Jul 29, 2021 · 1h 5m · capital-allocators
Julia Bonafede – Rosetta Analytics (Manager Meetings, EP.04)
0:00 / 0:00
⌖ your search result is the highlighted band (32:28–33:01). Playback starts there
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of Manager Meetings, Verger Capital CIO Jim Dunn interviews Rosetta Analytics co-founder Julia Bonafede to explore her trailblazing career at Wilshire Associates, the limitations of traditional factor models, and how deep reinforcement learning is transforming institutional equity investing and absolute return strategies.
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 10.6% of the talking time here. How this is scored →
Ted as informed peer 4.9 Guest teaching 3.3 Guest disagreement 0.3 Ted pushing back 0.1
speaking balance: gold is Ted, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 52:20 Calling out quant manager failures in March 2020
Bonafede bluntly critiques traditional quant and HFT firms, noting that their crowded factor models broke down severely during the early 2020 pandemic volatility.
Hardest push from Ted ▶ 43:56 Seatbelt analogy challenging slow institutional adoptionDunn challenges the industry's refusal to evolve from MPT by comparing institutional resistance to AI with drivers resisting seatbelt adoption in the 1950s.
Biggest teaching moment ▶ 32:30 Deconstructing factor models versus deep reinforcement learningBonafede delivers an extensive breakdown showing how traditional factor models leave alpha on the table compared to deep neural networks that dynamically adapt parameters.
Ted holds their own ▶ 5:57 Dunn exposes modern portfolio theory stagnationDunn articulates a sharp allocator critique, demonstrating deep domain knowledge on why computing power advancements have made traditional asset allocation frameworks obsolete.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Ted as informed peer | Guest teaching | Guest disagreement | Ted pushing back | Why |
|---|---|---|---|---|---|---|
| Verger Capital's Investment Thesis in Rosetta Analytics | 6 | 0 | 0 | 0 | Ted Seides interviews Jim Dunn about why Verger Capital decided to seed Rosetta Analytics. Dunn outlines his thesis around computing power outstripping traditional modern portfolio theory and treating Rosetta like an innovation skunk works. | |
| Julia Bonafede's Unconventional Path into Finance | 4 | 2 | 0 | 0 | Jim Dunn opens his interview with Julia Bonafede, asking about her background. Bonafede recounts growing up on a cattle ranch, attending USC, and joining Wilshire Associates over investment banking. | |
| Establishing Wilshire's European Presence as a Female Pioneer | 4 | 3 | 0 | 0 | Bonafede recounts opening Wilshire's first international office in London, navigating European fixed income markets and breaking barriers as a pregnant female executive on trading floors. | |
| Fostering Innovation and Scaling Institutional Consulting | 5 | 3 | 0 | 0 | Bonafede explains scaling Wilshire's consulting division by educating clients on hidden beta and managing slow institutional governance cycles. | |
| Fiduciary Responsibility and Institutional Active Management | 5 | 4 | 1 | 0 | Bonafede details the fiduciary reality of institutional investing, noting that short-term pressures force long-term investors into suboptimal active strategies and criticising peer group benchmarking. | |
| Post-GFC Market Shifts and Consulting Business Evolution | 5 | 3 | 0 | 0 | Bonafede walks through the aftermath of the 2008 GFC, explaining how fee compression and low scalability pushed consulting firms toward OCIO models. | |
| Macroeconomic Headwinds and Structural Cracks in Asset Management | 6 | 4 | 1 | 0 | Bonafede critiques the current asset management landscape, arguing that fixed income business models are broken in low-yield environments and private markets are becoming overly leveraged. | |
| The Evolution from Factor Models to Deep Reinforcement Learning | 5 | 6 | 1 | 0 | Bonafede delivers a deep breakdown contrasting static multi-factor models with dynamic deep reinforcement learning capable of autonomously updating parameters. | |
| Ridgeline Sponsor Message | 5 | 4 | 0 | 0 | Following the sponsor break, Dunn prompts Bonafede to define neural networks. She outlines backpropagation and nonlinear multi-period optimization. | |
| Deploying AI Strategies within Liquid US Equities | 5 | 5 | 0 | 0 | Bonafede explains how Rosetta applies deep reinforcement learning to liquid US equities to capture alpha while generating lower volatility than traditional equity indices. | |
| Institutional Pushback and Barriers to AI Adoption | 6 | 4 | 0 | 1 | Dunn compares slow AI adoption to historical resistance to car seatbelts. Bonafede discusses legacy institutional inertia and the difficulty of explaining non-linear models to committees. | |
| Behavioral Biases, Emotional Humility, and Model Discipline | 5 | 4 | 0 | 0 | Dunn and Bonafede discuss behavioral biases, emotional detachment, and the necessity of trusting model discipline rather than overriding algorithms during drawdowns. | |
| Differentiating Rosetta from Traditional Quant and HFT Firms | 5 | 5 | 1 | 0 | Bonafede differentiates Rosetta from high-frequency quant funds, pointing out how crowded traditional quant factor models suffered during the March 2020 crash. | |
| Transitioning from Demanding Consultant to Investment Manager | 5 | 2 | 0 | 0 | Dunn asks Bonafede about the shift from being a demanding institutional consultant to sitting on the manager side raising capital. Bonafede emphasizes empathy for allocator fiduciaries. | |
| Fundraising Hurdles for Emerging and Diverse Managers | 4 | 3 | 0 | 0 | Bonafede explores the high operational hurdles emerging and female-led managers face despite stated institutional interest, followed by leadership philosophy. | |
| Personal Daily Habits, Upbringing, and Finding Joy | 3 | 1 | 0 | 0 | The conversation closes with rapid-fire questions on gratitude, parental aphorisms, and distinguishing joy from transient happiness. |
Statements from this episode (22)
Dunn: Verger Capital seeded Rosetta Analytics with $5M
“So we started with sort of a five million dollar investment and it grew from there. So we now have Four strategies with Rosetta, and three of them have been very, very successful. One of them has been turned off.”
