About
About.
This note sets out the rationale behind the launch of The Japan Consumer Pod. It reflects several years spent in the investment industry, watching active management become steadily less convincing in the very areas where it once claimed an advantage.
The problem is not mysterious. Traditional active management is under pressure for two reasons. The first is chronic underperformance. The second is a fragile business model. These are separate issues, but they feed each other. Poor outcomes weaken the economics of the industry; weaker economics degrade the quality of the work; degraded work makes outperformance even harder. That loop has been running for some time.
Start with performance. A large part of the industry has drifted into a standardized form of investing: similar processes, language, guardrails, and portfolios. Many funds still charge active fees while operating under such tight benchmark awareness that true deviation has become rare. When everyone reads the same research, reacts to the same narratives, worries about the same factor exposures, and is judged against the same index in the short term, genuine differentiation disappears long before fees do.
Over time, this has hollowed out something more important than style. It has weakened investment philosophy itself. Too many portfolios are built without a clear view of where mispricing is most likely to exist, what kind of edge is being pursued, or what sort of mistakes the process is explicitly trying to avoid. The result is predictable: generic strategies, low-conviction portfolios, and a lot of expensive activity that does not add much value.
The misallocation of time and budget has made the problem worse. A surprising share of industry effort now goes to low-impact work: macro noise that rarely changes a company thesis, internal reporting layers, commercial pressure, administrative burden, and forms of process that are defensible institutionally but not especially useful analytically. Meanwhile, the actual craft of fundamental research, deep company work, longitudinal comparison, and disciplined follow-up, often gets squeezed.
The human economics of the business have also deteriorated. Financial alignment for portfolio managers and analysts has weakened in many places. The cost of maintaining strong buy-side research teams remains high. On the sell-side, the contraction has been visible for years, especially after MiFID II. Research teams are smaller, coverage is thinner, small and mid-cap names are often neglected, and much of what remains is too generic to be genuinely differentiated. None of this helps active managers produce better work. It merely lowers the average quality of the inputs while leaving the expectations intact.
Fees then complete the damage. If performance is mediocre and the cost remains meaningfully above passive alternatives, the comparison becomes brutal. Investors are not irrational in moving assets to passive vehicles, especially ETFs. Lower fees matter. Simplicity matters. Transparency matters. Tax efficiency matters. So does the growing distrust of discretionary decision-making after years in which much of the industry failed to justify its discretion.
That brings me to the second problem: the business model itself. Small and mid-sized active managers are being squeezed from both sides. Operating costs keep rising, while management fees come under pressure. Margins compress, which means fewer resources for research, technology, and talent. Fewer resources usually mean weaker coverage and lower-quality work. Lower-quality work makes it harder to generate alpha. And the cycle starts again.
The likely consequence is further consolidation and a clearer polarization of the industry. At one end will sit a handful of very large firms, protected by scale, distribution, and broad operating infrastructure. At the other end, I suspect, will be a smaller group of highly specialized firms with a clear identity, a narrow hunting ground, and a process that actually looks different from the benchmarked middle of the market. The segment most at risk is the one in between: too small to enjoy scale, too broad to claim true specialization, too expensive relative to passive, and too undifferentiated to command lasting loyalty.
I do not think active management disappears. I do think it has to earn its right to exist again. In practice, that probably means something narrower, more selective, and more explicit about where its edge is supposed to come from. Investors who want broad market exposure already have good tools for that. There is little point in offering them index-like portfolios with a more elaborate story attached. Active management only deserves capital when it is genuinely active, genuinely specialized, and operating in parts of the market where careful work still has a fair chance of being rewarded.
So where can active still make sense? Alpha is more likely to exist where market structure creates neglect, crowding, or simplification. The rise of passive has not made every part of the market perfectly efficient. In some places it may have done the opposite. Capital has become increasingly concentrated in the same benchmark-heavy companies. ETF flows can reinforce crowding. Sell-side coverage keeps shrinking in less fashionable areas. Smaller companies, less obvious sectors, more locally nuanced stories, and markets perceived as difficult or unglamorous can end up receiving far less serious attention than their economic importance would justify.
That is the backdrop for The Japan Consumer Pod. I am a former hedge fund analyst building a pod of AI agents focused on Japanese consumer equities. This is a deliberately narrow strategy. I am not trying to build a generalist platform with an artificial intelligence label attached to it. I am trying to build a better research system for a specific part of the market where I believe disciplined specialization still matters.
