Divergence
Divergence happens when a price chart and a technical indicator tell two different stories at the same time, a mismatch that experienced traders treat as an early warning sign. The clearest way to picture it: a stock's price climbs to a new high, but the RSI, an indicator measuring the strength of recent gains versus losses, scored 0 to 100, makes a lower high instead of matching the new price high. That combination is called bearish divergence, because it suggests the rally is running on weaker underlying momentum even though price is still technically climbing, often a sign that a top is close.
Bullish divergence is the opposite setup: price makes a new low, but the indicator makes a higher low, hinting that the selling pressure behind the decline is fading even as price keeps dropping. Nvidia's stock showed a version of this pattern in late 2018, where price kept sliding into December while momentum indicators had already started curling upward, a signal that preceded the eventual bottom. The practical use isn't to act on divergence alone, since it can persist for weeks before price actually turns, but to treat it as a reason to watch for confirmation, like a trendline break or a reversal candle, rather than assuming the current move will simply continue unchanged.
Divergence gets checked the same way across every liquid market, gold, major currency pairs, individual cryptocurrencies, because it's really a statement about the relationship between price and momentum, not about any particular asset. What changes is patience: divergence on a monthly chart can take months to resolve, while the same signal on a five-minute crypto chart might play out within the hour.
The most common mistake with divergence is not acting too early but seeing it everywhere. Any oscillator will disagree with price regularly, because an oscillator is bounded and price is not - a market in a sustained trend must eventually produce lower indicator highs against higher price highs, since the indicator has nowhere left to go. That is a mathematical property of the tool rather than evidence about the market. This is why divergences appear in clusters during strong trends, which is exactly the condition in which acting on them is most expensive.
Three habits separate the usable signal from the noise. Require the extremes being compared to be genuine swing points with something between them, rather than any two bars that happen to suit the argument. Require the divergence to appear at a level that already mattered - a previous high, a round number, the edge of a range - so that two independent reasons coincide. And require price to do something before acting: a failed attempt at a new extreme, a break of the short-term structure. Divergence is best understood as a reason to prepare for a signal rather than as the signal itself, which is also why it is so often described as early rather than wrong.