Jun 10, 2019 · 1h 18m · capital-allocators
Patrick O'Shaughnessy – O'Shaughnessy Asset Management (First Meeting, EP.01)
0:00 / 0:00
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In the premiere episode of First Meeting, host Ted Seides interviews Patrick O'Shaughnessy, CEO of O'Shaughnessy Asset Management, exploring the firm's systematic factor investing models, non-linear portfolio construction, machine learning applications, and the power of open-source research platforms.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Ted holds 17.4% of the talking time here. How this is scored →
Ted as informed peer 3.4 Guest teaching 5.3 Guest disagreement 0.5 Ted pushing back 0.9
speaking balance: gold is Ted, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 31:40 Patrick declares quants the enemy of fundamental stock pickers
Patrick emphatically states that quantitative models will inexorably eat discretionary stock picking whenever a repeatable data pattern exists.
Hardest push from Ted ▶ 43:20 Ted challenges Patrick on sharing alpha vs risking decayTed directly challenges Patrick on how publishing quantitative research does not immediately arbitrage away the firm's edge.
Biggest teaching moment ▶ 27:55 Patrick explains why ML fails to predict broad market returnsPatrick breaks down the mathematical limitation of applying ML to non-stationary financial returns, demonstrating why return prediction fails while narrow label classification succeeds.
Ted holds their own ▶ 22:46 Ted presses Patrick on competing with Renaissance and AQRTed leverages his deep allocator background to drill Patrick on how a boutique like OSAM can realistically compete for factor hypotheses against massive quant institutions.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Ted as informed peer | Guest teaching | Guest disagreement | Ted pushing back | Why |
|---|---|---|---|---|---|---|
| Introducing First Meeting and Guest Patrick O'Shaughnessy | 0 | 0 | 0 | 0 | Ted delivers an introductory solo monologue explaining the launch of the 'First Meeting' spinoff series and introducing his close friend and guest Patrick O'Shaughnessy. Host-side scores are zeroed out as Patrick does not participate. | |
| Upbringing, the 'Look It Up' Philosophy, and Joining OSAM | 2 | 3 | 1 | 0 | Patrick jokes about feeling uncomfortable on the other side of the microphone before detailing his childhood curiosity and serendipitous entry into OSAM. Ted facilitates with straightforward biographical prompts. | |
| The Four Core Factors and Three Sources of Equity Returns | 3 | 7 | 1 | 1 | Ted asks about OSAM's core factors, prompting Patrick to give a comprehensive breakdown of the three sources of equity returns (valuation, growth, return of capital) and explain why OSAM uses quality primarily as a negative screen rather than a positive factor. | |
| The Dynamics of Valuation vs. Price Momentum | 4 | 6 | 1 | 2 | Ted presses on the counterintuitive nature of price momentum versus operational business momentum. Patrick clarifies the differing holding periods and decay profiles between value and momentum alpha. | |
| The Three Silos of Quantitative Research at OSAM | 3 | 6 | 0 | 1 | Ted asks what changes Patrick has implemented as CEO. Patrick outlines OSAM's three research silos: factor refinement, hypothesis-driven factor exploration (citing the R&D failure), and machine learning label research. | |
| Differentiating OSAM from Academic Factor Quants | 5 | 6 | 1 | 2 | Ted challenges Patrick on how OSAM can compete against massive quant incumbents like Renaissance and AQR. Patrick explains OSAM's focus on pure alpha over tracking error and its non-linear modeling rather than linear regression. | |
| Non-Linear Portfolio Construction in the Factor Tails | 4 | 5 | 0 | 1 | Ted drills down on how OSAM builds portfolios differently. Patrick details how they strip out the worst deciles and only purchase the top decile, resulting in concentrated, low-overlap portfolios. | |
| Machine Learning Applications and Return Prediction Limits | 4 | 7 | 1 | 1 | Ted asks Patrick to clarify why quants cannot simply let machine learning predict returns directly. Patrick explains that market returns are non-stationary, so ML is better suited for predicting specific events like dividend cuts or parsing regulatory filings. | |
| Sponsor Message: Ridgeline AI-Native Investment Technology | 3 | 6 | 1 | 1 | After an ad break, Ted asks about the macro impact of quants on traditional stock pickers. Patrick bluntly calls quants 'the enemy' of discretionary stock picking whenever repeatable data patterns exist. | |
| Starting 'Invest Like the Best' and the Art of Interviewing | 3 | 4 | 0 | 0 | Ted shifts to discussing Patrick's podcasting journey. Patrick shares his origin story with Jeff Gramm and Michael Mauboussin and highlights the critical shift toward doing less prep and active listening. | |
