Academic research has catalogued more than a hundred behavioral biases, and Anastasia Buyalskaya's first piece of advice to financial advisors is not to memorize any of them.
The standard response to bad investment decisions is to train the bias out of the people making them. Buyalskaya says the research shows that does not work, so the fix has to be structural: change the size of the committee, the order people speak in, and what they commit to before the drawdown arrives.
"The solution is just to put some structure into the investment committee meeting such that the most senior person is the last one to speak"
Buyalskaya is an assistant professor at HEC Paris who is embedded in PIMCO's own investment process — she presents at its quarterly investment forums and has co-authored the firm's macro signposts column. Devin Ekeberg has spent his career teaching this material to advisors in the field.
The full interview is covered here so you can skip it. 51 minutes of audio, 18 minutes of reading.
Here are the 12 principles that matter.
👤 Guests: Anastasia Buyalskaya, Assistant Professor at HEC Paris and a behavioral finance adviser to PIMCO, who presents at the firm's quarterly investment forums; Devin Ekeberg, Senior Vice President on PIMCO's advisor education team
🎙️ Host: Greg Hall, who leads PIMCO's wealth management business in the United States
📰 Published: 14 September 2026 on YouTube (PIMCO U.S.)
🔴 YouTube | ⏱️ 51 min | ✅ Time saved: 33 min
Key Takeaways
Trying to debias people does not work, so redesign the meeting instead
Senior person speaks last; junior and contrarian views get guaranteed air time
Committees work at three, five or seven, and start losing accountability above that
A committee of 13 leaves nobody identifiably accountable when the call is wrong
Emotions are not noise in a decision — they are information, unless they are displaced
Anger about the debt crisis may belong in the allocation; anger about the drive to work does not
The most effective way to stop a client panicking is to make their former self the one giving the order
Large language models are built for engagement, which makes them flattering rather than accurate
Buyalskaya's counter is to prompt for a devil's advocate case against your own plan
Vanguard put an advisor's value at around 300 basis points a year, and Ekeberg says AI can already do six of the seven things on that list
The exception is behavioral coaching, which is why he is optimistic rather than worried
Behavioral awareness has not made another 2008 unlikely, because the people who lived through one keep retiring
1. Debunking Econ 101
Buyalskaya opened with where the field came from, and Greg Hall and Devin Ekeberg both pushed on what it replaced.
The field dates to the 1970s and one paper. Buyalskaya traced behavioral finance to the publication of prospect theory by Daniel Kahneman and Amos Tversky, which she said "really opened the floodgates" to what is now called behavioral science
The finding was that people do not behave the way the models say. Expected utility theory makes assumptions about how people treat losses and gains; taken into the wild and asked to make simple decisions, people turned out to be highly loss averse, and their risk preferences changed with the context they were in
Hall put it as the debunking of the perfectly rational actor he and Ekeberg were taught in Econ 101 — total information, optimal decisions, and lines that were "straight or like beautiful curves." Buyalskaya agreed it totally debunked the myth and said students are still taught it
Ekeberg's correction was that the problem is worse than a stale lecture, because the theory is still the machinery. Modern portfolio theory dates to Harry Markowitz in the 1950s, and mean variance optimization is what business schools teach and what the industry runs on. "Every asset management firm in the world uses some element of modern portfolio theory to construct their portfolios and so forth."
His framing of the tension is the one the whole conversation runs on. There is a balancing act between the mathematical exercise of investment management and the human side of it, and "And if you're ignoring one or the other, you can end up with some pretty terrible outcomes."
Hall's aside was that the average advisor listening is chuckling, because they could have told academics this thirty or forty years ago
2. The Crisis That Primed Her
Hall asked what in Buyalskaya's own experience made Kahneman's book land the way it did.
