CFA Institute

Foreign exchange (FX) markets are shaped by liquidity fluctuations, which can trigger return volatility and price jumps. Identifying and predicting abnormal FX returns is critical for risk management and trading strategies. This post explores two advanced approaches that allow investment professionals to better understand and anticipate shifts in market conditions. By integrating liquidity metrics with predictive algorithms, investors can gain deeper insights into return behavior and improve risk-adjusted decision-making. The first approach focuses on outlier detection, where robust statistical methods isolate periods with exceptionally large price movements. These detected outliers are then predicted using machine learning models informed by liquidity metrics, alongside key macroeconomic indicators. The second approach targets liquidity regimes directly, employing regime-switching models to differentiate high-liquidity from low-liquidity states. Subsequent return analysis within each regime reveals how risk is magnified in lower-liquidity environments. Observed patterns in major currency pairs suggest that periods of reduced liquidity coincide with abnormal price behavior. Researchers like Mancini et al. and Karnaukh et al. have demonstrated that liquidity risk, often measured through bid–ask spreads or market depth, is a priced factor. Others, such as Rime et al., highlight how liquidity and information proxies can improve FX forecasting. Building on these findings, there are two possible ways to tackle abnormal returns by leveraging machine learning methods and liquidity indicators. Tackling Abnormal Returns Outliers The first approach is to treat abnormal weekly returns, i.e., outliers, as the primary target. Practitioners could collect weekly returns of various currency pairs and apply either simple robust methods like the median absolute deviation (MAD) or more sophisticated clustering algorithms like density-based clustering non-parametric algorithm (DBSCAN) to detect outlier weeks. Once identified, these abnormal returns can be forecast by classification models such as logistic regression, random forests, or gradient boosting machines, which make use of liquidity measures (bid–ask spreads, price impact, or trading volume) as well as relevant macroeconomic factors (e.g., VIX, interest rate differentials, or investor sentiment). The performance of these models can then be evaluated using metrics such as accuracy, precision, recall, or the area under the ROC curve, ensuring that the predictive power is tested out of sample.  Liquidity Regimes The second approach shifts the emphasis to the identification of liquidity regimes themselves before linking them to returns. Here, liquidity variables like bid–ask spreads, trading volume, or a consolidated liquidity proxy are fed into a regime-switching framework, sometimes a hidden Markov model, to determine states that correspond to either high or low liquidity. Once these regimes are established, weekly returns are analysed conditional on the prevailing regime, shedding light on whether and how outliers and tail risk become more likely during low-liquidity periods. This method also gives insight into transition probabilities between different liquidity states, which is essential for gauging the likelihood of sudden shifts and understanding return dynamics more deeply. A natural extension might combine both approaches by first identifying liquidity regimes and then predicting or flagging outliers using specific regime signals as input features in a machine learning setup. In both scenarios, challenges include potential limitations in data availability, the complexity of calibrating high-frequency measures for weekly forecasts, and the fact that regime boundaries often blur around macro events or central bank announcements. Results may also differ when analysing emerging markets or currencies that typically trade at lower volumes, making it important to confirm any findings across a variety of settings and apply robust out-of-sample testing.  Ultimately, the value of either approach depends on the quantity and quality of liquidity data, the careful design of outlier or regime detection algorithms, and the ability to marry these with strong predictive models that can adapt to shifting market conditions. Key Takeaway Navigating FX market volatility requires more than traditional analysis. Liquidity-aware models and machine learning techniques can provide an edge in detecting and forecasting abnormal returns. Whether through outlier detection or liquidity regime modeling, these approaches help investors identify hidden patterns that drive price movements. However, data quality, model calibration, and macroeconomic events remain key challenges. A well-designed, adaptive framework that integrates liquidity dynamics with predictive analytics can enhance investment strategies and risk management in evolving FX markets. source

The Four Pillars of Investing, Second Edition: Lessons for Building a Winning Portfolio. 2023. William J. Bernstein. McGraw Hill Professional. In The Four Pillars of Investing, Second Edition: Lessons for Building a Winning Portfolio, William J. Bernstein, a retired neurologist and the cofounder of the investment management firm Efficient Frontier Advisors, provides a comprehensive guide that offers important insights and practical strategies for creating and maintaining a successful investment portfolio. The book, first published in 2002, gives investors a strong foundation in financial principles. Bernstein sets out four key pillars that serve as the bedrock: theory, history, psychology, and business. These pillars together function like the four legs of a chair and are the guiding principles for making good investment decisions. The first pillar, theory, includes comprehending the underlying concepts and principles that lead to successful investing. Bernstein discusses the need to create a well-diversified portfolio that strikes a balance between risk and return, tailored to individual financial goals, time horizon, and risk tolerance. He explores the intricate relationship between risk and reward, encouraging investors to thoroughly assess their risk appetite before making investment decisions.  The second pillar, history, stresses the importance of analyzing past market trends and historical data because history provides invaluable insights into the behavior of financial markets. History is my favorite of the four pillars. In my opinion, investors should spend more time analyzing financial history to understand what is possible in deriving their views on financial markets, instead of listening to “experts.” Based on historical events, including market booms/busts and recessions, the author illustrates the cyclical nature of markets and highlights the importance of a long-term investing approach. He discusses the implications of market efficiency for retail investors while advocating diversified portfolios as opposed to relying on market timing or individual stock selection strategies.  The third pillar, psychology, highlights the impact of human behavior on investment decisions since the presence of emotional biases can lead to irrational decision making. Bernstein discusses various biases and provides strategies for investors to overcome them. Keeping a disciplined approach to investing and avoiding emotional reactions to short-term market fluctuations are key messages that Bernstein provides throughout the book. Bernstein encourages investors to focus on long-term goals and to develop an investment plan based on solid principles while avoiding emotional decisions driven by noise or short-term trends.  