| The Knowledge Loop: Learn, Build, Share, Repeat | 3 | 6 | 0 | 0 | Ted asks how podcast learnings translate into investing. Patrick lays out the four-stage knowledge loop (Learn, Build, Share, Repeat) and explains why building and sharing refine understanding and attract inbound alpha. | |
| Sharing Behavioral Alpha vs. Proprietary Data | 4 | 5 | 1 | 2 | Ted presses Patrick on the paradox of publicly sharing quantitative research without destroying alpha. Patrick distinguishes between behavioral alpha (which persists when shared) and informational edge (which must be kept private). | |
| Applying Tech Platform Models and AWS Principles to OSAM | 3 | 6 | 0 | 0 | Ted asks how tech business models influence OSAM. Patrick describes adopting AWS principles by turning internal infrastructure into platforms and creating the OSAM Research Partners program. | |
| Twitter as an AGI and the Jesse Livermore Partnership | 3 | 5 | 0 | 1 | Ted inquires about sourcing collaborators on Twitter. Patrick explains Zach Cantor's concept of a high-quality Twitter network acting like an AGI and details collaborating with pseudonymous researcher Jesse Livermore. | |
| The Future of Quantitative Investing in Private Markets | 4 | 5 | 0 | 1 | Ted asks whether quantitative approaches can extend into private markets. Patrick evaluates the structural constraints of small sample sizes and relationship-driven deal structures while highlighting firms like CircleUp. | |
| External Capital Allocation and Backing Deep Basin Capital | 4 | 5 | 0 | 1 | Ted asks about OSAM's family office external allocations. Patrick explains using the podcast for partnership discovery and discusses backing long/short energy specialist Deep Basin Capital. | |
| Granular Energy Modeling and Brent Beshore's Private Equity Edge | 4 | 5 | 0 | 1 | Ted probes into what OSAM learned from Deep Basin's energy modeling and contrasts it with Brent Beshore's low-multiple private equity strategy. | |
| Due Diligence on Quants and the 'Research Graveyard' | 5 | 7 | 1 | 1 | Ted asks how allocators should evaluate quant managers. Patrick shares his best due diligence framework: inspecting the manager's 'research graveyard' of failed projects (such as factor timing) and assessing their data hygiene efforts. | |
| The Structural Hurdles of Quantitative Short Selling | 4 | 6 | 1 | 1 | Ted notes that short selling was absent from OSAM's framework. Patrick candidly explains their three failed internal attempts at shorting, citing borrow costs, availability friction, and volatility mismatch. |
Statements from this episode (37)
O'Shaughnessy: Key Investment Factors Historically Succeeded in Only 60-70% of Single Years
“In our research, luckily, those categories are honestly straightforward and simple. They're things like valuation, things like momentum, quality, return of capital. They're not crazy concepts. You just have to apply them with a lot of discipline, and I think t…”
O'Shaughnessy: Factor Strategies Like Momentum Should Never Be One-Off Trades
“So unless you really buy into these concepts at a deep level and are willing to stick with them through cycles, I think you probably shouldn't apply them at all. You shouldn't apply momentum into a one off trade. You should apply it in the right way systematic…”
O'Shaughnessy: Equity returns only stem from growth, multiple expansion, or capital returns
“When you're holding for that long, and you think about equities at an elemental level, I think there's just three ways to earn return. The businesses you own grow. In sales, earnings, EBITDA, whatever it is, free cash, the multiples of those fundamentals expan…”
O'Shaughnessy: Momentum forecasts fundamental growth better than valuation multiples
“What fascinates me about this is momentum does a better job of forecasting future fundamental growth than price does. So the highest momentum decile has better future fundamental growth than the most expensive, let's say PE decile or free cashflow yield decile…”
O'Shaughnessy: High quality metrics do not generate excess returns; low quality destroys them
“In the quantitative sense, high ROEs or gross profitability or whatever your measure of quality is really isn't indicative of strong future excess returns. What's really useful about quality is that very bad quality, so really over levered balance sheets, comp…”
O'Shaughnessy: Companies buying back 5%+ shares annually historically outperform peers
“Super normally high buyback yields, and a simple definition of this might be a company which is buying back five percent of its shares or more in a one-year period of time, which is a lot. That's a big deployment of capital. That kind of company historically h…”
O'Shaughnessy: OSAM Measures Momentum Across 3-12 Month Windows and Adjusts for Volatility
“What's fascinating about pure price momentum and the way that we think about this is three, six, nine, 12 month trailing windows of total return versus pierce. We think we should also care about the volatility of the momentum.”
