The Market Beneath the Market

Something very strange is happening in the stock market, but the most interesting part is probably being missed. The S&P 500 can be close to an all-time high while a surprisingly large part of the market behaves as if investors are living through a completely different environment. Goldman Sachs recently highlighted that almost half of S&P 500 stocks have moved into negative-beta territory, meaning they have increasingly tended to move in the opposite direction from the broader index, a pattern that has not been seen at this scale since the Tech Bubble. That statistic deserves much more attention because it tells us something deeper than the usual story about a few large companies carrying the index.

The first thing to understand is that the S&P 500 is no longer remotely close to being an equal representation of 500 companies. As of September 17, 2026, the ten largest positions accounted for roughly 40% of the SPY portfolio, while information technology alone represented more than 38% of the index. The index is deliberately constructed as a float-adjusted market-capitalization-weighted benchmark, so the companies that have become largest automatically acquire more influence over the number everyone watches. This creates an important distortion in how investors interpret the market because a record high in the index increasingly tells us about the behaviour of the largest companies rather than about the average company.

That distinction sounds academic until the market begins to diverge. Imagine two markets containing exactly the same 500 companies, with one version weighted equally and the other weighted according to market capitalization and then imagine that the largest companies rise while hundreds of smaller constituents stagnate or fall. The capitalization-weighted version can continue setting records while the equal-weighted version tells a much less comfortable story. Suddenly “the market is at an all-time high” becomes an incomplete statement because the market being measured is itself heavily tilted toward the companies that have already become enormous.

There is another phenomenon that is considerably more unusual. Goldman’s data show that the largest technology companies now have substantially higher market betas than the remaining constituents, while a growing proportion of other stocks have negative beta relative to the index. In simple terms, the market is increasingly splitting into two different populations: companies that amplify movements in the dominant index factor and companies that can behave as if they belong to a different risk regime altogether. When that happens, traditional assumptions about diversification become less reliable because owning 20 different stocks does not necessarily mean owning 20 independent sources of risk.

This is one of the least appreciated consequences of concentration. Investors often think diversification means owning more tickers, but diversification depends on what actually drives those securities. If hundreds of companies are increasingly being priced against one dominant factor while another group is moving in the opposite direction, the number of positions tells you surprisingly little about the number of independent bets being made. A portfolio can contain dozens of stocks and still be one enormous trade expressed through different securities.

Modern equity markets are not simply places where one person sells a share to another person on an exchange. Orders can be routed between exchanges, electronic communication systems, wholesalers, internalizers, dark pools, ETFs, futures and other trading venues, while sophisticated firms continuously arbitrage small price differences between related instruments. The SEC has described internalization as a process in which a broker or affiliated dealer can execute customer orders against its own inventory, while payment for order flow creates an economic incentive for brokers to route orders to particular market makers.

This is where the subject becomes much more interesting than the usual “algorithms are manipulating the market” argument. An algorithm does not need to receive an instruction saying “keep the index stable.” It can simply be following a portfolio objective, hedging an option position, maintaining a target beta, managing inventory, arbitraging an ETF against its underlying stocks or reducing exposure to one factor while increasing exposure to another. Thousands of independent systems can therefore respond to the same market movement in similar ways, producing an outcome that looks coordinated even when there is no central coordination.

That phenomenon could be called synthetic coordination. It is one of the most important concepts investors should understand. Markets do not require a conspiracy to produce herd-like behaviour when the participants are using similar benchmarks, similar risk models, similar derivatives, similar volatility targets and similar hedging mechanisms. The machines do not have to agree about the future. They only have to react to the same inputs under similar constraints.

Index arbitrage makes this even more interesting. Futures, ETFs and baskets of individual stocks are linked by traders attempting to exploit temporary differences in their relative prices, meaning that a transaction in one part of the market can create hedging or arbitrage activity somewhere else. CME describes this type of activity as the simultaneous trading of index products, ETFs and baskets of component stocks to capture temporary pricing differences.

This means the visible price movement in an individual stock can sometimes be the final expression of a much larger trade that started somewhere else:

• a stock can rise because somebody loves that company,
• it can also rise because somebody is hedging an index position,
• or because an ETF is being arbitraged,
• or because an options dealer needs to adjust its hedge,
• or because a portfolio manager is rotating factor exposure,
• or because a futures position has created a corresponding basket trade.

The chart shows the same green candle in every case. The reason behind it can be completely different.

This is precisely why looking at individual price movements without understanding the plumbing underneath them can be dangerous. Investors often assign a fundamental explanation to a move that may have been generated partly by positioning, hedging or mechanical flows. The longer these mechanisms operate, the easier it becomes to confuse a price signal with an information signal.

