Market divergence is a condition in which two market series or measures being compared for confirmation stop moving in a mutually consistent way, or fail to make corresponding highs, lows or directional moves. The comparison may involve price and momentum, an index and market breadth, two related markets, or another pair of variables linked by an analytical relationship. Divergence is therefore an observation of non-confirmation. It does not by itself establish causation, a coming reversal or a trade direction.
Divergence depends on what is being compared
There is no single universal market-divergence formula. A researcher must specify the two series, their transformation, the observation horizon and the rule used to identify the disagreement. A stock index making a higher high while an advance-decline measure makes a lower high is a breadth divergence; price reaching a new high while a momentum oscillator fails to do so is an indicator divergence; two markets that had been moving together can also diverge when one advances while the other stalls or declines. These observations share the idea of non-confirmation, but they are not economically interchangeable because each comparator measures something different.
The construction can materially change the result. A divergence may appear in price levels but disappear when returns are compared, appear on daily data but not weekly data, or look significant until an established lead-lag relationship is taken into account. Visual judgments can also depend on how swing highs and lows are defined. Reproducible research therefore requires an explicit divergence rule rather than a chart label alone.
What divergence can—and cannot—tell you
Divergence is useful because it asks whether a market move is being corroborated by another dimension of the evidence. Weakening breadth during an index advance may indicate narrower participation; weakening momentum during a new price high may indicate that the rate of advance has slowed; cross-market divergence may indicate that a previously useful relationship is changing. Each is an interpretation that still requires context.
The divergence can resolve through reversal of the first series, catch-up by the second, consolidation, ordinary noise or a more durable change in the relationship. It is also distinct from negative correlation: two series can remain positively correlated over a broader sample while temporarily failing to confirm one another around a particular move. The practical question is therefore not simply whether a divergence exists, but what stopped confirming what, over which horizon, and whether the assumed relationship still has a defensible economic or statistical basis.