full Self-Driving (FSD): Tesla’s advanced driver-assistance capability. Despite the name, current Full Self-Driving (Supervised) requires active driver supervision and the driver must be prepared to intervene.
Vertical take-off and landing (VTOL): The capability of an aircraft to take off and land vertically rather than requiring a conventional runway.
Quantum computing: A form of computing that uses quantum-mechanical properties to perform certain calculations. Its potential applications include optimization, scientific simulation, and some areas of cryptography.
Cryptography / decryption: Cryptography is the use of mathematical techniques to protect information and communications. Decryption is the process of converting encrypted information back into a readable form.
Disclosures
This material is provided for informational and educational purposes only and is not intended as investment advice, a recommendation, or an offer or solicitation to buy or sell any security or investment strategy.
Investing involves risk, including the possible loss of principal. Past performance is not indicative of future results. Historical averages, seasonal patterns, and prior market behavior do not guarantee or predict future performance.
Index performance is unmanaged, does not reflect fees, expenses, taxes, or transaction costs, and investors cannot invest directly in an index.
Market and index data are sourced from Bloomberg and are believed to be reliable; however, their accuracy or completeness is not guaranteed and the data have not necessarily been independently verified.
Technical analysis, including references to support and resistance, is subjective and does not assure that any price level will hold or that historical patterns will repeat.
Options, swaps, volatility-linked investments, and other derivative or volatility-based strategies involve additional risks, which may include leverage, liquidity, counterparty, pricing, and volatility risk, and may result in substantial losses.
The views expressed are for informational and educational purposes only and are subject to change without notice.
This material is not intended as, and should not be interpreted as, individualized investment advice or a recommendation to buy, sell, or hold any security, sector, industry, or investment strategy.
References to specific companies, securities, sectors, or industries are for illustrative purposes only and should not be construed as investment recommendations.
Investing involves risk, including the possible loss of principal. Investments in a specific industry or sector may involve greater risk and volatility than more diversified investments.
Past performance is not indicative of future results. No investment strategy can guarantee a profit or protect against loss.
Forward-looking statements, including views about future demand, pricing, supply, or industry cycles, are based on current expectations and assumptions and are subject to risks and uncertainties. Actual results may differ materially.
Data and information are believed to be reliable, but accuracy, completeness, and timeliness are not guaranteed. Source documents should be retained for factual claims, third-party research references, and company-specific data.
Portfolio holdings, allocations, and risk budgets are subject to change based on market conditions, client objectives, and investment guidelines.
The author, firm, clients, or related persons may hold positions in securities mentioned and may buy or sell those securities without notice, subject to applicable policies and regulations.
Securities offered through LPL Financial, Member FINRA/SIPC. Investment Advice offered through WCG Wealth Advisors, LLC, an SEC Registered Investment Advisor. WCG Wealth Advisors, LLC and The Wealth Consulting Group are separate entities from LPL Financial. Index performance is shown for illustrative purposes only and does not predict or depict the performance of any investment. Past performance does not guarantee future results.
All information in this report is believed to be from reliable sources; however, WCG Wealth Advisors, LLC, makes no representation as to its completeness or accuracy.
In general, stock values fluctuate, sometimes widely, in response to activities specific to the companies as well as broad market, economic and political conditions. Stock investing involves risks, including fluctuating prices and loss of principal. Value investments can perform differently from the market as a whole. They can remain undervalued by the market for long periods of time. (135-LPL) International investing involves special risks such as currency fluctuation and political instability and may not be suitable for all investors. These risks are often heightened for investments in emerging markets. (93-LPL)
The fast price swings in commodities will result in significant volatility in an investor’s holdings. Commodities include increased risks, such as political, economic, and currency instability, and may not be suitable for all investors. (122-LPL)
Rebalancing a portfolio may cause investors to incur tax liabilities and/or transaction costs and does not assure a profit or protect against a loss. (28-LPL)
There is no guarantee that a diversified portfolio will enhance overall returns or outperform a non-diversified portfolio. Diversification does not protect against market risk. (26-LPL)
Standard deviation is a historical measure of the variability of returns relative to the average annual return. If a portfolio has a high standard deviation, its returns have been volatile. A low standard deviation indicates returns have been less volatile. (131-LPL)
This is for educational / general purposes only, does not constitute investment, tax or legal advice and should not be relied on as such. This is not to be construed as an offer to buy or sell any financial instruments. Any strategies discussed are not intended to be relied upon as the sole factor in making an investment decision for any individual. As with all investments there are associated inherent risks. Please obtain and review all financial material carefully before investing. All material presented is compiled from sources believed to be reliable and current, but accuracy cannot be guaranteed. The opinions voiced in this material are for general information only and are not intended to provide specific advice or recommendations for any individual. All performance referenced is historical and is no guarantee of future results. All indices are unmanaged and may not be invested in directly. These comments should not be construed as recommendations but as an illustration of broader themes.
Forward-looking statements are not guarantees of future results. They involve risks, uncertainties and assumptions; there can be no assurance that actual results will not differ materially from expectations. In addition, forward-looking statements, including index targets or market scenarios, are hypothetical in nature, reflect current views and assumptions and are subject to change based on market and economic conditions and are not guarantees of future performance. This is a hypothetical example and is not representative of any specific investment. Your results may vary. (88-LPL) Scenario outcomes are illustrative and not predictive. This does not constitute a recommendation of any investment strategy or product for a particular investor. Investors should consult a financial professional before making any investment decisions.
The S&P 500 is a stock market index tracking the stock performance of 500 of the largest companies listed on stock exchanges in the United States. Indexes are unmanaged and cannot be invested in directly. (102-LPL)
Government bonds and Treasury bills are guaranteed by the US government as to the timely payment of principal and interest and, if held to maturity, offer a fixed rate of return and fixed principal value.
The opinions expressed are those of the author as of the date of publication and are subject to change without notice. Statements regarding the timing, development, adoption, capabilities, benefits, risks, or societal effects of AI, AGI, ASI, autonomous systems, and related technologies are forward-looking opinions and expectations. They are inherently uncertain, may prove incorrect, and should not be relied upon as guarantees of future events or outcomes.
This material is provided for informational and educational purposes only and should not be construed as individualized investment, legal, medical, engineering, safety, accounting, or tax advice. It is not an offer, solicitation, or recommendation to buy, sell, or hold any security, strategy, or investment product.
References to Tesla, Turo, Stanford University Human-Centered Artificial Intelligence (HAI), or other companies, organizations, products, or services are for illustrative purposes only and do not constitute an endorsement or recommendation. Product names and trademarks are the property of their respective owners.
Tesla Full Self-Driving (Supervised) is a driver-assistance system and requires active driver supervision. Capabilities and limitations may vary by vehicle, hardware, software version, location, road conditions, and other factors. Drivers should follow applicable laws, manufacturer instructions, and safety requirements and remain prepared to intervene.
The accuracy percentages and potential accident-reduction range discussed in the article are hypothetical examples used to illustrate a concept; they are not representations of measured performance, forecasts, or guarantees. Actual performance and real-world outcomes may differ materially.
References to AI-assisted medical analysis or earlier disease detection are illustrative and are not claims of clinical effectiveness or medical advice. Medical AI applications require appropriate validation, regulatory review where applicable, professional judgment, and human oversight.
AI systems can produce inaccurate, incomplete, outdated, biased, or unsupported outputs. Human review, independent verification, appropriate controls, and attention to the limitations of specific systems remain important, particularly in high-stakes uses.
Publication Date: September 11, 2026
The Wealth Consulting Group
For Public Use in the US
LPL 1173698