She was on a trading floor for the global financial crisis. She started her career just before it, was not managing P&L, and described herself as "a fly on the wall" just starting out
What broke was her assumption that expertise removes uncertainty. The illusion that an expert with a clear financial model faces very little uncertainty was, in her words, totally popped — and she came away interested in the fact that even experts with excellent analytics and robust engines are still deciding under uncertainty and still subject to bias
Thinking, Fast and Slow arrived two years after the crisis and she read nothing else until she finished it. "I think I just didn't do anything else except read that book until I had read it because finally somebody was explaining a much more accurate model of how people were really making decisions that accounted for human flaws and biases." Kahneman won the Nobel Prize in 2002; she put the book at 2011
3. A Core Subset of Biases
There is research on more than a hundred biases and no reason to learn them all. Buyalskaya said the field has produced great academic research on over a hundred, but that there is a core subset worth knowing rather than a list to memorize
Overconfidence is the one PIMCO's own CIO returns to. She named Dan Ivascyn as talking about it a lot, and put the problem structurally. "So anybody that's an active manager by nature is out there trying to beat the market, right? It's almost the definition of overconfidence." Her answer is an attribution system that separates skill from luck, so overconfidence is reined in where there is no demonstrated skill
Confirmation bias has two forms. Hall offered a distillation — confirmation bias is when you "see in the data what you want to see" — and she accepted it, then split it: going out to cherry-pick data that confirms a prior, or reading the same data differently because of the perspective you already hold
Her test for it is what lands in your inbox. "So, if you talk to an investor, they're much more likely to send you the sellside research that supports their view, right?" The same selection applies to who gets hired onto teams and seated on committees, which is where confirmation bias turns into groupthink
Ekeberg's count was higher and his framing was practical. He put the number of identified, repeatable biases at around 150, and argued that identifying exactly which bias is in play is hard while identifying which part of the brain it came from is easier — an emotional response calls for a different prescription than a cognitive one
4. Emotions Are Information
Ekeberg used Kahneman's system one and system two split, with system one as the automatic, nervous-system part that produces the emotional biases that are hardest to overcome, and the cognitive part as where information is received and processed — with, he said, plenty of opportunity to mess ourselves up there too. He then asked Buyalskaya whether the two-systems model still holds academically
Her answer reframed emotion entirely. Citing her former Caltech professor Ralph Adolphs and his book with David Anderson, she described emotions as functional states. "So rather than something kind of that's distracting you, emotions are information." They serve a function by directing attention
The distinction that matters is the source of the emotion, not its presence. Anger about the debt crisis might be genuinely informative about a portfolio allocation. "But anger because somebody cut you off on the highway on your way to work probably shouldn't inform the portfolio allocation." She said emotion gets its bad reputation from displaced emotion rather than from emotion as such
5. Structure Beats Debiasing
A committee is a different problem from an individual because of authority bias and groupthink. Buyalskaya noted the literature on wisdom of crowds and collective intelligence assumes the inputs are uncorrelated — that everyone in the room drew on different sources — and said that assumption often fails in an investment committee
The fix is not to insist the room is not intimidating. She said she sometimes hears investment teams claim authority bias is not a problem because the CIO is not intimidating, and rejected that as the answer. "The solution is just to put some structure into the investment committee meeting such that the most senior person is the last one to speak" — with air time protected for junior and contrarian voices
Survey the team before the meeting so people are anchored on their own view. Simple quantitative surveys taken in advance stop the conversation itself from setting everyone's position
PIMCO's own example is Tiffany Wilding's regional committee, which surveys the whole team on a set of macroeconomic variables and looks at the distribution before the discussion starts. Buyalskaya's argument for it: if the median view is also the most extroverted or most senior view, it will dominate the air time anyway, so the useful move is to go looking for the tails and structure the debate around them
This is the load-bearing claim of the interview. Her position is that guardrails and structure work, and that the research says trying to debias people does not
6. Three, Five or Seven
Committees get too big very quickly, and she has specific numbers. "And so the magic numbers that I always talk about are three, five, and seven."
Odd numbers because of ties. "Odd numbers work quite well because you have a tiebreaker." That matters most where the committee uses surveys or explicit voting
Above seven, nobody owns the call. In her words, "once you get bigger than seven it's very easy to have a dilution of accountability right so if you have a committee of 13 people all based in the US all fixed income experts making a call on duration then who is accountable if that's the wrong call"
The air-time arithmetic makes the same point. Thirteen people in a 45-minute meeting leaves two to three minutes each, which she said is not the best use of those people's time — and some of them are close enough in view that cutting the number costs very little information
The trade-off she wants people to watch is speed against quality. "Well, I think just recognizing that there's always a trade-off between decision speed and decision quality." More people slows the decision down in exchange for quality, and markets and clients do not always allow that. Her diagnostic: if three people's input already gets you to the answer, the room is too big; if the decision is being forced while questions are still open, it may be too small
7. Deciding With a Cool Head
Hall asked Ekeberg how this shows up in the advisor's world, and the answer converged with Buyalskaya's on pre-commitment.
