Investing Behavior Looks More Like Temperament Than Strategy

Sam Parr introduces the Swedish twin study as a way to think about investing behavior, not as an excuse for it. He learned about the topic through Jim O’Shaughnessy and describes an unusually useful research setup: Sweden had a large twin registry, originally useful for health research, and until 2007 also required adults to report financial assets under a wealth tax. That meant the government had records of portfolios, savings, stocks, mutual funds, and other financial holdings.

The comparison was between identical twins, who share effectively all of their genes, and fraternal twins, who share roughly half. Because both kinds of twin pairs often grew up in the same household, the study could compare people with similar upbringing but different degrees of genetic overlap. Parr says the researcher looked at six recurring investing biases: holding too few stocks, trading too often, buying whatever did well the previous year, over-investing in one’s home country, preferring lottery-like stocks, and refusing to sell losers.

The on-screen material included a summary-statistics table from the study, with 17,630 twins in the dataset and breakdowns for identical and fraternal twins by gender, socioeconomic characteristics, and savings behavior. Parr describes the broader study as examining roughly 30,000 sets of twins and also looking at outlier cases such as twins separated at birth.

The study’s conclusion, as Parr presents it, was that about 45% of savings and investing patterns were genetic. He treats that number as startling, but not as fatalistic. His interpretation is that investing behavior is a window into broader human behavior. What looks like a portfolio problem may be a personality problem expressed through money.

That is why great investors are compelling to Parr even though he and Shaan Puri do not consider themselves professional investors. They have spent years interviewing investors and reading people like Warren Buffett, but Parr argues that finance itself is secondary. Financial products, cycles, and fashionable vehicles change. Human nature does not change nearly as quickly. To be a world-class investor, in his view, one has to understand why humans behave the way they do.

That framing turns the investment discussion inward. Parr says the company a founder controls is often an extension of the founder’s personality. If a founder has trust issues, employees may experience that as micromanagement. If a founder has trouble committing, the company may experience that as absent-mindedness. If a founder is kind, the culture may become kind. He sees the same pattern in investing: a refusal to sell losers may not be only a stock-market bias; it may resemble attachment to failing relationships or projects. Home-country bias may reflect a broader preference for the familiar.

The conclusion is not that reading is useless, but that information alone rarely remakes behavior. Parr says the Swedish research found that after normalizing for education, simply reading more business books did not necessarily make someone a better investor. The meaningful non-genetic difference he highlights was work experience in finance — exposure to loss, feedback, and pain.

Parr connects that to Daniel Kahneman’s admission in Thinking, Fast and Slow: even writing a book on biases did not immunize him from them. Awareness is not cure. The practical response is to design around predictable failures.

His proposed defenses are concrete. First, operate in what he calls a “zone of genius,” citing Buffett’s line that “we invest in our zone of genius.” Parr interprets Buffett as a slow-and-steady person who invests in slow-and-steady things, which helps explain Buffett’s aversion to Bitcoin and why, in Parr’s view, Buffett would not naturally be the sort of 18- or 25-year-old today building an AI startup funded by aggressive venture capital.

Second, pre-commit so that future mood does not become future policy: write down hiring or firing thresholds in advance, then follow them. Third, shorten feedback loops so mistakes become visible quickly. Fourth, avoid games where one’s bias can be fatal. Parr says that as a control-oriented, relatively slow-and-steady person, a heavily venture-backed company requiring extreme speed and loss of control could be a structurally bad game for him.

Puri accepts two pieces of the argument: that problems in a company or life can mirror problems in psychology, and that the remedy is not merely better strategy. He connects it to Morgan Housel’s idea that wealth through investing is less about strategy than behavior. Parr puts it more directly, borrowing a line he attributes to Housel: “personal finance is more personal than it is finance.”

