Bouchaud argues that investors routinely abandon trend following after weak periods because performance chasing is deeply ingrained. CFM's historical research found that a simple strategy remained profitable across decades despite substantial year-to-year variation.

Trend following has experienced repeated flat periods
Weak recent returns do not prove structural decay
Performance chasing causes poorly timed exits
Long-run evidence should outweigh short-term disappointment

“since eighteen hundred, if you pay per trade, a very simple trend following strategy, you make money every decade with ups and downs.”
Jean-Philippe Bouchaud · 30:25

“it's ingrained in people's behavior to chase performance.”
Jean-Philippe Bouchaud · 30:50

#trend following#managed futures#behavioral finance#momentum

Myth Buster49:35

Why Investor Opinions Do Not Reliably Cancel Out

Efficient-market reasoning often assumes investors make independent noisy estimates, allowing errors to average away. Bouchaud argues that real investors imitate and influence one another, effectively behaving like a correlated aggregate rather than independent voters.

Independent estimation errors can average toward a useful consensus
Real investors observe and influence one another
Correlated behavior prevents errors from canceling cleanly
Social influence weakens a central intuition behind market efficiency

“The people are influenced by what other people are doing and what other people are saying.”
Jean-Philippe Bouchaud · 51:34

“they're kind of one guy who's doing only one thing”
Jean-Philippe Bouchaud · 51:43

#efficient markets#herding#behavioral finance#price discovery

Hot Take· 1

Hot Take18:53

A Quant Firm Should Leave More Than Investment Returns

CFM rejected a completely secretive model in which knowledge only entered the firm and never left. Bouchaud argues that commercial excellence should coexist with an intellectual legacy, public debate, and contributions to understanding markets.

CFM chose selective openness over total secrecy
Making money is not the firm's only objective
Research can create a durable intellectual legacy
Commercial and academic ambitions can reinforce each other

“We want to excel, but not at any costs.”
Jean-Philippe Bouchaud · 19:17

#quant funds#research#legacy#culture

Explainer· 4

Explainer02:42

Why Market Crashes Resemble Avalanches

Bouchaud compares markets with granular systems in which countless small interactions usually produce nothing dramatic but occasionally trigger a large avalanche. His physics-based view allows shocks to emerge endogenously from interacting traders rather than requiring an external cause.

Complex interactions can create surprising system-level outcomes
Small events may have no effect until the system reaches a critical state
Market crashes can be self-generated
Endogenous randomness differs from ordinary external noise

“how to interacting elements give rise to something surprising.”
Jean-Philippe Bouchaud · 02:49

“there can be self generated shocks, self generated randomness that come from large assemblies of individuals”
Jean-Philippe Bouchaud · 04:21

#econophysics#market crashes#complexity#systemic risk

Explainer21:35

Why AI Accelerates Quant Research Without Replacing Understanding

Bouchaud frames AI as an advanced continuation of data analysis rather than an entirely new activity for quantitative firms. Machine learning expands the ability to process order-book events, alternative datasets, and text, but CFM remains wary of placing opaque black boxes directly into production.

Machine learning extends longstanding quantitative analysis
Text and high-frequency data create new research possibilities
Black boxes may assist research without being production-ready
Understanding why a model works remains a major challenge

“AI is really an advanced form of data analysis”
Jean-Philippe Bouchaud · 21:35

“we're very uncomfortable with the idea of black boxes.”
Jean-Philippe Bouchaud · 23:13

#ai#machine learning#quant finance#interpretability

Explainer24:33

Can Generative AI Create Millions of Synthetic Markets?

Financial machine learning faces far less low-frequency history than language models receive text. Bouchaud suggests generative models might compensate by producing large numbers of fictitious financial-market paths, while acknowledging that lower-frequency predictability remains uncertain.

Low-frequency market history is a small dataset
High-frequency order books provide much richer data
Language structure may not transfer directly to prices
Generative models could create synthetic financial histories

“stock markets have only existed since I don't know, nineteen hundred or eighteen.”
Jean-Philippe Bouchaud · 26:39

“maybe you can use these gen AI models to generate a millionaires of fictitious financial markets.”
Jean-Philippe Bouchaud · 28:28

#generative ai#synthetic data#market simulation#finance

Explainer32:23

Why Volatility Alone Cannot Describe Futures Risk

A diversified futures portfolio contains a changing web of correlations and anti-correlations. Bouchaud says risk management must forecast both individual volatility and how positions interact, especially when leverage spans many markets and maturities.

