An index can rise while most of its constituents fall. This is not a paradox or a data error — it is a direct consequence of capitalisation weighting, and it is the reason a headline index level periodically stops describing the market it is supposed to represent.
How the arithmetic works
In a capitalisation-weighted index, each company’s influence is proportional to its market value. A company worth 3 percent of the index contributes thirty times as much to the index move as one worth 0.1 percent.
When a handful of very large companies rise sharply while the majority drift lower, the index rises. The average stock fell. Both statements are true, and they describe genuinely different things.
The gap matters because most investors do not hold the index in index proportions. A portfolio of thirty stocks chosen on their merits behaves far more like the average constituent than like a capitalisation-weighted benchmark. In a narrow market, such a portfolio underperforms badly while its manager is doing nothing obviously wrong.
The standard measures
- Advance-decline line. A running cumulative total of advancing minus declining issues. It weights every stock equally, so it describes participation rather than value. A rising index with a falling advance-decline line is the textbook divergence.
- Percentage above a moving average. The share of constituents trading above their 200-day average. Straightforward to interpret: it is the proportion of the market in an uptrend, irrespective of size.
- Equal-weight versus cap-weight. The cleanest single measure. Compare an equal-weighted version of an index against the standard one. When the cap-weighted version pulls ahead, gains are concentrated in the largest names.
- New highs minus new lows. Useful at extremes. An index near a record while more constituents make new lows than new highs is an unusual and historically unhealthy configuration.
What narrow breadth does and does not tell you
The conventional reading is that narrowing breadth is a warning — a rally sustained by fewer and fewer stocks is fragile, because the withdrawal of a small number of leaders removes the whole advance.
There is something to this, and it is over-applied. Two qualifications matter.
The timing is unreliable. Breadth can narrow for a long time before anything happens. As a signal it has poor precision — it identifies conditions under which a decline is somewhat more likely, not when. Positioning defensively on narrowing breadth alone has historically meant being early by a wide margin, and being early is expensive.
Concentration can be justified. If a small number of companies genuinely are capturing a disproportionate share of economic profit, an index reflecting that is not distorted. It is doing its job. The question of whether concentration reflects fundamentals or enthusiasm cannot be settled by the breadth statistics themselves — they describe the pattern, not its cause.
The index concentration problem
There is a second-order effect worth understanding. As the largest companies grow, they occupy a larger share of the index — which means index funds hold more of them, which channels a larger share of new inflows into them.
This is sometimes described as a feedback loop that mechanically inflates large companies. The characterisation is overstated. Index funds buy in proportion to existing weights; they do not change relative prices simply by being large, because they buy everything proportionally. What they do reduce is the pool of active capital available to express a contrary view, which can allow divergences to persist longer than they otherwise would.
The practical consequence for an investor is more immediate: a broad index fund is considerably less diversified than the number of holdings suggests. An index of 500 companies whose top ten constitute a third of its value has roughly the concentration risk of a portfolio with far fewer positions. The count is reassuring; the weights are what matter.
Using it sensibly
- Treat breadth as context, not a trigger. It describes the character of a move. It does not date the reversal.
- Check the equal-weight comparison before concluding “the market” did anything. It takes seconds and it settles most arguments about whether a rally is broad.
- Know your own concentration. Look at the top-ten weight in any index fund you own. It is published, and it is frequently higher than people expect.
- Do not use it to justify market timing. The evidence that anyone times entries and exits reliably is weak, and narrow breadth does not improve those odds enough to overcome the costs.
That final caution is the important one. Breadth analysis is genuinely useful for understanding what has happened and for setting expectations about a portfolio’s likely behaviour relative to a benchmark. It is much weaker as a basis for acting, and the gap between those two uses is where most of the damage gets done — a distinction that applies to nearly every market indicator, as we argued in what an inverted yield curve actually signals.