And then there is the part of market structure that receives surprisingly little attention outside specialist circles: internalization changes where price discovery happens. The SEC has repeatedly raised concerns that payment for order flow and internalization can fragment trading interest because some orders that might otherwise interact on public exchanges are executed elsewhere. The Commission has specifically discussed the possibility that information contained in those orders does not fully enter the public price-discovery process, while also acknowledging that internalization can provide retail investors with price improvement.

That creates a fascinating paradox, where the market can become more efficient for the execution of an individual retail order while becoming less transparent about where information is being incorporated into the broader market. Those two things can happen simultaneously.

This distinction is rarely discussed when people talk about “efficient markets.” Execution efficiency and information efficiency are related, but they are not identical. A retail investor receiving a fractionally better execution price does not necessarily mean that the overall market is doing a better job of discovering the fundamental value of the company.

The SEC has even documented historical cases where internalization algorithms failed to deliver the best available pricing. In 2017, the Commission charged Citadel Securities and announced a $22.6 million settlement over misleading statements concerning the pricing of retail orders, finding that two algorithms did not internalize orders at the best price observed or seek that price in the marketplace. The existence of such enforcement actions does not prove that the market is systematically manipulated, but it does demonstrate that the machinery through which millions of trades are executed deserves considerably more attention than it receives from ordinary investors.

There is another overlooked consequence. When liquidity is fragmented across venues and instruments, the price investors see on their screen is the result of an enormous amount of interaction between displayed liquidity and liquidity that is not displayed in the same place. The public order book therefore represents only part of the mechanism producing the final price. The market can still be highly competitive, but the relationship between visible orders and actual price formation becomes much more complicated than the old mental model of buyers and sellers meeting on an exchange.

This matters particularly during stress. The SEC has previously noted that during the May 2010 flash crash, the withdrawal of liquidity by OTC market makers that normally purchased and internalized large amounts of retail flow contributed to instability when those firms reversed their normal behaviour and large amounts of selling reached exchanges. The lesson is broader than the flash crash itself: liquidity that appears permanent during normal conditions can behave very differently when market makers, hedgers and systematic strategies suddenly face the same risk at the same time.

That is where the current divergence becomes interesting:

• a market can look extremely stable while the underlying sources of stability are becoming increasingly concentrated,
• a few enormous companies dominate the index,
• a large ecosystem of passive products references the index.

Futures and options allow institutions to express broad exposures without trading every underlying stock. Market makers hedge those exposures. Arbitrageurs connect ETFs, futures and individual securities. Retail orders can be internalized away from public exchanges. Systematic strategies respond to volatility, correlations, momentum and risk limits. None of these mechanisms is inherently sinister. Together, however, they create a market that behaves very differently from the simple world imagined by someone looking only at a price chart.

And that may explain one of the strangest features of the current environment: the index can become increasingly stable and increasingly misleading at exactly the same time. When a handful of companies dominate the benchmark, the index becomes less sensitive to weakness in the median constituent. When derivatives allow investors to hedge broad exposures efficiently, individual stocks can move for reasons that have little to do with their own fundamentals. When algorithms respond to common signals, securities can become correlated when the models say they should be correlated and negatively correlated when portfolio construction demands the opposite. When internalization removes part of the order flow from public exchanges, the displayed market becomes only one part of the price-discovery process.

The result is a market that can look calm from 30,000 feet while becoming increasingly complicated underneath. This is also why breadth deserves more attention than the headline index. If the S&P 500 rises, the next question should be how many constituents participated, whether the equal-weight index confirmed the move, whether correlations are rising or falling, whether dispersion is expanding and which sectors are actually responsible for the advance. The more concentrated the benchmark becomes, the less informative the headline percentage change becomes as a description of the entire equity market.

This is the part most investors miss. The question is not simply whether the S&P 500 is going up. The question is what mechanism is making it go up?

If earnings are improving broadly and participation is expanding, an index advance tells one story. If the index is rising because a narrow group of companies dominates its weighting while hundreds of constituents behave differently, the same one-percent gain tells a very different story. If derivatives, systematic positioning and cross-market arbitrage are simultaneously influencing the underlying securities, the interpretation becomes different again.

The number on the screen is identical, the market underneath it is not. So perhaps the most useful thing investors can do right now is stop asking whether the market is being manipulated and start asking a more uncomfortable question: how much of what looks like fundamental price discovery is actually the consequence of market structure, positioning and mechanical flows? That question requires looking underneath the index and that is precisely where the interesting information may be.