The fourth pillar, business, explores individual companies and their financial performance. Investors should conduct thorough research and gain a deep understanding of the businesses in which they choose to invest. The author stresses the importance of investing in undervalued assets, as well as the impact of fees and expenses on investment returns. He emphasizes the need to minimize costs because they can significantly erode investment performance over time. Bernstein advises investors to seek low-cost investment options that offer broad market exposure at a lower cost than that of using actively managed funds. Although the investment content in magazines, newspapers, social media, and market strategist interviews should be largely ignored, Bernstein recommends reading the Economist’s finance section and listening to the authors of academic papers referenced in this book on YouTube or podcasts, such as Eugene Fama, Zvi Bodie, and Robert Shiller. He supports his pillars with practical examples, case studies, and historical data, making the content accessible and understandable. The Four Pillars of Investing has received numerous accolades for its comprehensive approach and focus on evidence-based strategies. However, critics have argued that it may be too technical for beginner investors and overlook the possible benefits of active investing.  Private wealth investment professionals can use this book as a way to convey some basic investment concepts to individual clients who are not already familiar with them. Although the author argues that most brokers and advisers occupy the lowest rung in the hierarchy of investment knowledge, these same investment professionals can play a critical role in helping individual investors manage around their own psychology by “staying the course” and not overreacting to short-term fluctuations. This can be an important role played by brokers and advisers because the failure of just one leg of the chair can lead to the demise of the entire investment strategy.   In summary, The Four Pillars of Investing is an important tool for investors looking to design a more successful investment portfolio. Investors can make better financial decisions by comprehending the four pillars of theory, history, psychology, and business. This book highlights the importance of disciplined investing and a long-term diversified approach to managing risk and achieving financial goals. Because of its insights and practical guidance, this book remains a critical resource for those investors trying to navigate the complex world of investing.  If you liked this post, don’t forget to subscribe to Enterprising Investor and the CFA Institute Research and Policy Center. All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer. Professional Learning for CFA Institute Members CFA Institute members are empowered to self-determine and self-report professional learning (PL) credits earned, including content on Enterprising Investor. Members can record credits easily using their online PL tracker. source

When China restricted exports of gallium and germanium in 2023, markets were reminded that supply chains can be disrupted. These metals may not be household names, but they are critical to semiconductors, defense systems, and renewable energy, which is why the restrictions drew immediate market attention. Investors are again turning to supply chain resilience as a portfolio concern. Rare earth elements sit in the same category as gallium and germanium. Embedded in electric vehicles, advanced weaponry, and clean energy infrastructure, rare earth elements represent one of the few asset themes where geopolitics directly drives market outcomes. That reality was underscored in July, when the United States backed MP Materials, its only active rare earth miner, with a multibillion-dollar package including equity, loans, and a 10-year price floor on neodymium and praseodymium. The deal, discussed further in Winston Ma’s Enterprising Investor analysis of a potential US sovereign wealth fund, shows how policy is moving from rhetoric to concrete capital commitments. For investors, the right question isn’t whether rare earths can “beat the market.” It’s whether they can provide diversification and resilience in moments when traditional portfolios are vulnerable. A Portfolio Framing: Rare Earths as a Stress Hedge To evaluate this, I built a Maximum Sharpe Ratio portfolio using five ETFs: REMX – Rare Earth & Strategic Metals LIT – Lithium & Battery Technology ITA – Aerospace & Defense GLD – Gold (geopolitical hedge) IEF – U.S. Treasuries (defensive anchor) The goal was not to design a market-beating strategy, but to evaluate whether rare earth exposures add portfolio resilience. I used monthly returns from January 2018 to July 2025, a 36‑month rolling covariance matrix, and quarterly rebalancing. The results: Annualized Return: 11.45% vs. 14.53% (S&P 500) Volatility: 21.95% vs. 17.19% Sharpe Ratio: 0.43 vs. 0.73 If judged solely on Sharpe ratio, the portfolio underperformed broad equities. But this misses the real point: rare earths tend to outperform during geopolitical shocks and supply chain disruptions, precisely when traditional portfolios are most at risk. For investors, the practical takeaway is to test rare earths alongside other diversifiers, such as commodities, infrastructure, or defense equities, in a satellite sleeve. When Rare Earths Shine Looking at recent episodes of stress and transition highlights how rare earths can function as a hedge when traditional portfolios stumble. 2019 United States–China Trade Dispute: During the 2019 tariff standoff, rare earth and defense ETFs advanced even as the S&P stumbled. This divergence highlighted their value as a hedge against policy-driven supply chain risks. 2020–2021 EV Adoption Rally: As electric vehicle demand accelerated, lithium and rare earth exposures surged ahead of the market. For investors, this underscores their potential to capture secular growth trends while adding diversification. 2023 Export Controls: When China restricted exports of gallium and germanium, rare earth themes drew renewed attention and outperformed. The episode showed how policy shocks can create “thematic alpha” precisely when traditional markets are vulnerable. These bursts illustrate the real value: rare earths function as a shock absorber. They won’t replace equities, but they can provide a counterweight when macro risks flare. Figure 1. Practical Applications Thematic Diversification: Use rare earths as a satellite allocation that complements big secular themes: electrification, defense modernization, and the clean energy transition. These exposures can give portfolios targeted access to structural growth trends. Geopolitical Risk Premium: Recognize that policy shocks, not just market cycles, can drive returns. Export bans, tariffs, and supply disruptions often move rare earth markets independently of equities, giving investors a rare source of true diversification. Portfolio Construction:  Test rare earths as a 5% to 10% sleeve within a diversified portfolio. Pair them with gold and Treasuries to balance risk. The goal isn’t to outperform equities, but to add resilience when equities are stressed. Key Takeaways Rare earths are not a silver bullet, but they are a geopolitical hedge that investors can’t ignore. Traditional risk metrics (Sharpe ratio) understate their value: non-correlation and tail events. For allocators, the right framing is resilience, not return chasing. In a world where supply chains are vulnerable, rare earths are more than a commodity story. They are a portfolio strategy for managing geopolitical risk. The author declares no conflicts of interest. This article is based on publicly available ETF pricing data (2018 to 2025). It does not constitute investment advice and is intended solely for educational purposes. source