Advisors are moving toward systems as a form of default decision-making. Ekeberg's comparison was his brother in law enforcement, trained to manage emotion in favor of a standing operating procedure, on the theory that decisions made in advance are better decisions
In practice that makes the advisor part therapist. Clients arrive reading headlines about wars, inflation, tariffs and "a thousand other things" being fed to them, and Ekeberg described advisors working to change what those things mean to the client — replacing fear with something closer to interest in which asset classes now look more attractive, and steering toward small changes rather than drastic ones
Buyalskaya endorsed the premortem as best practice. "So before they're experiencing a big draw down, before they're in a position where those emotions are likely running high, sort of doing that premortem on their own portfolio" — asking what they would do in a 20% drawdown while their head is still cool
She noted that checklists and premortems are standard in medicine, law enforcement and the military, precisely because those professions know nobody has full cognitive capacity in an emergency
Ekeberg added the depletion argument. "It's very difficult to maintain that cognitive energy all day long." "And you've probably seen some of the studies on even judges in the courtroom will make certain decisions in the morning and very different decisions in the afternoon because their case load is so high and they're just completely drained." His conclusion: set the process in advance, red-team your assumptions when you are at your best, and then hold a very high bar for changing the decision later
Hall's own experience is the mechanism working on him. Three weeks after signing off on a policy statement he changes his mind on a headline, and the advisor's reminder that he laid out this plan himself is what holds
Buyalskaya uses the same line on investors. "No, your former self is telling you to make this decision because you committed to this, 3 weeks ago." She argued it is far more effective than intervening without the person having a stake in the decision
8. Who Feeds You the News
Hall picked up Ekeberg's word "fed" and turned it into a section.
Ekeberg's point is that information systems are curated more than people appreciate, sometimes deliberately, to influence the decision at the end of them. His recommendation to advisors is to be conscious of the source of what they consume and why it reached them
His example is his 16-year-old daughter and her apps. "there's a billion dollars worth of engineers on the other side of that app trying to keep you scrolling on that app and showing you information in very specific ways to influence your behavior" — a point usually made politically that he says applies just as squarely in finance
Buyalskaya described three eras of consuming information. A newsstand where you made an active choice between two papers and knew which prior you were confirming; digitally intermediated news, where the echo chamber shows you what resembles what you have already seen; and now AI, which does not tell you the source at all
Footnotes exist, she said, but nobody goes to them. "And so I think we're getting more and more removed from those primary sources." Whether that fuels overconfidence and confirmation bias or improves decisions she called an open question
Hall extended it from social platforms to the advisor's own firm. Consolidation in the RIA space means most advisors sit on a larger platform with its own imperatives, and the balancing act is between that external voice and the advisor's own independent thinking. Both he and Ekeberg volunteered that they have teenage daughters and that this is a running conversation at home
9. Gamified Dopamine
Gamification works because it hijacks a reward system, not because it is clever. Buyalskaya's research covers digital broker dealer apps that gamify financial decisions, and she said the rewards can be completely vapid and still release dopamine. "So, I think they've hijacked that kind of dopamine system really really well and we've seen it used for good." Her example of the good case was fitness apps like Fitbit
Hall's confession made the point better than the theory. "I check my I check my ring every morning to know how I'm feeling. I used to just wake up and know how I was feeling, but now I have to check an app to find out."
She gives gamification credit for bringing retail money into markets, and calls that a form of democratization — while flagging overtrading and excessive risk-taking that gets regretted later as the thing worth researching
Hall pushed back on the word democratization itself. His caution to advisors: when something is presented as a gift brought down to the huddled masses, the party bringing it usually has a lot to gain from getting you or your clients involved
Buyalskaya's answer was that this is exactly the advisor's job. The apps are good at engaging the automatic, reflexive part of the brain; an advisor engages the reflective part, running the checklists and premortems so the client makes decisions they are unlikely to regret
10. AI That Flatters You
Buyalskaya said behavioral scientists sit on both sides of whether AI helps or hurts decisions. The models are built for engagement, which shapes what they say back
The flattery is the design, not a quirk. Her illustration was the model telling Hall his question is exactly the question he should be asking about his financial plan rather than telling him it was a bad one. "So they do flatter you and they do have a tendency to find data and evidence to support what you want to hear." She tied it directly back to confirmation bias
The counter is in the prompt. Ask the model to argue the other side — "I need you to play devil's advocate and tell me why this financial plan is likely to fail" — and she said investors are using the tools this way to question their own decisions and run counterfactuals that would be prohibitively slow for human analysts. She noted the tools are sensitive to the exact wording, so it takes experimentation
She flagged a specific paper on the new biases. Lisa Messeri and Molly Crockett's work on AI and illusions of knowledge, and in particular the illusion of explanatory depth: a well-written one-page answer leaves you feeling you understand the subject. "And their point is you actually probably understand the tip of the iceberg and you might be a little bit overconfident because there's a whole bunch of stuff that you don't understand."