Japan consumer is a particularly attractive hunting ground because it sits at the intersection of three forces the market still does not digest evenly. First, Japan is no longer the static, deflationary market many investors still carry in their heads. Wage growth has remained unusually strong, the Bank of Japan has moved policy rates up to 0.75% and may tighten further, and draft revisions to the Corporate Governance Code are increasing pressure on balance sheets and capital allocation. In consumer sectors, that regime change matters company by company. Pricing power, wage absorption, traffic resilience, inventory discipline and capital allocation are no longer background variables; they are becoming central drivers of dispersion.
Second, the sector offers an unusually rich stream of high-frequency operating data. Many listed Japanese consumer companies disclose monthly sales, same-store sales, traffic and ticket trends, which makes it possible to track inflections in demand and execution with far more precision than in many other industries. That is especially valuable when combined with a research system that can monitor the sector continuously rather than episodically. Third, despite that data richness, analytical coverage remains uneven outside the largest names, particularly in mid-caps and domestically focused businesses, where language, local context and long institutional memory still matter. That combination — macro regime change, abundant live operating data, and patchy coverage — makes Japanese consumer a fertile area for original, specialist research.
The role of AI in this effort should be stated plainly. It is not magic. It is not a claim of invincibility. It is not a substitute for judgment, accountability, or risk control. And it is certainly not a promise that machines will somehow remove uncertainty from markets. The case for using AI is much more practical than that.
A well-designed pod gives a research effort something close to continuous analytical bandwidth. It can monitor filings, transcripts, presentations, guidance changes, pricing moves, distribution signals, local news flow, management language, and peer developments around the clock. It can retain a far larger memory of past company behavior than a small traditional team. It can compare opportunities across a wide universe, repeatedly and consistently, at a marginal cost that would have been difficult to imagine in a purely human setup. Most importantly, it can do this without fatigue, without boredom, and without the emotional drift that so often contaminates investment work.
That matters because some of the most expensive errors in fundamental investing are not informational in the classic sense. They are behavioral. Investors anchor on old views. They chase recent moves. They seek confirming evidence. They defend sunk work. They become reluctant to revisit a thesis once it has been socialized. They confuse activity with rigor. These tendencies are persistent because they are human. They are also one reason mispricings do not disappear. A properly governed AI pod can reduce part of that burden by preserving memory, forcing consistency, surfacing disconfirming evidence, and keeping comparative discipline alive even when attention would otherwise drift.
The real differentiator, though, is not simply that the system can read more. It is that it can compare more. Good investing often comes down to relative judgment: this company versus that one, this management team versus another, this margin structure, this pricing logic, this store economics, this capital allocation pattern, this promise versus this history of delivery. The ability to run deep transversal analysis across many opportunities at once, and to do so continuously, is where I think the research edge becomes meaningful. Outside of quantitative investing, this kind of simultaneous qualitative comparison was traditionally much harder to sustain at scale.
But access to AI models alone is not an edge. That will become obvious quickly. Many firms will use similar tools. If there is an edge, it will not come from the model itself but rather from the design of the research system, the sector taxonomy, the internal memory, the protocols for comparison, the rules of evidence, the governance of decisions, and the discipline with which the process compounds over time. Said differently, the edge does not come from having AI. It comes from how the research system is designed, and from the ability to apply it consistently over time.
The objective here is modest in wording, even if demanding in execution. I am not trying to sell a martingale. I am not claiming to be building the best investment platform in the world. The aim is more sober than that. It is to create a disciplined process, focused on situations where the probability of mispricing is meaningfully higher, and to use that process to outperform passive exposure over time. That is a high bar already. It does not require grandiosity. It requires clarity, consistency, and patience.
The approach is ambitious. The claims are not. Markets remain humbling. Judgment still matters. Human responsibility remains at the center of the process, especially when it comes to weighing soft information, calibrating risk, interpreting management, and deciding when not to act. What I am building is not a replacement for that judgment. It is a way to support it with a better research engine than the one most small and mid-sized active managers can realistically afford to build in the old model.
If active management is to remain relevant, I believe it will do so through narrower strategies, better tools, stronger philosophy, and a clearer sense of where mispricing still lives. That is the bar I am setting for this project.