As financial advisers, clients often come to us with various questions about GDP, unemployment, interest rates, consumer consumption, and how these numbers can affect the market and their investments. I like to be prepared and have the current figures ready for my clients as well as the context to help answer their questions. Lately, clients have noticed the rising costs across many of their expenses: groceries and rent to name a couple. Naturally, they may be frustrated and turn to us to help them understand what’s going on. Why is everything more expensive? What’s causing record-high inflation? How do the US Federal Reserve’s interest rate hikes help address this? Such discussions require that we have more than a quick stat or two at the ready. There is a lot of context we may need to fill in to help explain the current situation. We might have to sit down and explain the many in-between correlations, relationships, and effects of rising prices. What is really happening in the economy right now? How will central banks try to solve it? Can they? Here are a few tips to approach these conversations with clients: 1. Define Inflation First off, it may help to explain to clients what inflation is and why it matters in the long term. Put simply, inflation is the increase in the prices of goods and services. Deflation, on the other hand, is when these prices decline over time. So inflation raises the cost of living in an economy. This means that, over time, it takes more money to buy the same items and the consumer’s purchasing power declines. To be sure, consistent, incremental inflation is necessary for a healthy economy. If inflation is too low, that indicates a low demand for goods and services and can lead to a potential economic slowdown. However, inflation also becomes a problem when it is too high. Left unchecked, sustained high inflation can slow the economy and erode savings. This is why we need to work closely with our clients to help them find ways to sustain their purchasing power over time. 2. Explain How We Got Here The Consumer Price Index (CPI), published monthly by the Bureau of Labor Statistics, is the principal barometer of US inflation. The CPI stayed mostly flat in July versus June after gas prices declined for 57 straight days. But year-over-year, prices are up 8.5%. Food prices have been a key culprit: They are up almost 11% over last year. That poses a burden to many families. So, clients may ask, how did we even get to this point? Causes for inflation vary, but they tend to be products of the economic principles of supply and demand. While there are other variations, economists typically categorize inflation into two core concepts: Demand-pull: The demand for goods and services increases, but the supply does not keep pace. Cost-push: The supply of goods and services falls, but the demand for them does not. Today’s persistent inflation has no one single cause. Rather, multiple factors in the global economy contribute to it. According to research from the Federal Reserve Bank of San Francisco, supply factors are responsible for about half of the recent rise in inflation. So, what does that mean? Supply-chain issues created a shortage of goods and materials. This was exacerbated when many factories temporarily halted production in China due to the country’s zero-COVID policy. Meanwhile, trillions of dollars in US government stimulus propelled a robust recovery from the pandemic-fueled economic crisis and, in turn, increased both income and demand. Record low US unemployment and a tight labor market brought on wage growth. Then, the Russia-Ukraine war reduced the global supply of oil, wheat, and other commodities. 3. Explain What the Fed’s Rate Hikes Have to Do with This Why and how do interest rate hikes correlate to lowering inflation? The Fed has a dual mandate to promote maximum employment and stable prices. If it seems like inflation is driving up prices too quickly, the Fed will raise interest rates to try and contain it by increasing the cost of borrowing (e.g. credit cards, mortgages, etc.). This in turn reduces demand, which could lead to lower prices. But the Fed will also lower rates when it wants to spur economic activity. For example, in 2008, the discount rate was set to zero. We were in a financial crisis — a really bad one. To stimulate consumer consumption and inject liquidity into the economy, the Fed lowered rates so people would borrow to buy goods and services, start businesses or increase inventories. This is how it works in theory: More consumption leads to more spending, which leads to more growth, more people to hire, more paychecks cashed, and, again, more consumption. Today, by raising interest rates, the Fed wants to increase the cost of credit. That tends to make people less willing to borrow and, in turn, less willing to spend. For example, a client may decide to buy a new house with a 3% mortgage, but a 5% mortgage may push it out of their price range. As interest rates on savings accounts rise, more people may be encouraged to put their money in the bank.  The thought process goes something like this: higher rates mean a tighter and more limited money supply. Consumers will therefore spend less. Higher rates can “cool off” the economic landscape. To go back to basic economic theory: less demand means lower prices. 4. Help Clients Manage the Impact Everyone has different circumstances, priorities, and long-horizon goals. This is why it’s important for our clients to have a long-term financial strategy that aligns with their personal goals. Inflation can affect day-to-day expenses, but it also has implications on long-term planning. This is why we need to periodically review their allocations with them. Clients may ask if they should adjust their portfolio right now. And the truth is there isn’t one “right” answer for everyone. Inflation affects every sector differently. We need to talk to our clients

