My conversations with many investors today show that many groups of institutional investors and other larger asset owners are focused on their portfolios’ distance from broad equity indices, such as MSCI World, and are openly trying to close that gap by introducing a range of passive, semi-passive, factor or lower tracking error investment products. To me, this reflects not just late-cycle market dynamics but also practical, personal career considerations related to a pronounced concentration of the AI-led “Magnificent 7” cohort in global indices. While fundamentals in the AI space are strong and there are many good reasons to believe that this new technology is going to bring huge benefits to our society, the risks of excessive concentration of investment portfolios, reliance on tracking error measures, and delayed diversification are very real. These risks are clearly not new either as there is plenty of coverage of the benefits of diversification and dangers of concentration in the investment literature. This article aims to ground the discussion with evidence and industry research on the shortcomings of tracking error, the reasons for risk aversion in rising markets, the enduring value of diversification, and the dangers of waiting too long for a market reversal.
Flaws in tracking error as a risk metric
Many investors know that no measure is perfect when trying to assess a degree of active management in an investment portfolio. However, there is a clear demand for some sort of measure from all groups of investors, and when there is a demand, there will be a supply. Tracking error, defined as the standard deviation of portfolio returns relative to a benchmark, appears to be a clear first choice when it comes to comparing and analyzing active managers. Tracking error however was designed to evaluate passive funds’ ability to mimic their index—but its use as an active risk management metric can be fundamentally misleading. Tracking error fluctuates not only due to changes in portfolio active positions, but also with external factors such as market volatility, which is unpredictable and uncontrollable for portfolio managers.
It is also important to remember that the risk in tracking error is commonly defined as a) volatility and b) how differently a particular stock behaves relative to the index. The former, although often reflecting a genuine level of variability of the outcomes of the underlying business (i.e. when volatility is a good proxy for risk), can often simply indicate “past risks”. For example, a stock that had a few large falls and is now very cheap, or trading below its intrinsic value, will be viewed as highly risky by a risk model. Many investors however would logically view it as low risk from the perspective of losing invested capital. The latter is even less clear as it is not immediately obvious why difference should be viewed as bad. Diversification is best achieved by putting together a few investments that do not correlate to each other, so the different behaviour of prices should be viewed as a welcome attribute.
Using tracking error as a risk management metric can be fundamentally misleading.
However, we believe that there is even greater ‘risk’ in the use of tracking error-type measures during periods of market excess. When markets find themselves at the extreme end of the valuation pendulum, often the parts of the market that are most associated with that overvaluation represent a big part of the overall market. This big weight in that market also leads to those overvalued themes or groups of stocks dominating the behavior of the market in a way that they become the market. One can simply look at the correlation of the M7/AI group of stocks with the S&P, Nasdaq and MSCI World indices to see this very clearly. In this environment, any steps to reduce a portfolio’s tracking error to the index would simply result in the portfolio being more like that small group of stocks. In benign market circumstances this may or may not be desirable, but at the extreme end of valuation swings the same ‘risk reduction’ behavior exposes investors to material potential losses. To use an analogy, it is like trying to seek comfort and safety in a group of other ocean wave watchers on a beach as a huge wave hurtles towards you. Clearly the risk here is to be with the crowd. And the further you are from that group and away from the coastline, the better it is. In this context tracking error is just a measure of your proximity to the group, not the level of risk you take.
Why investors take less risk as markets rise
Bull market optimism can paradoxically prompt increased caution and lower risk-taking among investors—especially those approaching key career or portfolio milestones. The fear of underperforming a high-flying index, or missing out during a narrow rally led by a select group of stocks, reinforces herding behaviour. This dynamic is amplified by career risk: being different when AI or the M7 is the consensus trade can be seen as dangerous, especially if performance lags.
Almost by definition, the excesses of any market trend are the result of a prolonged period of euphoria. That often means that those who advocated the alternative view too early, or were true to their discipline and did not drift to this new theme, either lose their jobs, client support or both. Whoever replaces them adds more fuel to the fire and further supports the prevailing market trend.
Although some of this behaviour is not irrational, the point here is that it can exacerbate crowding and concentration in the riskiest parts of the market.
