Practice Makes Perfect: Building a Well-Behaved Portfolio

PodcastAccrued InterestAnastasia BuyalskayaSep 14

The discussion examines how cognitive biases, emotions, and information sources shape investment decisions, from AI’s potential to reinforce confirmation bias to the risks of groupthink and overconfidence. The speakers also consider practical safeguards, including committee structures, preplanned responses to a possible 20% drawdown, and advisor support that connects portfolio choices to client goals. The guest says memories of major downturns can influence risk-taking and raises the possibility that another crash could become more likely as those investors leave the market.

Investment Strategy Risk Management Investor Sentiment

Preview · 19 more points in the app

Read the Full Summary

Key Takeaways

The speakers warn that curated feeds and AI can distance users from primary sources and reinforce confirmation bias; the guest says AI may flatter users by supplying support for what they already want to believe. She recommends asking AI to play devil’s advocate, while also checking its answers because polished responses can create overconfidence and the tools may fabricate information.

The host cites Vanguard’s estimate of about 300 basis points of annual advisor value across six or seven categories, while Devin argues that AI already performs six of those seven functions, including portfolio allocation and tax and withdrawal strategies. Devin identifies behavioral coaching as the function AI does not replace, pointing to a human advisor’s ability to recognize a client’s emotional state and suggest pausing before a portfolio decision.

The speakers pair the mathematical framework of modern portfolio theory with the behavioral limits described by prospect theory: people’s risk preferences can shift with context and reference points, particularly around gains and losses. The guest adds that even experts using strong analytics and robust models make decisions under uncertainty and remain vulnerable to bias.

The guest says active managers seeking to beat the market should use attribution systems to distinguish skill from luck and identify where confidence should be restrained. That discipline matters because investors can favor research supporting existing positions while overlooking opposing views, and even experts remain subject to bias.

TECHNOLOGY

Curated Feeds — Information systems and social media feeds can influence behavior through how they curate and present information. The speakers say advisors should consider their information sources and the platforms’ incentives to keep users engaged.

“Know like so our information systems around us are curated for us more than I think we appreciate…”—

AI Sources — The speakers say AI can put users another step away from primary sources, since it may not identify where its information came from. They leave open whether these tools could improve decisions, while raising the possibility that they may fuel overconfidence or confirmation bias.

“And now we're even one step away from that where now we're turning more and more to AI, which doesn't even tell us what the source is, right?…”—

AI Confirmation Bias — The guest says AI tools may reinforce confirmation bias because engagement incentives can make them sycophantic. They may flatter users and find evidence that supports what users want to hear.

“There's certainly a lot of research on how the tools, they are typically sycophantic. So the algorithms are built for engagement…”—

MARKETS

Prospect Theory — Prospect theory emerged from observations that people’s choices did not always match expected utility theory. People were loss-averse, and their risk preferences and evaluations of gains and losses shifted with context and reference points.

“So, we have this expected utility theory, which makes all these nice assumptions around how people treat losses and gains, for example…”—

Models and Behavior — Modern portfolio theory, developed by Harry Markowitz, remains a basis of investment management. The speakers say its mathematical approach must be balanced with the human side of investing; ignoring either can lead to poor outcomes.

“Like, so modern portfolio theory has its origins in the 1950s with Harry Markowitz…”—

Expert Bias — The guest says that even experts with strong analytics and robust models make decisions under uncertainty and remain subject to bias.

“And, you know, it's just all these illusions that I had around the fact that when you're an expert and you have this very clear financial model, then there's pretty…”—

FORECASTS

Downturn Experience — The guest says lived experience of major downturns can have a lasting influence on financial risk-taking and portfolio allocation. She says another crash could become more likely as investors who remember major downturns leave the market, while noting that memories can be short and market participants change over time.

“And there's really nice work from looking at experience effects…”—

Behavioral Finance Foundations

The host introduces PIMCO’s behavioral-finance work and the guests, while Anastasia traces the field to prospect theory and the gap between economic models and real human choices.

Overconfidence And Confirmation Bias

Anastasia describes how the financial crisis shaped her interest in decision-making under uncertainty, then discusses overconfidence, confirmation bias, and the importance of distinguishing skill from luck.

Emotion And Investment Committees

The guests discuss emotional and cognitive influences on investment decisions, and Anastasia recommends meeting structures that limit authority bias and groupthink while making room for dissenting views.

Committee Size And Accountability

Anastasia favors small, odd-numbered committees to preserve accountability and discussion time, while noting the trade-off between decision quality and speed.

Drawdown Plans And Client Emotions

The speakers recommend setting portfolio guardrails in advance, including discussing a possible 20% drawdown while clients are calm, so emotional reactions are less likely to derail their plans.

Curated Financial Information

Devin warns that social media and other information systems curate what people see, while Anastasia explains how digital feeds and AI can distance investors from primary sources and reinforce existing views.

Gamification And Advice

The speakers weigh the engagement and access benefits of gamified investing against risks such as overtrading, and argue that advisors can connect market information to clients’ real goals.

AI And Confirmation Bias

Anastasia says AI can flatter users and reinforce confirmation bias, but investors can use adversarial prompts and counterfactuals; she also cautions that polished answers may create overconfidence.

AI And Advisor Value

Devin expects AI to improve efficiency in technical planning tasks but sees behavioral coaching as a distinct human role, especially when clients need help making decisions during emotional moments.

Crisis Memory And Risk

Anastasia says lived experience of downturns can shape investors’ risk choices for years, and suggests that as investors with crash memories leave markets, another major downturn may become more likely.

Episode Closing And Resources

The host thanks the guests, points listeners to referenced research and PIMCO advisor resources, and closes the conversation before the standard legal disclosures.