For more on artificial intelligence (AI) applications in investment management, read The Handbook of Artificial Intelligence and Big Data Applications in Investments, by Larry Cao, CFA, from CFA Institute Research Foundation. ChatGPT has launched a new era in artificial intelligence (AI). The chatbot built by OpenAI and powered by the GPT-3 and GPT-4 families of large language models (LLMs) responds to natural language prompts much like a very well-informed human assistant and has continuously evolved with the introduction of GPT-4 and ChatGPT APIs and plugins. Other tech giants haven’t sat idly by. Google and NVIDIA, among others, have shown their commitment to the rapidly evolving technology by announcing a series of innovative generative AI (GenAI) services in recent months. Indeed, each week it feels like the AI industry is experiencing a year’s worth of progress. But what does it mean for investment management? How will all the ChatGPT- and LLM-related developments affect how investment professionals work? ChatGPT: An Overview ChatGPT is an AI language model developed by OpenAI using a technique called reinforcement learning from human feedback (RLHF) that processes natural language prompts and provides detailed responses based on human input. GPT stands for Generative Pretrained Transformer architecture. It is a type of GenAI that can produce new data based on the training data it has received. The leap from natural language processing (NLP) to natural language generation represents a significant advancement in AI language technology. The model pre-trains on vast amounts of data to learn how to respond quickly to queries. For example, GPT-3 has over 175 billion parameters. GPT-4 has even more. Nevertheless, both models are limited by their training data’s cutoff date and cannot incorporate new and time-sensitive information in real time. The transformer architecture is a deep learning technique applied by both ChatGPT, to extract and analyze textual data, and the Bidirectional Encoder Representations from Transformers (BERT) language model, developed by Google. The different components of the GPT architecture work in synchrony to achieve better outcomes. ChatGPT Learning Methods ChatGPT is a conversational AI model built on the GPT series, either GPT-3.5 or GPT-4, for use in conversational applications. Fine-tuned on conversational data, it can better generate relevant, engaging, and context-aware responses. The GPT model is first trained using a process called “supervised fine-tuning” with a large amount of pre-collected data. Human AI trainers provide the model with initial conversations between a questioner and an answerer. This process is like personal training for an AI assistant. After this, the model undergoes reinforcement learning (RL), which involves creating a reward mechanism and collecting comparison data consisting of two or more model responses that are ranked by quality. To further refine the model, OpenAI collected data from conversations between AI trainers and the chatbot. It randomly selected a model-written message, sampled several alternative completions, and had AI trainers rank them. Using these reward models, OpenAI fine-tuned the model with Proximal Policy Optimization (PPO) and performed several iterations of this process to improve the model’s performance. ChatGPT’s Limitations ChatGPT’s shortcomings are well-known. It may provide plausible sounding but incorrect or nonsensical answers due to the limitations of RL training. OpenAI acknowledges that there is currently no single source of truth for RL training and that ChatGPT is designed to answer questions to the best of its abilities rather than leave them unanswered. The quality of its responses depends on the question’s phrasing and the information ChatGPT has learned through supervised training. ChatGPT does not have values in the same way that humans do. While it has been trained to ask clarifying questions to ambiguous queries, it often guesses at the user’s intended meaning. OpenAI has made efforts to prevent ChatGPT from responding to harmful or inappropriate requests, but the LLM may exhibit biased behavior at times. That’s why it’s crucial to avoid illegal, unethical, aggressive, or biased suggestions and forecasts. ChatGPT can also be verbose and overuse certain phrases, often stating that it’s a “large language model trained by OpenAI.” The training data used to develop the model has biases and over-optimization issues, and trainers may prefer longer answers that appear more comprehensive. While ChatGPT and other language models are generally excellent at summarizing and explaining text and generating simple computer code, they are not perfect. At their worst, they may “hallucinate,” spitting out illogical prose with made-up facts and references or producing buggy code. LLM Scaling Laws, Few-Shot Learning (FSL), and AI Democratization Potential GPT models offer unique features that distinguish them from BERT and other mainstream AI models and reflect the evolution of AI applications for NLP. Like GPT, BERT is a pre-trained model that learns from vast amounts of data and is then fine-tuned for particular NLP tasks. However, after pre-training, the models diverge. BERT requires fine-tuning with task-specific data to learn task-specific representations and parameters, which demands additional computational resources. GPT models employ prompt engineering and few-shot learning (FSL) to adapt to the task without fine-tuning. With GPT-4’s pre-training data, GPT models can generate appropriate outputs for unknown inputs when given example tasks. Scaling laws, which Jared Kaplan, et al., have highlighted, are among GPT models’ essential features. Performance improves as model size, training dataset size, and the computing power used for training increase in tandem. Empirical performance has a power-law relationship with each individual factor when not bottlenecked by the others. GPT-4 follows this law and can achieve high performance without fine-tuning, sometimes exceeding previous state-of-the-art models. Moreover, scaling laws work with other media and domains, such as images, videos, and mathematics. The features of GPT models represent a paradigm shift in AI development away from traditional models trained for each specific task. GPT models do not require large local computational resources or additional training data, and tasks are tackled through FSL rather than model fine-tuning or retraining. However, a limited number of players — Google, Amazon, and the like — could control the supply of large language models (LLMs) on cloud computing platforms, which could create an oligopoly that hinders the democratization of AI

Introduction Before 2020, the threat of a global pandemic shutting down the world economy was not a top-of-mind concern for most investors. Pandemics were nothing new, of course, but no outbreak in recent history had suggested anything near the magnitude of COVID-19. SARS had broken out in 2002 and Ebola in 2014, for example, but both were contained rather quickly, and their corresponding toll in economic disruption and human lives hardly hinted at what COVID-19 would bring. Before 2022, even fewer investors saw a third world war as a serious possibility. And while such an outcome is still very unlikely, the Russian invasion of Ukraine has increased the odds. A limited nuclear exchange, let alone a global nuclear war, would have enormous consequences for humanity as a whole to say nothing of the securities markets. Nevertheless, it is worth considering what a simple escalation of the current conflict might entail. Intuitively, war implies economic damage and falling stock markets. But so does a global pandemic. Yet the S&P 500 was significantly higher one year after COVID-19 went global. Which raises the question: How do stocks — specifically sectors and factors — perform during times of war? Stock Performance in Major Wars To answer this question, we analyzed the US stock market, which has the best dataset for individual securities and indices. In particular, we evaluated the performance of US stocks during three of the country’s most significant wars: the US Civil War, from 1861 to 1865; World War I, from 1917 to 1918; and World War II, from 1941 to 1945. These three wars had major implications for the US population and economy. Countless thousands died. Infrastructure was both built and demolished. Major cities were razed. Some parts of the economy collapsed while others boomed. Yet despite all the war-related misery and destruction, the US stock market expanded in both the US Civil War and World War II. Only in World War I did it suffer a net decline.  