Money Stuff: The Podcast
Episode: Re-Run: Cliff Asness | A Deep Dive into Quant Investing, Market Efficiency, and Machine Learning
Date: February 13, 2026
Host: Matt Levine (Bloomberg Opinion / Money Stuff Columnist)
Co-host: Katie Greifeld (Bloomberg News)
Guest: Cliff Asness (Co-founder, AQR Capital Management)
Overview
In this special re-run episode, hosts Matt Levine and Katie Greifeld are joined by legendary quant investor Cliff Asness, co-founder of AQR Capital Management. The discussion spans the role and function of quant investing, the evolution of financial markets, debates about market efficiency, the behavioral versus risk-based sources of market anomalies, the rise of machine learning in finance, and frank commentary on hedge fund fees and private equity. The tone is witty, conversational, and candid, with Asness offering both technical insight and humorous personal anecdotes.
Key Discussion Points & Insights
1. What Do Hedge Funds Do for a Living?
[02:04]
- Asness distinguishes: Between what quants like AQR “do for a living” (attempting to predict security prices) and what they “do for the world.”
- Market contribution:
- By taking the other side of trades, quants help move prices toward more accurate values or bear unwanted risks (e.g., holding merger arb after the main pop).
- Active managers may not always think in grand “market contribution” terms while trading.
- "They're thinking, is Nvidia undervalued?" — Cliff Asness [02:49]
2. Momentum Trading and Market Efficiency
[04:33]
- Competing academic explanations:
- Underreaction vs. overreaction: momentum profits may stem from people not fully incorporating news or from herd behavior/fomo.
- Asness finds it “embarrassing” these explanations are opposites—yet both may coexist.
- "It sounds embarrassing that the two major explanations, one is under reaction and the other is overreaction. When you've narrowed it down to two things that at least feel like the opposites, you should feel a little shame for a second." — Cliff Asness [05:09]
- If momentum arises from underreaction, momentum traders actually make prices more efficient; if overreaction, they might worsen efficiency.
3. Personal Anecdote: Meme Stocks, Twitter, and Trolls
[06:41]
- GameStop/AMC Shorting: Asness is candid about not remembering whether AQR was long or short GameStop during the meme stock mania, illustrating how diversified quant portfolios are.
- Twitter fallout: After disclosing a short on AMC, received intense flak from the “meme stock” crowd.
- "I discovered two things. They're not gonna like a short period. And crazy people don't always like being called crazy." — Cliff Asness [08:23]
4. Quants vs. Graham & Dodd Value Investors
[12:23]
- Meta-Graham & Dodd: Modern quant models are, increasingly, a holistic system similar to traditional value investing (cheap with positive catalysts).
- Differences remain: Graham & Dodd use in-depth, situation-specific knowledge; quants bet on characteristics working on average.
- Evolution: Quant value has moved closer to this broad, “quality + safety” approach, away from just low multiples.
- "Over time I've gotten a little less hubris about this." — Cliff Asness [14:38]
5. Machine Learning: NLP, Alternative Data, and the Arms Race
[16:51]
- NLP in quant investing: Quants once counted “good” and “bad” words in earnings calls; now they use natural language processing to extract more nuanced predictive signals.
- Alternative data arms race:
- Asness admits some data sets are kept proprietary for competitive advantage: "It would be a fireable offense from AQR to publish something that we thought was truly proprietary." [25:36]
- Academic publishing dilemma: Waiting too long to publish a finding risks getting scooped; publishing too early could spoil a lucrative edge.
6. The Role of PhDs and Firm Culture at AQR
[28:30]
- Unlike high-frequency shops (which hire more pure math/physics), AQR values a blend of finance, economics, and math skills for their holding periods.
- Asness’s “tortured analogy”: high-frequency is to quant investing as quantum mechanics is to Newtonian physics.
7. Renaissance, Market Makers, and Capacity Constraints
[30:22]
- Asness reflects on Renaissance’s Medallion Fund’s mystique—admiring their edge but noting it can’t scale to institutional money.
- Discusses market “fee for service” roles—e.g., Jane Street, and how many quant strategies are ultimately constrained by capacity.
8. Behavioral vs. Risk-Based Explanations for Anomalies
[34:06]
- Shifting views: Asness moved from favoring risk-premia explanations (75/25) to favoring behavioral explanations (now 25/75).
