Sep 26, 2026 · 1h 6m
AI portfolio agents face the test of investment reality
SI419: What Happens When AI Starts Building the Portfolio ft. Nick Baltas
The conversation examines whether agentic AI can improve portfolio construction without sacrificing reproducibility, accountability, or human judgment.
- 1Trend following benefited strongly from rates, but dynamic lookbacks showed little advantage over a simpler ten-month approach.
- 2Volatility scaling appears more practical than trying to predict reversals or time trend-model speed around market regimes.
- 3Agentic AI can structure investment debate, yet look-ahead bias, stochastic outputs, and oversight remain unresolved challenges.
Don't miss
Nick Balthas explains how specialized AI agents could debate assumptions and portfolio choices in natural language, while still requiring human oversight.
The brief
Nick Balthas joins Alan Dunne to assess strong trend-following results, rising global bond yields, market concentration, and where dispersion may create opportunities.
A volatility-conditioned trend model sounds adaptive, but the evidence discussed offers little reason to replace a simpler ten-month lookback, especially under a limited fixed-notional design.
The more durable lesson is operational: volatility scaling can shape exposure, while predicting reversals or choosing the perfect trend speed remains difficult.
Andrew Ang’s agentic AI framework divides asset allocation among specialized agents that analyze conditions, set assumptions, construct portfolios, and debate through natural language.
The standout tension is whether AI-managed products can be trusted when backtests risk look-ahead bias and model outputs remain stochastic, requiring human accountability.
What was said on this episode
10 statements · 7 positive · 3 negative
AI dominance is creating an internal index and reducing correlations among remaining stocks.
“The AI dominance is effectively building its own kind of many index within the index, and all the rest are becoming less correlated to it.”
Listen at 3:03
Equity decoupling is creating opportunities for dispersion strategies.
“this decoupling. in the equity space is creating opportunities for dispersion.”
Listen at 3:54
Agentic AI could automate or abstract parts of institutional asset-management processes.
“the transition into agentic um ai could have implications in many things that you know historically we used to do using people and processes and time and various parts of that process could be now abstracted.”
Listen at 33:39
Human portfolio analysts and builders are limited in scale and attention.
“Human beings will do the analysis historically. Human beings will build their portfolios and so on and so forth. But the scale at which they can operate and the attention they can have is limited by the fact that we operate in real time and in human being capacities.”
Listen at 43:26
Agentic AI should augment human supervisors rather than replace them.
“there's an element of supervision rather than replacement now that should allow us to elevate the level of abstraction”
Listen at 43:41
Agentic portfolio systems can produce board memos with fully auditable decision trails.
“getting us the output, from the CIO agent the board memo for the actual board to then decide upon with an audit trail through to be fully auditable.”
Listen at 43:55
Mixed economic-regime evidence leads allocators to favor greater diversification.
“When the economic regime is more mixed in terms of evidence and data points, to categorize it with some confidence, we typically vouch more for diversification.”
Listen at 47:47
Investors should diversify more when macroeconomic outcomes are unclear.
“should we instead... be more prone to diversify at the times that we cannot tell how the macro outcome is clear”
Listen at 48:29
Unclear regimes reduce confidence in capital-market assumptions.
“I'm not trusting my capital market assumptions because the regime is not Very clear.”
Listen at 49:26
Unclear regimes warrant tilting allocations toward risk-based and heuristic frameworks.
“I should then rather tilt my allocation towards risk-oriented and more heuristic allocation frameworks rather than trusting the data as much.”
Listen at 49:34
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.