US Stock Market Performance in Major Wars Sources: Finominal and StooqReturns are based on close prices unadjusted for dividends. Factor Performance in Wartime Unfortunately, stock analysis suffers from something of a recency bias: The further back in time we go, the poorer the quality of securities data. As a consequence, the remainder of our analysis will focus on World War II–related data. The long–short performance of the size, value, and momentum factors was positive between 1941 and 1945, albeit just barely for momentum, according to data from the Kenneth R. French Data Library. The value factor generated a CAGR of 16%, and the size factor 11%. Theoretically, such returns would have generated attractive diversification benefits for a traditional portfolio inasmuch as they represent excess returns from long–short portfolios. But practically, these returns were calculated before transaction costs and at a time when shorting stocks was an inefficient process at best. Thus, these results need to be evaluated with a grain of salt. Performance of Factors (Long–Short) in World War II Sources: Finominal and Kenneth R. French Data Library Performance of Industries during Wartime But what about sector performance? Did any one in particular stand out during World War II? By analyzing the 49 industries from the Kenneth R. French Data Library, we zeroed in on the top and bottom 10. We expected the top 10 to be dominated by industries that were deeply involved in the war effort — heavy machinery and defense companies, for example. But the best-performing sector was actually printing and publishing, followed by alcoholic drinks and personal services.  Performance by Industry in World War II: The Top 10 Sources: Finominal and Kenneth R. French Data Library The worst-performing industries held some surprises as well. Though they generated positive returns, tobacco companies came in dead last. This creates something of a paradox given the beer and liquor sector’s 723% windfall. Did people drink more and smoke less during the war? It’s also hard to understand why steel, chemical, and aircraft companies wouldn’t have fared better. We don’t have any answers except to say that financial markets are full of surprises and never do what’s expected of them. Which is why active management is so difficult and creates so little value. Performance by Industry in World War II: The Bottom 10 Sources: Finominal and Kenneth R. French Data Library Asset Class Performance in Wartime How did bonds fare relative to stocks during World War II? Equities generated the highest nominal returns between 1941 and 1945, but short-term and long-term Treasuries as well as corporate bonds all yielded positive returns, although after inflation, only corporate bonds had positive real returns.  Of course, the United States and its allies won the war. The Axis powers financed themselves by selling government bonds to their citizens. When they lost the war, those became worthless.  World War II Performance: By Asset Class, 1941 to 1945 Source: Finominal and Professsor Aswath Damodaran Further Thoughts While investors made money with stocks in two out of the three largest US wars, this analysis is backward- rather than forward-looking. It is difficult to imagine a third world war that doesn’t involve the deployment of nuclear arms. Yet these weapons could destroy much of human civilization let alone the capital markets. Few investment options have much appeal in such a cataclysmic scenario. Maybe productive farmland in such faraway destinations as Australia or New Zealand would be viable options, although even here, the goal would be more capital preservation than capital growth. For more insights from Nicolas Rabener and the Finominal team, sign up for their research reports. If you liked this post, don’t forget to subscribe to the Enterprising Investor. All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer. Image credit: ©Getty Images / gece33 Professional Learning for CFA Institute Members CFA Institute members are empowered to self-determine and self-report professional learning (PL) credits earned, including content

The Canadian pension plan system has long been lauded for its robust returns and resilience, especially in the face of volatile markets. One key aspect contributing to this success is the incentive frameworks that Canadian pension funds use to attract and retain top investment talent. In this post, we explore how the largest Canadian pension funds have structured their compensation plans to drive exceptional outcomes while managing market fluctuations and ensuring long-term sustainability. The insights here are derived from Southlea’s 2024 Asset Management Survey. The Canadian model provides a framework for asset managers globally. Large Canadian pension funds manage most of their assets in-house, with the eight largest (the Maple 8) managing 80% of their investments internally. Key Components of Canadian Pension Plan Incentive Structures Incentive designs are the “secret sauce” in Canada’s pension plan system’s success. The incentive designs used by these organizations follow a multi-layered approach to ensure that individual, divisional, and overall corporate objectives are aligned. Some common components of these incentive frameworks include: Corporate Metrics: These typically include performance measures tied to overall investment returns but also consider broader organizational objectives like talent development and client satisfaction. Division/Asset Class Metrics: By aligning the incentive structures with specific asset class outcomes, pension plans can ensure that teams are focused on achieving their unique objectives while contributing to the broader goals of the organization. Individual Performance: Pension plans also evaluate individual performance based on both the “what” (e.g., results) and the “how” (e.g., leadership and values). This holistic approach ensures that the right behaviors are incentivized across all levels of the organization. In addition, both absolute and relative performance metrics are used to ensure that compensation aligns with market expectations and benchmarks. This balanced approach encourages investment teams to deliver not just in terms of returns but also in relation to the broader market conditions. Adapting to Market Volatility The past few years have underscored the need for flexibility in incentive design. With market volatility becoming the new normal, Canadian pension plans have been adjusting their frameworks to remain competitive while ensuring they retain their top talent. For example, relative total fund returns are commonly used to measure performance. This benchmark helps to ensure that pension plans are not only generating returns but outperforming the market. However, given the variability of market performance, more sophisticated models are being used to assess relative returns, ensuring that the chosen benchmarks are appropriate and reflective of the organization’s specific investment strategy. Another major adaptation has been the increasing focus on risk metrics. Pension funds are now incorporating additional risk measures into their incentive plans, moving beyond simple return measures. These risk-adjusted metrics, often assessed in consultation with the Chief Risk Officer, ensure that undue risk-taking is penalized and stable, long-term performance is rewarded. Elongating Performance Horizons Canadian pension funds have also adapted their incentive structures by extending performance periods. Historically, many plans have operated with three- to four-year performance windows, but more recently, these horizons have been elongated to five or even seven years. This longer-term approach aligns more closely with the long-term objectives of pension funds, smoothing out the impact of short-term market downturns and ensuring that compensation outcomes reflect sustained performance. Judgment-Based Incentives vs. Quantitative Metrics In a move away from rigid, formulaic compensation structures, many pension funds are now introducing an element of judgment into their incentive decisions. This shift allows for greater flexibility in compensation outcomes, particularly in volatile market conditions where strictly quantitative approaches may lead to skewed results. By allowing for informed judgment, pension plans can ensure that compensation decisions better reflect both the financial and operational realities of the organization. Compensation Trending Down Southlea’s 2024 Asset Management Compensation Survey highlights a notable trend: actual compensation levels for Canadian pension plan employees decreased by about 6% year-over-year, with senior employees seeing even larger declines. This is largely attributable to challenging market conditions, with senior employees — whose compensation is more heavily weighted toward long-term incentives — being the most affected.   All Employees Senior Employees Junior Employees All Investment Asset Classes -6% -11% -3% Private Asset Class -7% -15% -3% Public Asset Class -6% -14% -1% Private asset classes, such as private equity and real estate, saw some of the largest year-over-year declines in compensation, reflecting the challenging conditions in 2023. However, it’s important to note that these trends are not isolated to one pension fund but are consistent across the asset management industry. When looking at specific private asset classes, amongst these senior employees, private equity and real estate pay dropped more significantly compared to natural resources/infrastructure which is reflective of the challenging market conditions of 2023. Below are the year-over-year decreases in actual pay for the senior employees of the following private asset classes: Private Equity: -28% Real Estate: -14% Natural Resources / Infrastructure: -3% A More Balanced Labor Market The Canadian pension sector is also seeing changes in labor market dynamics. The labor market is more balanced between employers and employees than it has been in the recent past, with turnover significantly down and offer acceptance rates significantly up. At median, total turnover decreased by roughly 25% to 8.9% and voluntary turnover rates decreased by approximately 45% to 5.4%. This significant decrease is reflective of the wider market conditions. Many firms across the market have slowed their hiring compared to previous years when they hired large numbers of employees, especially in the aftermath of COVID hiring freezes. When looking at investment jobs, it was interesting to note that the time to offer acceptance and time to start increased year over year, but acceptance rates increased from 95% to 100% at median. This indicates that while it is taking longer to fill these investment roles, the search for these roles is resulting in more success hiring a candidate. It is also worth noting that the number of jobs being filled by internal candidates increased by 5% year-over-year (21% to 26%) and external hiring rates and the use of external recruiters are down.

ChatGPT and other natural language processing (NLP) chatbots have democratized access to powerful large language models (LLMs), delivering tools that facilitate more sophisticated investment techniques and scalability. This is changing how we think about investing and reshaping roles in the investment profession. I sat down with Brian Pisaneschi, CFA, senior investment data scientist at CFA Institute, to discuss his recent report, which provides investment professionals the necessary comfort to start building LLMs in the open-source community. The report will appeal to portfolio managers and analysts who want to learn more about alternative and unstructured data and how to apply machine learning (ML) techniques to their workflow. “Staying abreast of technological trends, mastering programming languages for parsing complex datasets, and being keenly aware of the tools that augment our workflow are necessities that will propel the industry forward in an increasingly technical investment domain,” Pisaneschi says. “Unstructured Data and AI: Fine-Tuning LLMs to Enhance the Investment Process” covers  some of the nuances of one area that is rapidly redefining modern investment processes — alternative and unstructured data. Alternative data differ from traditional data — like financial statements — and are often in an unstructured form like PDFs or news articles, Pisaneschi explains. More sophisticated algorithmic methods are required to gain insights from these data, he advises. NLP, the subfield of ML that parses spoken and written language, is particularly suited to dealing with many alternative and unstructured datasets, he adds. ESG Case Study Demonstrates Value of LLMs The combination of advances in NLP, an exponential rise in computing power, and a thriving open-source community has fostered the emergence of generative artificial intelligence (GenAI) models. Critically, GenAI, unlike its predecessors, has the capacity to create new data by extrapolating from the data on which it is trained. In his report, Pisaneschi demonstrates the value of building LLMs by presenting an environmental, social, and governance (ESG) investing case study, showcasing their use in identifying material ESG disclosures from company social media feeds. He believes ESG is an area that is ripe for AI adoption and one for which alternative data can be used to exploit inefficiencies to capture investment returns. NLP’s increasing prowess and the growing insights being mined from social media data motivated Pisaneschi to conduct the study. He laments, however, that since the study was conducted in 2022, some of the social media data used are no longer free. There is a growing recognition of the value of data AI companies require to train their models, he explains. Fine-Tuning LLMs LLMs have innumerable use cases due to their ability to be customized in a process called fine-tuning. During fine-tuning, users create bespoke solutions that incorporate their own preferences. Pisaneschi explores this process by first outlining the advances of NLP and the creation of frontier models like ChatGPT. He also provides a structure for starting the fine-tuning process. The dynamics of fine-tuning smaller language model vs using frontier LLMs to perform classification tasks have changed since ChatGPT’s launch. “This is because traditional fine-tuning requires significant amounts of human-labeled data, whereas frontier models can perform classification with only a few examples of the labeling task.” Pisaneschi explains. Traditional fine-tuning on smaller language models can still be more efficacious than using large frontier models when the task requires a significant amount of labeled data to understand the nuance between classifications. The Power of Social Media Alternative Data Pisaneschi’s research highlights the power of ML techniques that parse alternative data derived from social media. ESG materiality could be more rewarding in small-cap companies, due to the new capacity to gain closer to real-time information from social media disclosures than from sustainability reports or investor conference calls, he points out. “It emphasizes the potential for inefficiencies in ESG data particularly when applied to a smaller company.” He adds, “The research showcases the fertile ground for using social media or other real time public information. But more so, it emphasizes how once we have the data, we can customize our research easily by slicing and dicing the data and looking for patterns or discrepancies in the performance.” The study looks at the difference in materiality by market capitalization, but Pisaneschi says other differences could be analyzed, such as the differences in industry, or a different weighting mechanism in the index to find other patterns. “Or we could expand the labeling task to include more materiality classes or focus on the nuance of the disclosures. The possibilities are only limited by the creativity of the researcher,” he says.  CFA Institute Research and Policy Center’s 2023 survey — Generative AI/Unstructured Data, and Open Source – is a valuable primer for investment professionals. The survey, which received 1,210 responses, dives into what alternative data investment professionals are using and how they are using GenAI in their workflow. The survey covers what libraries and programming languages are most valuable for various parts of the investment professional’s workflow related to unstructured data and provides valuable open-source alternative data resources sourced from survey participants. The future of the investment profession is strongly rooted in the cross collaboration of artificial and human intelligence and their complementary cognitive capabilities. The introduction of GenAI may signal a new phase of the AI plus HI (human intelligence) adage. source

What does it take to be a successful investor? A healthy balance of technical skill, emotional intelligence, and intentional habits can help. This three-legged stool requires us to derive precision from knowledge and analysis and perspective from curiosity and discipline while developing processes to nudge us toward our fullest potential. The investment community is in constant search of new tools that facilitate this critical process. But as technology across the financial world has evolved exponentially in recent years, one of the most valuable investment tools has been around for millennia: writing. Clear writing and clear thinking go hand in hand. While the exercise may seem trivial, articulating our ideas through words on a page rather than in our heads alone is a revealing experiment: Our views may start the same but often materialize differently. That’s because writing encourages us to slow down, calibrate our thoughts, and test the true health of our ideas. Could writing be the most underrated investment skill? I believe so. Indeed, writing should serve an active role throughout an investment process. Here’s why. 1. Writing reveals what we know versus what we understand. It’s all too easy to think we fully grasp a given investment topic. As we consume information throughout the day, it’s difficult to assess its broader implications while being bombarded by news alert after news alert. Instead of being critical thinkers, we often become “headline experts,” regurgitating catchy fast facts without a deeper awareness. We might know many things but understand far fewer. Geopolitics is a prime example. War, public health, natural disasters, trade policy, the climate, and elections, among other topics, attract no shortage of attention. Our natural tendency is to rush in and immerse ourselves in these topics as they arise to learn as much as possible, gauge potential investment implications, and adapt. The urge to “do something” frequently scatters the investment community on frenetic quests to become experts in extraordinarily complex topics. While being properly informed is a noble goal, we should be careful when we align how well we understand a topic with how much conviction we have in our investment response. Writing helps us bridge this gap and find our blind spots faster. 2. Writing enhances self-awareness. While the quantitative side of investing is challenging enough, the emotional hurdles are often far steeper. Self-awareness is critical if we are to spot our biases and keep our emotions under control. A sound investment process systematically incorporates a series of checks and balances that optimizes our decision making. But seeing exactly where and how we can improve requires additional perspective. In other words, without a mirror, we can’t always tell if something is stuck in our teeth. Writing serves as that mirror by reflecting our mindset in the moment and across time. It creates a healthy emotional distance from ourselves that helps us become more objective and confirm our convictions — and if we need to, floss. 3. Writing improves our ability to discern insight from noise. Writing builds healthy investment research habits. It sharpens our “insight-noise filter” by using an intentional framework to detect helpful information. For an overly basic example, consider mid-2023 headlines celebrating a remarkably strong start to the year for the S&P 500. “Stocks are doing great” might be an easy takeaway. But were they? Just seven companies powered most of the gains. The average stock had hardly budged. So, a theme’s underlying mechanics are often far more nuanced than what appears on the surface. A simple writing prompt to describe the health of the stock market would have quickly offered context. 4. Writing serves as “lane assist” for our investment process. For drivers and investors alike, when we don’t keep our eyes on the road, the likelihood of veering off course skyrockets. It’s all too easy for hot topics to quietly steer us away from the disciplined course we mapped out for these very situations. After all, even the best investment process is only as effective as how well we follow it. In May 2023, an impending “US Debt Ceiling Crisis” evoked concerns about whether the Treasury would default on its own debt and send the global economy into a tailspin. News networks dedicated entire segments to guessing the probability of collaboration among the different factions in Washington, DC, based on the daily schedules of political figures. Major headlines blared widespread fear about systemic risks for weeks on end. This time was going to be different even though the debt ceiling had been raised 77 times since 1960. It was understandably hard to keep focus. But writing helps us home in on our process when it matters most. Structuring intentional prompts as we contemplate new themes gives us a checklist to ensure we are seeing more clearly. Moreover, writing helps us proofread our own ideas and serves as our own second opinion. 5. Writing sheds light on the quality of our decisions. Performance results alone are not enough to gauge the true quality of our investment decisions. Was our analysis sound? Did the results occur for the reasons we expected? Were we right or lucky? Wrong or unlucky? Without considering the input, we’re not fully equipped to assess the output. More importantly, by focusing only on results, we ignore the learning opportunities that can collectively enhance the longer-term impact we aim to achieve. Hindsight may be 20/20, but remembering how we actually thought and felt at any given time can be blurry — unless we have a process to document it. Writing helps us be more intentional about capturing these moments. It creates a time capsule of feedback that provides deeper context and accelerates our ever-evolving learning curves. So, How to Begin? As an investment, writing is well worth its J-curve. As with any fitness routine, patience and effort help build writing muscles. So, here are a few ways to begin: Start small. Consider the timing. Align length with purpose. Pay attention to your emotions. Review periodically. If you liked this post, don’t forget

The following is based on “Redefining the Optimal Retirement Income Strategy,” from the Financial Analysts Journal. Retirement, like life, is fundamentally uncertain. That’s why we need to provide clients with more context about what missing their retirement-income goals might look like and do it in a thoughtful way. In my two previous articles, I explored how retirees tend to have more flexibility in their retirement spending than the conventional models imply and discussed a basic framework to dynamically adjust their spending. Here, I examine how commonly used financial planning metrics — the probability of success, in particular — are flawed and why we should consider other outcomes metrics that can offer additional and better insights into clients’ retirement income situations. The Rise of Monte Carlo Financial advisers often use Monte Carlo projections to demonstrate the uncertainty associated with funding retirement income and other retirement goals. The element of chance, or randomness, is the key differentiator with Monte Carlo projections compared to time value of money calculations and other methodologies. While showing the likelihood that a goal may not be achieved is important, so too is outlining the array of potential scenarios. The probability of success is the most common outcomes metric in Monte Carlo tools and refers to the number of runs, or trials, in which the goal is fully accomplished in a given simulation. For example, if a retiree wants $50,000 in annual income for 30 years, and that goal is achieved 487 times in 1,000 runs, there’s an estimated 48.7% chance of success. Success-related metrics treat the outcome as binary, however, and don’t describe the magnitude of failure or how far the individual came from accomplishing the goal. According to such metrics, it doesn’t matter whether the retiree fails in the 10th or 30th year or by $1 or $1 million dollars. All failure is treated the same. So, a retiree can have a relatively small shortfall yet also a low probability of success, especially when their retirement income goal is primarily funded through guaranteed income and for a relatively extended assumed period, say 30 years. Goal Completion But a financial goal is not a discrete set of pass or fail outcomes. It is a spectrum of possibilities. That’s why adding context about the degree of potential failure is so important. The percentage of the goal that is completed is a critical metric. The chart below illustrates this effect with an assumed goal of $100 a year for 10 years. Percentage Chance that $100 a Year for 10 Years Goal Is Met Courtesy of David Blanchett, PhD, CFA, CFP In runs 1 to 5, for example, the goal is only partially met. The percentage varies across the five simulations, but each run constitutes a “failure” based on success-related metrics. Other metrics tell a different story. Using the average goal completion, 90% of the goal is covered, on average, while success rates indicate a 50% chance of success. Though based on identical data, these two metrics give very different perspectives about the safety of the target level spending. The relatively low success rate suggests reaching the goal is far from assured. But the goal completion score offers a much more positive picture. This is especially important with extended-duration goals like retirement where “failure” is most likely in the final years of the simulation. Diminishing Marginal Utility While goal-completion percentages demonstrate a more colorful perspective on the results of Monte Carlo simulations, they also fail to account for how the disutility, or pain, associated with missing a goal may vary. For example, not funding essential expenses like housing or health care will likely lead to more dissatisfaction than cutting back on travel or other flexible items. The concept of diminishing marginal utility describes this relationship: The pleasure of consuming, or funding, something typically increases, but at a decreasing rate. This may explain why people buy insurance even though it reduces wealth on average. They guarantee that they will be able to fund some minimum level of consumption. Goal-completion percentages can be further modified to incorporate diminishing marginal utility, whereby the implied satisfaction associated with achieving a given level of consumption changes, especially depending on whether the consumption is discretionary or nondiscretionary. I developed a framework for making these adjustments based on prospect theory. These values can be aggregated across years within a given run, and across all runs. This yields a goal-completion score metric that may necessitate much different advice and guidance than modeling based on probability-of-success rates.  Working with What We’ve Got Our industry must deploy better outcomes metrics in financial plans. Such metrics must consider goal completion and more directly incorporate utility theory. To be sure, relatively few instruments accomplish this today, so financial advisers may have to offer improved guidance using the current toolset. Those financial advisers who continue to rely on success rates should dial their targets down a bit. According to my research, 80% is probably the right target. This may seem low: Who wants a 20% chance of failure? But the lower value reflects the fact that “failure” in these situations is rarely as cataclysmic as the metric implies. Clients also need more context around what exactly a bad outcome entails. As financial advisers, we can explain how much income is generated in the unsuccessful trials. How bad are the worst-case scenarios? Will the client have to generate $90,000 at age 95? This is much more meaningful than a success rate and demonstrates just how poorly things could go if they don’t go well. Conclusions The probability of success may be the primary outcomes metric for advisers using Monte Carlo projections, but it completely ignores the magnitude of failure. Success rates can be especially problematic for retirees with higher levels of longevity-protected, or guaranteed, income and for those with more spending flexibility. Alternative-outcomes metrics can help us fill in the gap and ensure we provide reasonable and accurate information to clients to help them make the best financial decisions possible. If you liked this post, don’t forget to subscribe to the Enterprising Investor.