Volatility is only one dimension of portfolio risk
Correlations determine how positions combine under stress
Large futures portfolios contain subtle dependence structures
Upside and downside are both treated as risk outcomes

“you really have to think about how all these products interact with one another, talk to one another.”
Jean-Philippe Bouchaud · 32:44

#risk#correlation#futures#portfolio construction

Story· 2

Story05:13

How the 1987 Crash Pulled a Physicist into Finance

A banking contact noticed that Bouchaud's work on physical systems with large jumps resembled financial crashes. Bouchaud reconsidered Black-Scholes in a world with non-Gaussian moves, developed software around the idea, and met CFM founder Jean-Pierre Aguilar, leading to Science and Finance.

The 1987 crash exposed the limits of crash-free assumptions
Physics research on large jumps transferred naturally to finance
Bouchaud rebuilt the problem using non-Gaussian statistics
A chance industry connection led to a rapid partnership

“I read black Shoals and I thought it can't be right.”
Jean-Philippe Bouchaud · 06:24

“why don't we create something together”
Jean-Pierre Aguilar, quoted by Jean-Philippe Bouchaud · 07:15

#black-scholes#1987 crash#econophysics#cfm

Story52:08

The Physicists Who Gave Bouchaud Permission to Cross Fields

Benoit Mandelbrot, Pierre-Gilles de Gennes, and Phil Anderson showed Bouchaud that valuable ideas need not remain confined to their field of origin. Their examples helped him overcome doubts about speaking on finance as a physicist and encouraged his cross-disciplinary career.

Mandelbrot challenged Gaussian views of markets
Leading physicists worked beyond their original specialties
Cross-disciplinary transfer can reveal overlooked mechanisms
Strong evidence matters more than conventional credentials

“you shouldn't be stuck to your own field.”
Jean-Philippe Bouchaud · 53:54

“what you learn from one field can be very useful understanding another field.”
Jean-Philippe Bouchaud · 54:04

#mentorship#physics#interdisciplinary#career

Q&A· 1

Q&A33:39

Why CFM Does Not Trade Prediction Markets

Despite the growth of platforms such as Kalshi and Polymarket, CFM does not use them. The constraint is practical rather than philosophical: their liquidity cannot accommodate a fund operating at CFM's scale.

CFM has evaluated the category at a high level
Prediction-market liquidity remains insufficient
Institutional strategies need executable size
An interesting signal is not necessarily a tradable opportunity

“They're not liquid enough for us.”
Jean-Philippe Bouchaud · 33:44

“you need size and they can't provide it exactly.”
Barry Ritholtz · 33:47

#prediction markets#liquidity#polymarket#institutional investing

Takeaway· 2

Takeaway09:29

Why CFM Keeps One Foot in Academia

Bouchaud says academic work satisfies his drive to understand difficult problems while also helping CFM identify and recruit exceptional researchers. Publishing useful findings also lets the firm contribute data and insight that conventional academic teams may not possess.

Continued research supports intellectual curiosity
Academic visibility attracts PhD-level talent
Publishing creates an intellectually rewarding workplace
Data-rich firms can contribute to public research debates

“I feel I'm a researcher's researcher at heart”
Jean-Philippe Bouchaud · 09:29

“to attract talents, you need to identify them, you need to attract them, you need to be their professor at one point.”
Jean-Philippe Bouchaud · 10:11

#research culture#talent#academia#leadership

Takeaway56:52

Bouchaud's Advice to Aspiring Quants and Physicists

Bouchaud recommends studying theoretical physics, learning everything connected to data, and investigating an overlooked idea one genuinely believes in. The scale of the idea matters less than making the effort to build and test something real.

Study theoretical physics for analytical foundations
Develop strong data literacy
Choose an underinvestigated question
Build something rather than stopping at belief

“study theoretical physics and study everything that's related to data, and pay attention to data.”
Jean-Philippe Bouchaud · 57:06

“Make the effort of building something you strongly believe in.”
Jean-Philippe Bouchaud · 57:25

#career advice#quant finance#physics#data

Frameworks in this episode

Backtest-to-Production Model Gate

Data-to-Model Research Loop

Flow-First Market Model

Model Blind-Spot Override

Insights & moments

⚡Myth Buster· 2

Trend Following Is Not Dead After a Few Flat Years