The persistent importance of diversification
Diversification remains one of investment’s most powerful, timeless tenets. Allocating capital across regions, sectors, and asset classes smooths returns, reduces event risk, and ensures portfolios are not overly reliant on the fortunes of a single theme or group. Foundational studies [1] demonstrate that portfolio returns are overwhelmingly determined by the asset allocation decision to a much greater degree than stock picking or timing [2]. Diversification offers protection against “unknown unknowns” and helps temper risk, especially during regime shifts, cyclical corrections, or sector-specific reversals.
Diversification remains one of investment’s most powerful, timeless tenets.
One potentially important change in this cycle when compared to the last significant one in the late 1990s is an unusually high level of similarity among the major equity indices. Back in 1999 the composition of S&P500, Nasdaq and MSCI World was rather different with only one stock – Intel – included in the top 10 of all three indices. Naturally, the overlap between the S&P500 and the MSCI world was higher at five stocks in the top 10, but the weight of those five US stocks in the world index was materially lower [3]. In other words, global investors had a decent degree of diversification between their US growth, US other and global equity managers despite a similar level of concentration in the Nasdaq and S&P indices to what we observe today. Today, however, all these three major indices look very similar with nine out of 10 top names matching [4]. So, to any global investor, a more passive or lower tracking error (but not ‘less risky”) portfolio might look well diversified, but it would in fact have allocated 30-60% (range of concentration of those three indices in top 10 names) to the same 9-10 names. This must raise questions. One must have a strong conviction to put so much of what is supposed to be a diversified portfolio into such a small group of companies. Just a quick look at the history of persistence of the companies remaining in the top 10 of various indices over time would tell us that this overreliance is not backed up by evidence.
Also, at this extreme level of index concentration, institutional investors probably need to treat their overall portfolios as a single concentrated portfolio where risks and future prospects really depend on a few names – something that normally is only considered at the single strategy or fund level, not the overall asset owner level. I would draw attention to a very interesting paper by Antti Petajisto[5] where he finds that, over a 96-year period to the end of 2022, of the US stocks that performed in the top 20% of their universe over a five-year period, 86% lagged the index over the subsequent 10-year period. This share rises to 93% in the post-World War II period. This effect is more pronounced in sectors that experience a more rapid change or a faster technology cycle (Tech, IT, Healthcare). And before we all take a bit of comfort and assume that most of this should be happening in smaller capitalization segments of the market – yes, the biggest effect is in fact for smaller firms – the effect is somewhat reduced for mid caps and then rises again for larger firms.
For many, diversification today means allocating away from equities all together (most dramatic action), away from US equities (to overseas developed and emerging equities) or at least away from US large caps (to overseas and potentially mid and smaller caps). Many today point to the fact that US exceptionalism resulted in a greater valuation of most of the US equity market, not just the larger tech/AI element. If one accepts this and seeks to best diversify a US equity portfolio, then the best solution would be to go overseas and seek unloved, undervalued areas that would complement the US portfolio most.
Outperformance of some sectors in Europe, and some emerging markets are the signs of the changing tide.
Falling opportunity costs
A new argument for investors – one that was arguably absent 12 months ago – is that the opportunity cost of moving away from US large cap growth/M7/AI stocks is lower than ever. Even for those who believe in their ongoing business case, valuations today are fully reflective of future growth prospects, and concerns around these high multiples are rising. For these stocks to continue to outperform, they must deliver above already high expectations – a tall order as 2025 has shown, with performance broadening to other areas.
This year, multiple regions and sectors have delivered impressive returns, and investors are increasingly seeking diversification. To those waiting for the right moment to de-risk or diversify away, this should be a sign that the others are already moving and acting. Mind you, given how much of the overall investor capital has been directed over the years towards M7/AI, I would expect that it would not take much (in terms of % of the investors’ existing US large cap holdings) to move to some of the unloved corners of the market.
Outperformance of some of the cyclical and value sectors in Europe, and of some emerging and frontier markets are all the signs of the changing tide. The irony is that these first price moves themselves are likely to create their own cycle of “fear of missing out” that might draw more and more capital to those markets. Once again the relative size of those allocations matters a lot. Even for the strongest believers in AI, it must be challenging to accept that the market cap of just one of those AI stocks – Nvidia – could be comparable with that of all countries included in MSCI EM index, to say nothing about Frontier countries (c.$4.3-5 trillion for Nvidia vs c. $10 trillion for the entire EM index – yes, this includes China!) [6]. In other words, a 10% divestment from this single company alone could bring 5% net new inflow into the whole EM index… One can point to the obvious fact that those countries (EM and Frontier put together) represent c. 85% of the global population [7]and 60% of GDP [8]. Moreover, if we assume that Nvidia would ultimately need customers for their chips, then the balance between one excellent company’s prospects and that of a huge part of its target market should not be where it is today.
The risk of waiting for the rally to end
Experience shows that sitting out a long rally, waiting for the ‘right moment’ to diversify or de-risk, is fraught with peril. Those with more experience in the market also know that market upswings can persist longer than rational analysis or fundamental maths can justify, with bull markets historically tending to be longer and stronger than bear markets[9]. Furthermore, research[10] suggests that the psychology of fear-of-missing-out is a powerful force for sustaining buying, even as valuations stretch. Signs of excess—such as extremely narrow leadership, valuation extremes, or overconcentration – can persist, so waiting too long for the reversal also risks significant missed upside. This to me is the prevailing theme of the dozens of year-end prediction papers I read and indicates a consensus view today. In other words, many people see and willingly talk about the high or even excessive valuation multiples of the US equity market and some of the larger technology stocks, but then find enough comfort and justification to stay invested. This implies either a superior ability to switch at the right time, a strong belief in the longer cycle (higher multiples) this time or both.
But eventually, corrections do occur, often abruptly and with little advance warning. History warns that overvalued stocks are susceptible to sharp pullbacks, which can erase years of gains in a matter of weeks or months. Another good study, this time from JPMorgan [11] indicates that the widely-held belief that it is only troubled companies that fall and destroy investor’s capital is very far from the truth. The change is often difficult to see in advance and comes from a wide range of sources, within the company or outside. Among other things, this study finds that the vast majority of companies that go through what they define a “catastrophic loss of value” – 70% drop in share price without a full recovery – had either “buy” or “strong buy” recommendations at the beginning of those deratings.
This is the moment to refocus on portfolio resilience, not just the narrow chase for returns.
Conclusion: The call for portfolio resilience
The trends driving concentration in AI, US large cap growth, and the M7 are not without merit, given their strong fundamentals. However, evidence from history and industry literature overwhelmingly supports the virtues of diversification and prudent risk management. We argue that a reliance on tracking error can distort portfolio construction, and the psychological pressures of career risk and market consensus are real – but cannot replace the disciplined benefits of owning a thoughtfully diversified portfolio. This is the moment to refocus on portfolio resilience, not just the narrow chase for returns. Diversification, humility about timing, and clarity of objectives are, in our view, the most robust defenses as markets move deeper into the late cycle and the risk of abrupt reversals grows.
Key Information
No investment strategy or risk management technique can guarantee returns or eliminate risks in any market environment. Past performance is not a guide to the future. The prices of investments and income from them may fall as well as rise and investors may not get back the full amount invested. Forecasts and estimates are based upon subjective assumptions about circumstances and events that may not yet have taken place and may never do so. The statements and opinions expressed in this article are those of the author as of the date of publication, and do not necessarily represent the view of Redwheel. This article does not constitute investment advice and the information shown is for illustrative purposes only.
[1] Brinson, Hood and Beebower ‘Determinants of Portfolio Performance’ (1986) and ‘Determinants of Portfolio Performance II: An Update’ (1991)
[2] Bloomberg, 26 November 2025
[3] Bloomberg
[4] Bloomberg, 26 November
[5] ‘Underperformance of Concentrated Portfolios’ (2023)
[6] Bloomberg, 26 November
[10] Bonaparte, ‘Catching the FoMO Fever: A Look at Fear in Finance’ (2021) and ‘Global FOMO: The Pulse of Financial Markets Worldwide’ (2025)
[11] The Agony and Ecstasy series – 2004, 2014, 2021, 2024
