Divergence has an appealing logic. Price makes a new high, the oscillator does not, and you conclude that momentum is fading before price shows it. Sometimes that is exactly what happens. Often it is not, and the trader who shorts every divergence in a bull market learns that lesson expensively.
The concept is worth keeping. What needs work is the filtering.
What divergence measures
Momentum oscillators such as RSI and MACD measure the rate of change in price, not price itself. Divergence occurs when those two disagree.
If price grinds to a marginal new high while the rate of that advance slows, RSI will print a lower high even though price printed a higher one. The oscillator is telling you the move required more effort for less result.
That is genuinely useful information. What it is not is a reversal signal. Momentum can fade for a long time while price continues higher, and a market losing momentum can simply consolidate rather than reverse.
Regular divergence
Regular divergence points towards a possible reversal.
Bullish regular divergence: price makes a lower low, the oscillator makes a higher low. Selling pressure is weakening even as price makes new lows.
Bearish regular divergence: price makes a higher high, the oscillator makes a lower high. Buying pressure is weakening into new highs.
This is the version everyone learns first, and it is also the version with the worst standalone results, because it asks you to trade against the prevailing direction.
Hidden divergence
Hidden divergence points towards trend continuation, and it is underused.
Bullish hidden divergence: price makes a higher low, the oscillator makes a lower low. The pullback shook out momentum but price held structure. In an uptrend this is a continuation signal.
Bearish hidden divergence: price makes a lower high, the oscillator makes a higher high. In a downtrend, a continuation signal.
Because hidden divergence trades with the trend rather than against it, its risk profile is better for most traders. If you are going to use one type, this is the one to start with.
Why divergence trades fail
Trend strength
In a powerful trend, divergence appears repeatedly and means nothing. A parabolic advance will print bearish divergence on every timeframe for weeks while price doubles. Every one of those is a losing short.
The fix is a trend filter. Do not take regular divergence against a trend that is still structurally intact. Wait for price itself to break structure first, then use the divergence as supporting evidence rather than as the trigger.
Cherry-picking the pivots
Divergence requires two comparable swing points. Which swings you choose determines whether a divergence exists at all, and it is very easy to select the pair that supports the conclusion you already hold.
Fix this with a rule. Use only pivots that are clearly defined by a minimum number of bars on each side, and compare consecutive pivots rather than any two you like the look of.
Trading it in isolation
Divergence tells you about momentum. It says nothing about location. A bearish divergence in the middle of a range is close to worthless. The same divergence at a well-tested resistance level, into a prior supply area, with a rejection candle, is a different proposition.
Filters that actually improve results
- Require a structure break. For a regular bearish divergence, wait for price to break the most recent higher low. Now the divergence is confirming a change that has already begun rather than predicting one.
- Require location. The divergence should occur at a level that matters: a prior high or low, a higher-timeframe zone, a measured move target.
- Prefer higher timeframes. Divergence on the 4-hour and daily charts carries far more weight than on the 5-minute chart, where oscillators whipsaw constantly.
- Watch for the third push. Divergence that develops across three successive pushes is generally more reliable than across two, because exhaustion has had time to build.
- Check the oscillator's absolute level. A bearish divergence forming with RSI above 70 says something different from one forming at 55.
RSI versus MACD for divergence
RSI is bounded between 0 and 100, which makes its highs and lows directly comparable across time. Divergences are easy to spot and the overbought and oversold context is built in. The downside is that RSI can sit pinned at an extreme during strong trends, producing divergence after divergence.
MACD is unbounded and built from moving averages, so it responds more slowly. Divergences take longer to form, which means fewer of them and typically better quality. The MACD histogram is particularly useful, since a shrinking histogram into a new price extreme is a clean visual for fading momentum.
Neither is better in general. RSI gives you more signals with more noise, MACD gives you fewer signals with more lag. If you are being shaken out repeatedly, MACD may suit you better.
Building the trade
Take a bearish regular divergence on the 4-hour chart as an example.
Price makes a higher high into a prior resistance area. RSI prints a lower high. So far this is analysis, not a trade.
You wait. Price rolls over and breaks the most recent swing low, confirming a structural change. That break is your trigger. You enter on the retest of the broken level, stop above the divergence high, and target the previous swing low or the nearest demand area.
Notice the sequence: divergence first as a reason to watch, structure break as the trigger, retest as the entry. Traders who enter at the divergence itself are guessing at the top. Traders who wait for all three are trading a confirmed change with a defined invalidation.
The uncomfortable truth
Divergence on its own does not have a reliable edge. Published tests of naive divergence rules generally show results around or below breakeven once costs are included.
That is not an argument against learning it. It is an argument for treating it as one input among several. Divergence is excellent at telling you when a move is running on fumes. It is poor at telling you when the move will stop. Those are different questions, and confusing them is what costs money.