O'Shaughnessy: Top Decile Momentum Stocks Show Supernormal Operational Growth in Year One
“All we know is that of the highest momentum decile stocks, just in very, very simple terms, those stocks in the next year have super normally high operational growth.”
O'Shaughnessy: Value Stocks Continue Generating Marginal Alpha for Up to 10 Years
“So if you buy in value today, you, on average, earn a little bit of alpha for a really long time. Most of it's early on, but you continue to earn marginal monthly alpha for up to 10 years.”
O'Shaughnessy: Momentum Alpha Exhausts Within One Year and Reverts to Negative Returns
“So you earn all of your momentum, all of your alpha, rather, in the first one year holding period, and then you actually need to get the hell out, because it reverts. And typically, depending on where you're looking at the data, within two years or so, it's ac…”
O'Shaughnessy: R&D spending factor added zero incremental alpha to master model
“So we did a lot of work around like R and D type spending. And when you plug it in, it Is both value and momentum in some interesting different ways and sectors. But when you plug it in into the master model as a factor, it was completely gone. It literally di…”
O'Shaughnessy: Almost No Funds Have Pure Machine Learning Models in Live Production
“To be clear, I don't think almost anybody has a lot of that in production today, but those tools are becoming more and more useful.”
O'Shaughnessy: Reducing tracking error is much easier than finding new alpha
“It's actually far easier to shape down tracking error than it is to find new alpha. I could run a much lower tracking error process with not that much additional research. I can't just go find new alpha.”
O'Shaughnessy: OSAM has almost zero portfolio overlap with AQR and LSV
“Our overlaps with other quants are shockingly low. Like we have this little cool program that we can show people this on the fly and they won't believe us. Like they'll ask about AQR and LSV and these other great managers, you know, really, really, really stro…”
O'Shaughnessy: Negative stock returns are concentrated in the worst factor decile
“Most of the really interesting return, in this case, negative, bad return that you want to avoid, is concentrated in the tails, in the worst decile.”
O'Shaughnessy: Machine learning algorithms always fail at predicting stock returns directly
“You can feed all the best data in the world to whatever ML algorithm you choose. Let's say it's like a decision tree or something. And if you're trying to predict returns, it always falls apart. It never works.”
O'Shaughnessy: OSAM Uses Machine Learning to Predict Specific Events Like Dividend Cuts
“So as a result, we have started to focus more on applying these techniques to focus on something like a dividend cut. Let's use that as an example. Build a model that just forecasts or predicts dividend cuts, and that's it.”
O'Shaughnessy: Quants will inevitably replicate any repeatable investment strategy
“If there is a repeatable strategy where we can build a big enough sample size and do the right kind of research, a quant is going to figure it out. And the question is when, not if.”