Bonafede: Emerging markets did not open to US investors until 1992
“The emerging markets didn't open up to U.S. Investors until 1992, and so being able to be on the forefront of that and run that business essentially for a few years really gave me a huge view as to how non-U.S. Investors and U.S. Investors looked at investing”
Bonafede: Every investment fundamentally reduces to lending or owning
“When you put a portfolio together that there is hidden beta in everything because everything is either the pure stream. You're either Lending or your owning, right? And then you put a lot of complication around those two concepts in whatever you're structuring…”
Bonafede: Pension funds move like icebergs, taking years to adopt ideas
“Pension funds work like icebergs. Okay, so you might start the concept and then many years later you might get to where you actually get movement. And so understanding that you can't have instantaneous decisions and outcomes with clients, especially clients th…”
Bonafede: Most institutions cannot simply index their portfolios
“Jack Bogle had a view of Really, you should index everything, and over the long term, you'll get there at much lower fees, and everything will compound, but the reality is, is that most institutions can't invest that way for just by their very structure, or by…”
Bonafede: Peer group benchmarking is consultants' biggest disservice
“Probably the biggest disservice that consultants have ever done is putting everybody in peer groups so that they compete against each other, even though that their liabilities are completely different and their circumstances are very different.”
Bonafede: Consultants enter OCIO to capture asset management profit margins
“Consulting today, that's why you've seen a lot of consultants go into the OCIO business because they need to find a business that has asset management margins because you really can't scale the business.”
Bonafede: The active fixed income business model is in trouble
“So how do fixed income managers actually charge active fees on top of duration risk and low yields, where investors are going to be looking at most of their portfolios, and you're earning close to zero, and on a real basis, you're earning negative. So that who…”
Bonafede: Asset management must undergo a major structural correction
“In my view, portfolio, the asset management industry is really has to go through a big correction before any of this will shake out, but you've seen unprecedented mergers and acquisitions.”
Bonafede: Traditional factor models leave substantial returns on the table
“The traditional way of measuring relationships really leaves a lot of information on the table. You're either describing a small portion of that return, or you're just building portfolios that don't have, aren't dynamic in terms of how those relationships chan…”
Bonafede: Rosetta Analytics uses deep reinforcement learning for portfolio optimization
“So in our case, we use neural networks to make a directional prediction, but we have actually taken the neural network and we're using a concept called deep reinforcement learning, which is, you can think of it as a decision framework, but an optimization that…”
Bonafede: Deep reinforcement learning discovers relationships traditional optimizers miss
“But in a deep reinforcement learning process, you're using that neural network, and it has the ability to learn new relationships where all other optimization functions will only use the parameters or the factors or the framework that the model itself is capab…”
Bonafede: Allocators typically sacrifice returns to diversify risk
“There are many allocators who have an absolute return component of their portfolio. You're, Typically trying to diversify risk with that allocation, but you end up sacrificing return in the process.”
Bonafede: Quantitative factors are spurious because timing them is impossible
“Now we all know that factors are spurious, but that's because they appear and they disappear at certain times, and you don't know exactly when to make that bet.”
Dunn: In 1955, 90% of US car buyers rejected a $27 seatbelt
“37,000 Americans died in car accidents in 1955. That's six times today's rate adjusted for miles driven. And it was because seat belts were a 27 dollar upgrade And 90% of people didn't want that upgrade.”
Bonafede: Business schools still teach static linear models for portfolio construction
“All of the skills that are being taught are based on modern portfolio theory, which all of those, nearly all of those equations are in an APT framework or a linear regression framework, and even traditional optimization methods used for portfolio construction.…”
Bonafede: AI investing will stay in absolute return before going long-term
“I think it's going to have to stay in places like absolute return, because really what these models are capable of, there's going to take a lot of research, but they're shorter term right now. Eventually the research will change to longer term.”
Bonafede: Traditional investment processes fail to justify active management fees
“If you think about how rudimentary the other processes are that literally trillions of assets are invested in, at this point in this seat, I'm not quite sure why everybody feels so comfortable with their current frameworks. Because again, you could index and p…”
Bonafede: Traditional quant managers faltered in March and April 2020
“If you see how they performed during March and April, 2020, they faltered with all of the other quant managers and the products that they were selling to clients.”
Bonafede: The 2020 quant breakdown would have worsened without Fed intervention
“Had the Fed not stepped in, you probably would have seen the true aftermath of what Can happen with concentrated investments in the same kind of relationships in the same kind of data.”
Bonafede: Only reinforcement learning can dynamically moderate portfolio risk
“As it learns to maximize the reward, the model, I don't mean to give it human characteristics, but it will moderate risk because it can see that taking full advantage of the underlying patterns may not lead to the best reward, and there's No other optimization…”
Bonafede: High onboarding costs hinder allocators from backing emerging managers
“I think it's difficult to onboard a manager in general, so I think probably the biggest hurdle to overcome is when we're an emerging women-led manager, so there's a lot of assets that are looking for opportunities to be able to help firms like us grow. However…”