Her test for anyone who doubts it is to use the tool in their own field. "So I'll use AI a lot and ask it questions about behavioral finance and it'll sort of make up papers, right?" She is placed to catch that in her own domain and would not be in someone else's
Hall's analogy was a former colleague who won every argument and was frequently wrong — articulate enough to turn your mind around without being right — which he said makes the failure mode of these models look very human
11. The One Thing AI Won't Do
Ekeberg has seen this cycle before. Advisors have always chased efficiency, and the robo-advisor years produced the same fear that the technology would take over
He anchored the value question on a number. "And they landed they quantified it literally. They landed at around 300 basis points a year." That is Vanguard's Advisor's Alpha study, which broke the value of an advisor into six or seven categories — portfolio allocation, withdrawal strategies, tax strategies among them — plus behavioral coaching
His observation is that AI already does most of that list. Six of the seven categories advisors took credit for are things he says AI is doing now and will do better, leaving behavioral coaching as the one he does not expect it to take
The illustration is a walk, not a report. An AI model will produce a long response on what the client should do; the advisor recognizes an emotional response first. "Maybe we should actually just take a walk around the block first, right?"
Hall's recollection of the robo era supports it. Consultants declared it game over; what actually showed up later was that as clients crossed wealth thresholds and the stakes became real, they turned to a human being to bridge the logical and the emotional
Ekeberg's read on all of this is optimistic rather than defensive. His argument is that the advisor's value is taking abstract information the client can do nothing with and connecting it to what they actually care about — liquidity for a purchase, college in a few years, a retirement lifestyle, a legacy — which he called a grounding mechanism
12. Memory Is Short
Hall's closing question was whether the spread of behavioral awareness helps explain why there has been no repeat of 2008.
Buyalskaya would not take the compliment. "I think that's a very generous hypothesis." She would love it to be true, but said memory is short and the composition of market participants changes every decade
The research she pointed to is on experience effects. Ulrike Malmendier's work shows that living through an event, rather than reading about it, is what gives it lasting weight in financial decisions and in how someone thinks about risk and allocation
That implies the risk rises as the survivors retire. "So, it's possible that as those of us that remember some of these big downturns kind of leave the market, right, and retire and more of those actors are leaving the market that we do get another crash." She called that more likely than not
What she will grant is awareness. Far more market participants know about behavioral finance and about their own and others' biases than twenty years ago
Hall's list of what today's extrapolators never lived through ran from the Asian crisis and the Russia crisis to Long-Term Capital, the dot-com boom and bust, the telecom bust, Enron and the financial crisis — then he cut himself off, noting that fixed income people tend toward pessimism and the conversation was too good to end on a dull note
Buyalskaya's bottom line is that individual willpower is the wrong lever: the reliable way to improve an investment decision is to change the structure around it before the decision has to be made.
Bonus Insights
Hall said he had used AI to prepare for this very conversation, and called it a fantastic way to get ready — while Buyalskaya was in the middle of explaining how the same tools mislead
Hall's theory of why models converge on a view is that they return the answer with the most writing behind it, so volume of published opinion becomes the answer
The advisor business has consolidated into teams, which Hall noted is one of the biggest changes of the last two decades — it creates the chance to make better decisions and imports every committee dynamic that goes with it
Buyalskaya was introduced as having co-authored or ghost-authored PIMCO's macro signposts column, usually written by Tiffany Wilding
Hall's reason Ekeberg lives in Colorado is that it is the only place you can get "everywhere in the country in a three-hour flight." Ekeberg's answer: he can get anywhere with a direct flight
Buyalskaya agreed to come back next year for an update on what has moved in the field
Products, Companies & Tools Mentioned
PIMCO (The host's firm, which Hall says has spent the last few years embedding behavioral finance principles in its own investment decision-making; Buyalskaya presents to its quarterly investment forums)
Vanguard (Author of the Advisor's Alpha study Ekeberg uses to quantify what an advisor adds)
Fitbit (Buyalskaya's example of gamification used for good — health and well-being apps that run on the same reward loop as trading apps)
HEC Paris (Where Buyalskaya is an assistant professor)
Caltech (Where she studied under Ralph Adolphs, whose work reframed emotions as functional states)
Books & Resources Mentioned
Thinking, Fast and Slow – Daniel Kahneman (She read nothing else until she had finished it; Kahneman won the Nobel Prize in 2002 and she put the book at 2011)
Prospect theory – Daniel Kahneman and Amos Tversky (The 1970s paper she credits with opening the floodgates to behavioral finance)
The Neuroscience of Emotion – Ralph Adolphs and David Anderson (Her former Caltech professor's book, which treats emotions as functional states rather than distractions)
Vanguard's Advisor's Alpha (The study Ekeberg cites for around 300 basis points a year across six or seven categories, one of them behavioral coaching)
Ulrike Malmendier's research on experience effects (Why living through a crash, rather than reading about one, changes how someone allocates risk)
AI and the illusions of understanding in scientific research – Lisa Messeri and Molly Crockett (The paper Buyalskaya plugged on the new biases AI introduces, including the illusion of explanatory depth)
PIMCO's Macro Signposts column (Typically written by Tiffany Wilding; Buyalskaya has co-authored it)
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: