The Gravitational Pull of the Index
Why We Refuse to Let a Benchmark Dictate Decisions
By the team at Banyantree Investment Group
Philosopher Giving a Lecture on the Orrery, c.1768. Joseph Wright of Derby. Art UK (Yale Center for British Art). Cropped and colour-treated.
The NASDAQ peaked on 10 March 2000. Twenty days later, Julian Robertson closed Tiger Management.
At its peak, Tiger had been one of the most successful hedge funds in the world: twenty-two billion dollars in assets, two decades of disciplined, value-oriented returns. By early 2000, Robertson was haemorrhaging capital. His sin was not bad stock-picking. His sin was refusing to own Cisco at 150 times earnings, refusing to own Yahoo at prices that valued eyeballs over cash flows, refusing to buy companies whose valuations required believing six impossible things before breakfast. While the NASDAQ doubled and doubled again, Robertson held what he understood at prices he could defend.
He looked, in the language of the industry, ‘wrong’.
Within two and a half years, the NASDAQ would fall 78 per cent. Cisco would lose 88 per cent of its value and take twenty-five years to recover its peak price. The fund managers who had owned it at 150 times earnings would have explanations. Robertson would have been right. But by then he was out of business, his investors’ patience exhausted twenty days before the market proved him correct.
What would you own if there were no benchmark?
───
In physics, gravitational pull is proportional to mass. The larger the object, the harder it is to escape its orbit. Index weights work the same way, and the place you feel their pull most clearly is not in a performance table. It is in a meeting room.
The scene repeats itself with modest variation across fund types and decades: a global equity strategy, underweight the largest and most expensive benchmark names on valuation grounds, three quarters into a period of relative underperformance. The committee pack lands on the table. Someone flips to the benchmark comparison page and asks, politely, the question that echoes across the industry: ‘Why don’t we look more like this?’
Another member notes that tracking error is ‘elevated’ — the word chosen carefully, as though deviation from the index were a fever requiring diagnosis. The portfolio manager explains the rationale: valuation concerns in the largest benchmark constituents, concentration risk, liquidity under stress. The explanation is correct. It is also, in that room, inadequate. No one demands the manager hug the index. They simply ask why the portfolio looks different, and keep asking until it doesn’t.
This is how gravitational pull operates. As a stock rises, its benchmark weight increases, and the energy required to stay underweight grows with it. A 2 per cent position that triples becomes a 6 per cent position in the index. Underweighting it now costs three times as much career risk as it did before. The pull compounds. It is not a one-time force to be overcome; it is a continuous field, growing stronger the longer the divergence persists and the larger the benchmark constituent becomes. Escape requires not just conviction but a governance structure that explicitly permits divergence — established before the divergence becomes uncomfortable.
Without both, the benchmark wins by attrition.
───
Benchmark-hugging is often the rational choice for the person making it.
Research from State Street’s Center for Applied Research, surveying two hundred institutional investors across endowments, foundations, pensions, and sovereign wealth funds, found that the majority cited career risk as the single largest determinant in their decision-making. Not market risk. Not liquidity risk. Career risk — the fear of being fired, passed over, or blamed for outcomes that look conspicuously different from peers.
The asymmetry is brutal. Bold and wrong ends careers. Timid and mediocre extends them.
A manager who underperforms the benchmark by four percentage points while everyone else underperforms by three faces uncomfortable questions. A manager who underperforms by three alongside everyone else faces sympathy. The mathematics of career survival therefore favour conformity, even when conformity means owning assets at prices that make no sense. This is why, in early 2000, so many professional investors owned Cisco at 150 times earnings. They knew the price was detached from any reasonable expectation of future cash flows. They owned it anyway, because being underweight a 4 per cent benchmark position that keeps rising is a career problem. Being wrong together is not.
In gravitational terms, career risk is the force that prevents escape velocity. An object needs to move fast enough to break free of a body’s pull. A portfolio manager needs enough institutional support to withstand the discomfort of looking different. Most don’t have it. Most aren’t designed to. The governance structures that hire active managers are often the same structures that punish active behaviour, and the contradiction resolves itself, inevitably, in favour of the larger body. The portfolio drifts back toward the index. The manager, relieved, calls it ‘risk management’.
───
‘Benchmark-unaware’ does not mean ignoring benchmarks. It means refusing to let an index dictate decisions.
In a benchmark-aware world, risk means tracking error: how much the portfolio deviates from the index. Success means not looking different. The logical response is predictable: portfolios tilt toward large benchmark weights regardless of valuation, avoid off-benchmark opportunities regardless of merit, and converge toward crowded positions because standing apart carries career risk. The benchmark becomes not a reference point but a ceiling and a floor — the boundary of a space from which the portfolio is never permitted to stray far enough to justify its fees.
Benchmark-unaware portfolios start elsewhere. The mandate specifies what is binding: objectives, risk tolerance, liquidity requirements, concentration limits, time horizon. Those are the guardrails. The benchmark shows what the market owns at a point in time — useful for context, for diagnosis, for explaining relative performance. Not a portfolio instruction. The distinction is between a map and a destination. We use the benchmark as a map of where the market has been. We refuse to treat it as a map of where the portfolio should go.
A related term — ‘benchmark-agnostic’ — sounds similar but means something different, and the practical consequences matter. Benchmark-agnostic implies indifference: to reporting, to governance, to the reasonable expectation that a manager can explain why results diverged. A benchmark-agnostic manager might say ‘we don’t report against any index,’ which makes oversight nearly impossible. Against what would a committee measure outcomes? A portfolio that cannot be evaluated cannot be trusted, regardless of the quality of the thinking behind it. Benchmark-unaware keeps the benchmark visible. It just stops being the steering wheel.
───
Here is where the gravitational pull becomes tangible.
A dominant enterprise software company represents 7 per cent of the benchmark. It trades at 32 times earnings, a premium to its own history though not absurd given the growth profile. Recurring revenue, expanding margins, a competitive position that rivals have failed to dislodge. The stock has tripled in four years. The bull case writes itself.
A benchmark-aware manager looks at this and asks: what is my tracking error if I underweight by 3 per cent? If the stock rises another 20 per cent, that underweight costs 60 basis points of relative performance. Do that with three similar names and you have explained most of a difficult year. The gravitational response is predictable. Own 5 to 6 per cent. Stay close. Manage the career risk.
A benchmark-unaware manager asks different questions. At 32 times earnings, the market is assuming five more years of 15 per cent growth. Is that realistic? What happens in a stress scenario — multiple compression to 20 times, a 10 per cent earnings cut? That implies a 44 per cent drawdown. At benchmark weight, this single position could cost the portfolio more than 3 per cent in a bad year. Does that fit the loss budget? And here is the question the benchmark-aware framework never asks: what else could you own with that capital?
Exhibit 1: Same Stock, Two Frameworks (Hypothetical)
Same stock, two frameworks. Hypothetical example for illustrative purposes only.
The trade-off is explicit. If the stock rises another 30 per cent, the benchmark-unaware portfolio trails. The committee meeting will be uncomfortable. But the question in that meeting should be: was the decision consistent with mandate, valuation, and loss budget? That is a conversation about process. Not a conversation about why the portfolio failed to orbit the same centre as everyone else.
───
The freedom to look different is not free. Robertson’s story is the honest version of what it costs, and it would be dishonest to put that story at the beginning of this essay and then avoid its implications. His fund faced redemptions rather than tracking-error scrutiny — the mechanism was different from a long-only mandate governed by a committee — but the outcome was identical: capital withdrawn before the thesis was proved, at precisely the moment when patience had the highest expected value.
Relative returns can diverge from the benchmark for years — not months, not quarters, but the kind of multi-year stretches that outlast the patience of most oversight structures. Here is what that feels like from the inside. Quarter one: the portfolio trails. The commentary is measured. ‘Positioning is intentional.’ ‘Valuation discipline will be rewarded over the cycle.’ The committee nods. Quarter two: the gap widens. The tone shifts. ‘Can you walk us through the underweight in [largest benchmark name]?’ The question is polite. The implication is not. Quarter three: the email arrives. Not hostile, not even critical — just a request for ‘an updated note on how the current positioning relates to our peer group.’ Peer group.
By the fourth quarter, the conversation is no longer about the positions. It is about the manager. The committee is no longer asking ‘why don’t we look more like the index?’ They are asking, in the silence between polite questions, whether they made a mistake hiring someone who doesn’t. This is the specific gravity of the benchmark: it doesn’t need to be right. It just needs to be what everyone else owns, because the committee’s own career risk is entangled with the manager’s. Firing a manager who looks like the index is hard to justify. Keeping a manager who doesn’t is harder.
This is why governance must be stronger when benchmark constraints are looser, not weaker. If the benchmark is not constraining the portfolio, the mandate and process must: documented reasoning, pre-committed signposts, explicit limits on divergence that are agreed before the divergence becomes uncomfortable. A benchmark-unaware label that becomes licence to drift is worse than useless. And if a committee cannot tolerate meaningful divergence over a realistic horizon, it should not hire a benchmark-unaware manager. That is not a judgement. It is clarity about the trade-off.
───
If gravitational pull is the force, closet indexing is what happens when the force wins completely and nobody admits it.
In 2015, the Norwegian Financial Supervisory Authority examined DNB Norge, the country’s largest fund manager. DNB had the lowest Active Share of any known collective investment fund in Europe — holdings so closely resembling the benchmark that it had virtually no chance of meaningfully outperforming. Yet it charged active management fees. The regulator concluded that DNB had failed to deliver the service its customers paid for. Norway’s Consumer Council filed a class action on behalf of 180,000 investors, seeking 690 million kroner in compensation.
DNB won at first instance. Oslo City Court ruled in January 2018 that the legal bar for compensation had not been met — even though the regulator had already concluded the fund breached its conduct obligations. Investors appealed. The Court of Appeal reversed that finding in May 2019. Norway’s Supreme Court upheld the reversal in February 2020, ordering DNB to repay approximately 350 million kroner to the 180,000 investors. Think about the full sequence. A regulator found that the product did not do what it claimed. A first court agreed the description was inaccurate but found no grounds for compensation. Two higher courts overturned that, and five years after the original investigation, clients received a partial refund of the fees they had overpaid. The quiet bargain was eventually broken. But it held, across three courts and half a decade, before it did.
Research from ESMA, the European securities regulator, found that between 5 and 15 per cent of actively managed equity funds could be closet indexers. In the UK, the Financial Conduct Authority forced managers to pay £34 million in compensation across 64 funds. The pattern is consistent across jurisdictions and decades: portfolios that charge for escape velocity and never leave orbit.
The test for whether a manager is genuinely active is not whether the portfolio looks different in a single quarter. It is whether the manager can sustain difference when it is uncomfortable — when the benchmark is pulling hardest, when the committee is asking the most questions, when the career-risk calculus most strongly favours conformity. That is the moment that separates a portfolio with genuine active risk from one that is merely in a slightly eccentric orbit around the same centre as everyone else.
───
What would change this view?
Two live debates would genuinely require the framework to be revisited. The first concerns passive's growing market share. The standard critique of benchmark weighting is that it is backward-looking — it reflects what has risen, not what is cheap. But if the continued growth of passive investing makes large benchmark weights self-reinforcing through mechanical inflows, then underweighting the largest index constituents may carry more structural risk than the essay allows. That is not a settled empirical question, and a CIO who has watched mega-cap concentration compound through passive flows has reasonable grounds to press on it. The second concerns Active Share itself — the metric on which the DNB Norge case and much of the closet-indexing literature depends. Cremers and Petajisto’s original research linked high Active Share to outperformance, but subsequent work has found the relationship is sensitive to time period, fee level, and how opportunity set is defined. If Active Share does not reliably predict what it is claimed to predict, the case for benchmark-unaware mandates becomes harder to defend on quantitative grounds, even if the qualitative logic remains intact. Neither argument dismantles the framework. Both deserve honest engagement.
───
What would you own if there were no benchmark?
The question is worth asking not because benchmarks are useless but because they have become gravitational. They pull portfolios toward them through the accumulated weight of questions asked in meeting rooms, career incentives that punish deviation, a measurement culture that treats ‘different’ as synonymous with ‘risky.’ The pull compounds. Escape requires energy most governance structures are not built to supply.
A benchmark tells you what the market owns. It does not tell you what the market is paying, or what happens when crowded positions unwind, or whether liquidity will exist when you need it.
A portfolio that must look like the index is not an active portfolio. It is an index fund with a higher fee. A portfolio that can look different — with the governance to sustain difference and the discipline to explain it in terms of mandate, valuation, and downside — that is what active fees are for.
The index knows what everyone owns. It does not know what anything is worth.
General information only. Not personal advice. Past performance is not indicative of future performance. Examples are illustrative and hypothetical. This material is intended for wholesale and professional investors.