The same logic applies to careers and businesses. Shaan Puri brings up Mohnish Pabrai, who had appeared twice on the show and, in Puri’s telling, became a Hall of Fame guest. Before becoming known as an investor, Pabrai ran a moderately successful company — Puri recalls something like $6 million in annual revenue — but did not love it. He assumed the feeling was ordinary entrepreneurial burnout.

A deep personality assessment changed the frame. According to Puri, the assessment involved conversations with Pabrai, coworkers, parents, and others around him. The conclusion was that he was miserable because the game he was playing did not fit his personality. The assessors described him as someone suited to “solo player, competitive, number-based games.” Running a company was multiplayer, operational, and not primarily a numbers game. Investing was the opposite: a solo competitive numbers game.

Pabrai later found the same pattern in other domains. He had an edge in casinos, though Puri says Pabrai described it off-camera rather than on-camera. Pabrai was not card counting, Puri says, but had developed a system for specific casinos and specific ways of playing. The money was not the point; the thrill was beating the house.

Philanthropy became another version of the same game. Rather than participate in gala culture or social philanthropy, Pabrai looked for the highest impact per dollar. Puri describes his work identifying smart children in rural India, funding competitive math education for roughly $3,000 a year, and helping them reach IIT, which Puri describes as harder to enter than Harvard or Stanford. In Puri’s telling, a relatively small intervention could substantially increase a family’s earning power.

The point is not Pabrai’s exact domain. It is the danger of spending life in the wrong arena. Puri says another mentor, James Currier, made the same point through sports and business: Currier had played competitive soccer for years but later realized his body and coordination were better suited to racquet sports. Puri generalizes the insight: there may be billions of people who never discover the game they are unusually good at, never enter it, and let life pass by.

Sam Parr’s version of the right game is research. He says he is most alive when he can take a new topic, go deep, obsess over it, talk to experts, study what others did, find where they failed, and synthesize conclusions. Puri notices the same pattern in both of them: a default instinct to reverse engineer. A book developer once told Puri that the most interesting thing about him was not a specific business idea, but his tendency to reverse engineer almost everything — business, family, creativity, investments — by studying history, interviewing people who had done it, extracting principles, ignoring noise, and building his own system.

Parr says that instinct shows up when founders ask for advice. Instead of giving a generic prescription — raise venture capital, hire this role, try this tactic — the first question should be: what would success look like in five years? Define the game first. Then work backwards.

That is the thread connecting genetics, investing, and founder psychology. Parr does not argue that the right personality assessment or study removes difficulty. He argues that the wrong game magnifies one’s flaws, while the right game turns temperament into leverage.

Shaan Puri is skeptical of how much weight to put on a study that implies genetic constraints. His filter is not “is this true?” but “does this serve me?” He calls his preferred beliefs “productive placebos”: the beliefs one needs in order to do the things that produce a good life.

He gives Howard Marks’s S&P 500 example as a warning about statistical interpretation. Marks, Puri says, asks people what the S&P’s average annual return is; they answer roughly 10%. Marks then points out that a 10% year almost never happens. Annual outcomes vary widely, and an average can mislead if it becomes an expectation. Puri uses the familiar image of drowning in a river that is “on average” five feet deep. Averages can be true and still not guide action well.

So when Sam Parr says genetics explain a large portion of financial behavior, Puri’s instinct is to discard any belief that makes him feel less in control. He notes the replication crisis and the possibility that “facts” from a study may not hold up. But even if the claim is true, his response is: if 45% is genetic, then 55% is not. “So you’re saying there’s a chance,” he jokes, invoking Dumb and Dumber. What he wants to retain is the useful version: identify behaviors that make money, identify behaviors that lose money, do the former, avoid the latter, and be especially alert where one’s temperament pulls in the wrong direction.

Parr’s reaction is different. If the facts show a predisposition, he wants to use that knowledge to design systems that still get him what he wants. The disagreement is not about whether behavior matters. Both agree it does. The difference is epistemic: Parr wants to work from the fact pattern; Puri wants to work from the belief that produces agency.