O'Shaughnessy: Qualitative analysis and portfolio concentration protect discretionary managers
“If you can handicap future scenarios based on all that you know about a business or an industry or a trend or whatever it might be, that's going to be hard for quants to replicate. And if you can concentrate In portfolio construction, that's also going to be v…”
O'Shaughnessy: OSAM Strictly Prohibits Researchers From Discarding Failed Studies
“And in our case as a business building, we've got a pretty strict rule around OSAM when it comes to building. You can't just do a research study and if it doesn't work, just like forget about it. You need to bring it into our ecosystem in a way that can be use…”
O'Shaughnessy: Publicly sharing quant findings produces stronger signals via feedback
“In quant research, this is really interesting, which is that you actually get a stronger signal. We found when you share your general findings, not necessarily the code or the particulars, but the general findings with a broad audience, then all of a sudden yo…”
O'Shaughnessy: All OSAM investment factors have behavioral explanations
“So I would argue all the factors that we use here have a behavioral component, or at least a compelling behavioral explanation for why they work.”
O'Shaughnessy: Factor alpha comes from execution discipline, not awareness
“That says that the source of the excess return is not knowing about the thing. It's doing the thing.”
O'Shaughnessy: No asset manager is applying tech platform principles holistically
“Asking whether or not we can apply those principles to a very simple old school asset management business. And the answer, at least in my view, unequivocally so far as yes, and nobody else is doing it. Or at least they're not doing it holistically as part of h…”
O'Shaughnessy: OSAM Has Seven External Research Partners, Including Jesse Livermore
“We now have seven of these people. It started with a very well-known anonymous writer who goes by Jesse Livermore online. He was sort of our pilot case.”
O'Shaughnessy: A High-Quality Twitter Following Functions Like Having an AGI
“That having a big, smart, diverse group of people that follow you on Twitter is literally like having an AGI. I believe that that is true. It's the best search engine in the world. So if you're looking for something, ask a big group of people, you're going to …”
O'Shaughnessy: 30% to 50% of Top Thinkers in Any Field Are on Twitter
“Something like, 30 to 50% of the most thoughtful people in that domain are gonna be on Twitter talking about what they do.”
O'Shaughnessy: Pure quantitative strategies fail on concentrated portfolios
“If you're going to build a concentrated portfolio of a handful of positions, you can't do quant. It can't be pure quant. The stats don't work. You need to build a pretty diversified portfolio for a pure quant approach to make sense.”
O'Shaughnessy: Quant methods will never fully replace relationship-driven private equity
“Then your relationship with the seller, your skill in negotiation and deal structure, all of these things are critical and Again, can quant help inform some of that? Probably, but I don't think it's ever going to completely eat the whole thing.”
O'Shaughnessy: Quants typically avoid sector-specific data in favor of broad cross-sectional metrics
“Typically quants don't do deep industry sector specific
Data sets outside of like financials, which has been like a common problem for quants. Quants typically are looking for something you can compare cross-sectionally across the entire population of stocks.”
O'Shaughnessy: 4x Free Cash Flow Signals Distress in Public Equities, Not Private
“And, you know, when he first told me that he was buying businesses at a little bit north of four times free cashflow, you know, I kind of laughed. I thought, okay, like, so they're going out of business in three years. I mean, in the public markets, that multi…”
O'Shaughnessy: Best quant researchers are defined by extensive failure experience
“The best way to know that someone's going to be an effective and efficient researcher going forward into the future is that they know what not to do. And I think the only way to do that is to build up that experience of failure through time.”
O'Shaughnessy: OSAM failed to make factor timing work despite massive research effort
“We have burned an insane amount of our time Capital, human capital, stress, just every kind of capital trying to make this work, and it has not.”
O'Shaughnessy: OSAM team spent three full years cleaning primary dataset
“We once spent three full years, the entire team spent three full years just cleaning and scrubbing data in our primary data set.”
O'Shaughnessy: Predictive modeling will get easier, shifting advantage to input data
“I believe that the predictive modeling part of this is going to get easier and easier. So what that means is what you put into those models is of critical importance.”
O'Shaughnessy: OSAM Failed Three Times Attempting Short Strategies Internally
“We've done it with our own capital three times and failed miserably each time, not an absolute return sense, but certainly an opportunity cost sense because it's been during this great run up. So we've lost a lot of return with our own capital trying to do thi…”
O'Shaughnessy: Borrow Costs and Volatility Eliminate Quantitative Short Paper Alpha
“Most of the stuff that screens as screaming shorts and in a paper portfolio works incredibly well. You have incredible problems with borrow availability, with borrowing cost, which often instantly removes the paper alpha and with volatility.”