Divergence Between Price And Indicators

Divergence has no clock

This is the expensive part. A divergence describes a change in the character of a move. It says nothing about when, or whether, price will respond.

Markets routinely produce three or four consecutive divergences during a strong trend, each one looking exactly like the last, while price keeps going. A trader who treats the first as a reversal signal takes a position against a trend, gets stopped, sees the second, and does it again. Four small losses in the same direction during a move that never turned is a common way to spend a month.

The correct framing is that divergence changes what you do with an existing position, not what you do about a new one. Stop adding. Tighten management. Require more from the chart before taking the next entry in that direction. Those are all reasonable responses to weakening momentum. Reversing is not.

There is one more distinction worth holding. The gap between price and indicator widens gradually, so a divergence that has been building across five bars is a different observation from one that appeared on the last two. The first describes a sustained loss of force. The second may simply be a fast bar that the indicator has not caught up with.

The right-hand swing does not exist yet

• You are measuring from a swing that is not confirmed. A high only becomes a high once price has moved away from it, which is several bars later.
• The divergence looks obvious afterwards and ambiguous in real time, because in real time you do not know whether the current bar is the second peak or the middle of a longer advance.
• Different indicators disagree. The same two swings can produce a divergence on RSI and none on MACD, since the two measure different things.
• Changing the period changes the answer. An RSI(7) divergence and an RSI(21) divergence over the same swings are not the same observation.

Making it tradeable with a level and a stop

Divergence becomes workable when it stops being the reason for the trade and becomes a condition on a trade you would have considered anyway. The structure is simple enough to write down.

First, require a level. A divergence at a price that means nothing is worth less than the same divergence at a prior swing high, a session pivot, or the edge of a range. Second, require a price event. A rejection candle, a failed break, or a close back inside a level gives you an invalidation point that divergence alone never provides. Third, size the position from that invalidation rather than from how convincing the divergence looked, which is what the position size calculator is for.

That sequence also fixes the timing problem. You are no longer trying to call the top from a momentum reading. You are waiting for price to agree with the reading, and accepting that a share of these will fail regardless, which is the arithmetic the risk management guide is built on. The same discipline applies to any early-warning reading, which is why reading a signal properly always starts with where the idea is wrong.