That disagreement becomes more concrete when they talk about their own investing biases. Puri says he is guilty of all the listed biases, especially overactivity. Parr says he does not have much performance chasing and tends to stick to what he knows, but he does refuse to sell losers and becomes emotional about things he likes. The self-diagnosis fits their prior descriptions: Puri is generative, active, idea-rich; Parr is slower, more attached, more inclined toward familiar domains.

Shaan Puri asks Sam Parr what happened after Parr sold The Hustle in February 2021 and received a large amount of capital. Parr says roughly 80% went into the S&P 500. That portion of the portfolio, he says, is up whatever the S&P is up; he describes it as roughly doubled or up around 75%, depending on how framed.

Puri presses him on Howard Marks. Marks had come on the show, Puri says, and told them the S&P 500 was a bad bet at that moment because valuation was high — Puri recalls something like a 23 times price-to-earnings ratio. Marks’s argument, as Puri summarizes it, was that when investors buy at that valuation, history suggests the next 10 years of returns may fluctuate around minus 2% to plus 2%.

Parr says he did the math: since Marks made that comment in August, the market was up 12.5%. Puri notes that Marks was not making a five-month forecast; his point was a 10-year valuation-based warning. Parr’s reply is that he does not care, because his personal plan is longer than a decade. He says he studies America and history, and expects history to continue repeating well enough during his lifetime. More specifically, he argues that the S&P 500 is no longer simply an American index but effectively a global index. He wants an 8% nominal annual return over a very long period. If he gets that, his plan works.

Parr says he set a target at age 21: if he had a certain amount of money by 30, and that compounded at 8% annually, he could spend what he wanted, still have money left, and eventually give or leave money away. Marks’s warning does not break that plan unless the long-run compounding assumption fails. “What I look at is till I’m 100,” Parr says. His disagreement with Marks is therefore not a claim that Marks’s 10-year valuation analysis is wrong. It is that Parr’s personal constraint is different. A bad 10 years is tolerable if the 40-year plan remains intact.

Puri sees the distinction: Marks is using historical data about entry valuations; Parr is using a longer time horizon. Puri jokes that if Parr wrote a book on investing, the first page would simply say “America.” Parr complicates that, arguing that the S&P 500 is not cleanly American in the way people imply. Apple may be headquartered in the United States, but its revenues, supply chains, and markets are global. Puri asks Claude and reports that roughly 43% of Apple’s revenue comes from American sales and roughly 54% of operating income. Parr’s point is that if the figures are around half, the national label is less meaningful to him.

The exchange exposes a deeper difference in wealth accounting too. Puri says his net worth is up roughly 40 times since selling, but about 70% came from companies he started or owned, not public-market investing. He includes private companies in net worth, while separating liquid and illiquid net worth. Parr says he does not count private companies that way. Puri does, but marks them mentally because they can move sharply: his e-commerce business, for example, became worth less in his internal assessment after tariffs, weaker consumer sentiment, lower margins, and plateauing performance.

Parr’s S&P decision is consistent with his personality model. He chose a game he could tolerate: broad exposure, low activity, long horizon, and no need to outguess valuation warnings from a superior investor over short periods.

Shaan Puri connects investing overactivity to company overactivity. He suggests that one reason great investors read so much may not be the knowledge itself. Reading, bridge, pickleball, and other activities may simply prevent the most damaging behavior: sitting at a screen and pushing buttons. For investors, “doing nothing” can be the discipline.

Sam Parr applies the same idea to business. If a company is going well, he says, the best thing he can often do is nothing. Meddling creates problems. If a social-media strategy works because posting three times a day drives growth, the answer may be to keep doing that, not constantly add new tactics. Founders may need a separate playground where they can experiment without damaging the main thing.

Puri illustrates the problem with a Jeff Bezos story. Bezos said Jeff Wilke once told him, “You have enough ideas to destroy Amazon.” The issue was not that the ideas were bad. It was that every idea created a backlog, a queue, a distraction, and work-in-process. Bezos had to learn to release ideas at a rate the organization could absorb.

Puri pleads guilty. He says he has drowned growth and marketing teams with ideas: 14 marketing ideas, then two more each day, then a tweet from someone else, then a person to meet, then an event to attend, then a tactic to retry. A generative founder can suffocate an organization with plausible, even good, ideas. That is another case of a personality trait becoming an operational defect.

Parr’s practical fix at Hampton is a containment ritual. His co-founder Joe, who is CEO, set up one Thursday meeting where Parr can bring any idea. They keep a running Notion list. In that room, they can discuss anything. Outside it, Parr is not supposed to Slack ideas into the organization. The system does not suppress his instinct; it queues it.

That is the more mature version of “know yourself.” The goal is not self-expression. The goal is architecture. If a founder knows he is overactive, overattached, avoidant, controlling, or conflict-averse, the company needs mechanisms that prevent those traits from becoming default operating procedure.

Shaan Puri turns to Y Combinator’s Requests for Startups, which he has followed for about a decade. In earlier years, he says, many requests sounded normal: better software for lawyers, a tool to help a known workflow, incremental but useful businesses. Sam Parr notes that some past requests were already ambitious, including one about ending cancer. Puri’s larger observation is that the list has become increasingly brain-breaking, and that this may signal where the ground is moving.

The first idea is not from the YC list directly but from Daniel Gross: aesthetic data centers. Puri argues that the world needs many more data centers and far more power to “win the AI race,” but local opposition is growing. People may use AI while disliking AI companies, AI art, AI music, data-center water use, and the effect on power bills. Puri says the data would suggest golf courses use far more water than data centers, but golf courses do not draw the same protests. His point is that data centers face a not-in-my-backyard problem.

Gross’s idea, as Puri relays it, is that because data centers already cost billions of dollars, the incremental cost to make them architecturally beautiful, locally welcome, or publicly beneficial may be small. Puri frames it as a kind of corporate public art: infrastructure that earns permission through beauty or civic value.

Parr’s historical analogy is Rockefeller Center. John D. Rockefeller, he says, had a reputation problem after Standard Oil, monopoly backlash, and mining-town conditions tied to company operations. Parr describes John D. Rockefeller Jr. visiting a poor mining town after a disaster and later, during the Great Depression, backing a major architectural project in Midtown Manhattan that would employ thousands and help revitalize the city. Parr calls Rockefeller Center a case of “reputation laundering”: taking reputational harm and countering it with a civic monument that still shapes public memory.

Puri adds cell towers disguised as palm or pine trees — monopalm and monopine towers — as a more literal precedent. People disliked ugly steel towers, so companies made them look like trees. He also mentions Andrew Carnegie building 2,500 libraries while Carnegie Steel faced brutal labor practices and major strikes. Parr adds that Carnegie was hypocritical, citing his focus on costs and his treatment of workers. The point is not moral absolution. It is that infrastructure and corporate power have often needed aesthetic, civic, or reputational packaging.

Parr is receptive. He points to pre-war buildings in New York and venues like the Paramount Theatre in Oakland as evidence that utilitarian or commercial buildings used to be far more beautiful. Puri names David Perell and The Cultural Tutor as part of a broader online movement asking why the modern world looks bland, noting that the theme has attracted millions of views. Aesthetic data centers sit at the intersection of AI infrastructure, local politics, and a renewed appetite for beauty.

Puri uses drone-swarm defense as another sign that the opportunity set has shifted. He cites a YC request around low-cost drone defense and relays a story he had heard: Iranian drones damaged an AWS data center that lacked drone protection. The economic asymmetry is the problem in his telling: the U.S. military may use million- or two-million-dollar missiles to destroy $200 drones. An adversary will take that trade all day.

For Puri, this means war is changing faster than institutions built for the last war. He is careful to be respectful of decorated generals, but questions whether people trained in a prior technological paradigm are best positioned for the next one. Future conflict may look more like cyberattacks, drones, and video-game-like systems than battlefield images from the past. Parr adds that Alex Karp of Palantir has argued Silicon Valley should be more involved with the U.S. military and government. To Parr, it is a strange and meaningful culture shift: the nerds are now interested in defense and state power.

Puri frames this as part of a recurring pattern in startups. The next important wave rarely looks like the last. When he moved to Silicon Valley, many people were still pattern-matching from Facebook, Twitter, and Dropbox. Then the big companies were real-world marketplaces like Airbnb, Uber, and Lyft, which required city-by-city playbooks, driver acquisition, local incentives, and operational hustle. Then crypto created wealth through an entirely different set of ideas: money, fiat currency, cryptography, double-spending, and buying tokens rather than equity. Then AI shifted the ground again. OpenAI and Anthropic looked less like traditional startups and more like research labs, in some cases nonprofit or open-source-adjacent in ways that would have sounded strange to earlier Silicon Valley founders.

The warning is that expertise can expire. The people studying the last winners may be optimizing for a game that has already moved. Puri says the next decade’s interesting companies may be in hardware, robotics, defense, AI management systems, and other areas that did not fit the previous startup template. Palmer Luckey, in his telling, helped make defense cool in a culture where, years earlier, someone wearing a suit at a startup meetup might be jokingly treated like a narc.

The AI idea Shaan Puri dwells on is the “company brain.” He says most people still think of AI as a smart assistant. You ask it a question; it retrieves an answer. You assign it a task; it attempts the task. In a slightly more advanced model, AI becomes a set of digital employees — tireless workers who perform functions under human direction.

Puri thinks the direction may reverse. The AI may not work for the company; the company may work for the AI.

He knows people will dislike that sentence, so he defines it operationally. AI is good at absorbing broad information, having “eyes and ears everywhere,” reading everything, and making decisions. Today the CEO or manager sits at the center while AI tools feed information into that person. Jack Dorsey, as Puri and Sam Parr describe his argument, suggests a different structure: AI is the brain at the center, and humans are nodes around it. Humans provide context, gather real-world information, execute tasks, and feed results back. The AI integrates the information and decides.

Parr says Dorsey described the old model as a human in the circle, with agents and machines as surrounding nodes. The new model puts AI in the center. Humans become the nodes. The AI is the thinker; humans add information and often execute. Puri qualifies that humans will likely retain editing capabilities. Neither sees it as absolute replacement. But both recognize the inversion.

Puri says his own use of AI has shifted in that direction. He began with “tell me this” and “go do this.” Increasingly, he asks AI to question him, gather his answers, and help him make decisions. The AI becomes less like a junior assistant and more like a decision partner.

He then connects this to a Citrini Research scenario shown on screen as “The 2028 Global Intelligence Crisis.” The visible document describes itself as “a scenario, not a prediction,” and frames the memo as a thought exercise about what could happen if AI bullishness is right but the result is economically bearish. As Puri summarizes it, the scenario imagines AI-driven productivity gains that reduce employment. Fewer workers receive wages; lower wages reduce spending; lower spending reduces revenue; lower revenue leads companies to cut more staff. Productivity rises, but the economy can still spiral downward because gains accrue to owners of compute rather than circulating through households. Parr asks AI to summarize the thesis in three sentences and reads a version of that: if AI works exactly as bulls predict, white-collar jobs are gutted, displaced workers earn less, consumer spending collapses, and gains flow to compute owners.

Puri says there was a large market sell-off around that discussion, but he leaves the causal link uncertain: he says he does not know whether the Citrini piece “triggered or created or maybe just coincidentally” coincided with the sell-off. He also says people got angry at Citrini, either because they accepted the grim conclusion or because they thought the firm was spreading doom and gloom. The point for him is the shape of the argument: a world can become more productive and still become economically unstable if income stops circulating through workers.

Puri says Citrini later drew attention again by sending a researcher to the Strait of Hormuz to gather first-party information. In Puri’s telling, the analyst crossed into Oman, paid for a boat, and observed that the Strait was neither simply open nor closed; it functioned more like a toll route, where the right country and the right payment could get through. Citrini’s point, as Puri presents it, was that people first got angry about the claim that white-collar jobs are at risk, then got angry when Citrini showed the kind of analyst work that might survive: fieldwork that collects real-world data to feed an AI engine.

That is the job redefinition Puri finds important. The analyst of the future may not be someone reading spreadsheets and earnings reports. AI can do that at scale, simultaneously. The analyst may be someone who obtains better proprietary context from the physical world, then feeds it into the model.

Parr already uses a tool in this category, Viktor, which connects to company tools and lets people ask questions in Slack. Viktor’s site, shown on screen, describes it as “Not a tool. A hire,” an AI project manager that reports, builds dashboards, writes code, creates campaigns, queries tools like Stripe, Meta Ads, Notion, and GitHub, and learns from company conversations. Another Viktor page frames the distinction this way: “ChatGPT tells you how to audit your ad spend. Viktor audits it. Hands you the PDF.”

Puri does not use such tools because he does not trust startups with sensitive company information. Startups, he says, are often leaky buckets; employees may see far more than customers assume. Parr agrees with the unease, describing the odd feeling of buying something online and receiving a friendly founder email — fine for shoes, uncomfortable if the purchase were more private.

The same access that would make an AI company brain useful would also make it sensitive. Puri’s caution is simple: a system that can manage the company needs to see the company.

The final example is personalized medicine through AI agents. Sam Parr reads from a prompt-like description: intelligent agents can analyze personal health data — diagnostic tests, scans, electronic health records, wearables — and produce accurate, user-specific recommendations. Shaan Puri tells a story he attributes to Nat Friedman. Friedman gave Claude Code his genetic data and blood-test data, and it inferred that he was likely chronically dehydrated. Friedman then told it to do whatever it needed to make him not dehydrated.

The story becomes absurd and revealing. Friedman had connected the system to screens, cameras, voice assistants, WhatsApp, and his home environment. The AI could see that he had been sitting for hours without water, tell him to get up, watch him drink, and congratulate him. Parr asks why a Post-it note on the refrigerator would not suffice. Puri agrees the jokes write themselves — tech people taking things too far — but sees it as a preview of AI-managed health.

Puri adds another Friedman story: when AI told him he needed a supplement and he did not have it, the system, connected to his Tesla, rerouted the self-driving car to a nearby Whole Foods. Friedman noticed the directions change and bought it. Puri presents this not as ordinary consumer behavior but as a glimpse of agency: software moving from recommendation to intervention.

Parr brings up Sid from GitLab, a billionaire founder who said he was going “founder mode” on his cancer. Puri says Sid cured his own cancer with AI; Parr is more cautious and asks whether it is cured before moving to a separate example. Parr mentions a friend with cancer who connected with Sid and others, works in AI, and has been using AI in the process. Parr does not disclose specifics, but says the friend has been getting very good news every couple of months and that the cancer is going away, though not gone.

Puri’s reaction is that this should be a much bigger public story. If AI is already helping people organize cancer treatment or uncover health interventions, he thinks the attention level is strangely low. Parr says one viral example — a person who treated a dog’s cancer — made it to Australia’s version of the Today Show, not to the level of recognition the breakthrough might imply.

Puri quotes the familiar line: “the future is already here, it’s just unevenly distributed.” He says that feels more obvious now than ever. Some people are already wiring AI into their bodies, homes, cars, companies, and medical decisions. Most people are not. The future is present, but unevenly installed.