- Extreme markets: Personal and professional experience with the tech bubble and COVID-era meme stocks has shifted his worldview toward greater belief in extreme behavioral mispricings.
- "If the first one didn't make you start to go, this behavioral stuff may be real and maybe bigger than it used to be..." — Cliff Asness [36:12]
9. Fun in Markets: Meme Coins, Bragging Rights & Non-Economic Utility
[38:39]
- Discusses “non-economic” enjoyment of owning certain stocks (e.g., meme stocks, coins).
- Asness critiques market efficiency definitions that stretch to accommodate such phenomena.
- On the durability of non-fundamental bubbles: “The more absolutely unsubstantiated by anything something is, the longer the craziness can go on.” — Cliff Asness [43:45]
10. Active Management: Its Inherent Arrogance and Fees
[45:15]
- Being an active manager is “an inherently arrogant act”—the need to believe in excess ability/alpha but charge less for it than its value.
- Fee Setting:
- Highly unique alpha can justify higher fees; standardized factors (e.g., price/book) should have low fees, sometimes close to zero.
- Many pod shops charge what the market will bear, sometimes “insane” fees if their alpha is unique or hard to access.
- Industry Critique:
- Asness is clear: “Yes, [alts strategies] are too expensive.” [49:35]
- Study findings show that after fees, adding alts to a balanced portfolio often provides commensurately less return, not alpha.
11. Private Equity & Volatility Laundering
[53:13]
- Private equity seems attractive partly because it reports smoothed returns, not market volatility.
- Asness calls out “volatility laundering” and argues this leads to unfair fee advantages.
- "[PE managers] can mark it just like we market. They're brilliant at valuing companies… So it's like an institutional legal quirk that they get to do it one way and we have to do it another.” [53:49]
- Concern about expanding retail access to illiquids: “I think it's a terrible idea.” [58:28]
12. Machine Learning at AQR: From Skepticism to Adoption
[60:02]
- Asness admits he was initially hesitant, perhaps even slowed AQR’s ML adoption.
- Describes the shift: machine learning shifts the data vs. theory balance (from 50/50 to 75/25), and “throwing more at it” requires careful theory-guided constraint.
- "I had to be convinced it wasn't a light bulb moment." — Cliff Asness [60:08]
- Finds in practice that ML sometimes outperforms linear regression in modeling true return functions.
Notable Quotes & Memorable Moments
-
On finance and memes:
“Don’t you understand? You have to like every movie or else you’re anti-America.” — Katie Greifeld (about AMC) [10:14] -
On the competitive edge:
"It would be a fireable offense from AQR to publish something that we thought was truly proprietary." — Cliff Asness [25:36] -
On active management:
“Any active management is an inherently arrogant act.” — Cliff Asness [45:18] -
On alternative strategies pricing:
"Yes, they are [too expensive]." — Cliff Asness [49:35] -
On behavioral markets:
"I have a very cynical view…The more absolutely unsubstantiated by anything something is, the longer the craziness can go on." — Cliff Asness [43:45] -
On ML adoption:
"I think I probably slowed us down by a couple years in machine learning. I think it probably cost us some money." — Cliff Asness [60:08]
Timestamps for Key Segments
- [02:04] - Economic function of a quant/hedge fund
- [04:33] - Do momentum strategies improve or hurt market efficiency?
- [06:41] - Meme stocks, AMC, and Twitter fallout
- [12:23] - Quants vs. Graham & Dodd investing philosophy
- [16:51] - Natural language processing and the evolution of quant tools
- [23:14] - Publishing finance research vs. keeping proprietary edges
- [28:30] - AQR vs. other quant shops in personnel and approach
- [30:22] - The Medallion Fund paradox and quant “toll-keepers”
- [34:06] - Behavioral vs. risk explanations for market anomalies
- [38:39] - Markets as “fun” and meme coin investing
- [45:15] - Arrogance and fee philosophy in active management
- [53:13] - Private equity, “volatility laundering,” and retail expansion
- [60:02] - Asness’s changed mind on machine learning
Final Thoughts
This episode offers an uncommonly transparent look inside the mind of one of quant finance’s most prominent practitioners. Asness is forthright about his own learning curve, the limits of quant models, the reality of behavioral extremes in markets, and the importance of matching fees to true value-add. The conversation balances technical content with wit and practical market wisdom, making it an essential listen for finance professionals and enthusiasts alike.