What Is RSI Divergence?
Relative strength index (RSI) divergence is when price makes a higher high while RSI makes a lower high, or price makes a lower low while RSI makes a higher low. It is read as a warning that the move is losing conviction, though the reading measures the fourteen-bar balance rather than the size of the latest push.
Covered on this page: TradingView.
The most-taught reversal signal in technical analysis, and the one where measuring what it actually does changes the story most.
How it works
Compare two highs in price with the two RSI readings underneath them.
In the worked example price went from 100.80 to 101.41 — a higher high — while relative strength index (RSI) went from 60 to 55. Price up, indicator down. That disagreement is the divergence.
Bearish divergence compares two highs. Bullish divergence compares two lows. Comparing a high with a low is not a divergence, it is two unrelated points.
What it does not mean
The usual explanation is that the second push is the weaker one — the move running out of strength. That is what the example above describes.
Measured, the second push covered 1.10 and the first covered 1.01. The second advance was larger, and RSI still fell.
The explanation is in the indicator, not the market. RSI averages the up moves against the down moves over fourteen bars — which includes the pullback between the two highs. A deep pullback loads the down side of that ratio, so the reading at the second high can be lower even though the push into it was stronger.
So “divergence means the move is weakening” is a story about price that is actually a fact about a fourteen-bar average. Sometimes the two coincide. Here they did not.
How often they happen
8 of 54 bars on an ordinary price series, using a fixed rule: a higher high than any of the previous four to twelve bars, on an RSI reading at least three points lower.
That is one every seven bars. They are not rare events, and a signal that appears every seven bars needs a very high hit rate to be worth acting on.
Hidden divergence
Regular divergence: price higher high, indicator lower high — read as a reversal.
Hidden divergence: price lower high, indicator higher high — read as continuation.
Notice what that does. Between the two definitions, almost any disagreement between price and RSI has a name and a directional interpretation, and the two interpretations are opposite.
A framework that can explain both outcomes explains neither, and this is the clearest example of that on the site. It does not make hidden divergence wrong; it means the pair cannot both be evidence, and you have to decide which you are claiming before the fact.
Which two points, and who chose them
The part that decides everything, and the part no definition pins down.
A divergence needs two highs. Nothing tells you which two. Take a slightly different pair from the same chart and the comparison flips — the same freedom the trend lines page identifies as its central weakness, arriving here through a different door.
The count above used a fixed rule — a higher high than any of the previous four to twelve bars, on a reading at least three points lower — precisely so that a person could not choose. 8 in 54 bars is what the rule found, not what a person looking for a trade would have found, and a person would have found more.
So the honest procedure is to fix the rule before you look. Which highs count, how much lower the reading has to be, how far apart the two may sit. Write it down, then apply it.
If you cannot state it, you are not finding divergences — you are noticing the ones that suit you, and there are enough of them on any chart to supply whatever you already believed.
A worked example
Only look for it if you already have a read. Divergence is a reason to doubt a position, not a reason to open one.
Compare like with like — two highs, or two lows, both of them turns you would have marked as swing points anyway.
Treat it as a reason to tighten, not to reverse. Moving a stop up is proportionate to the strength of the evidence; going short into a trend is not.
And take the invalidation from price — above the second high, the divergence is finished.
The original data
Across our study of 24,971 trading videos, 40 have RSI divergence in the title. The median one gets 10,295 views, 70% never pass 50,000, and the median length is 9.6 minutes.
That is a strong median on a small field — better than RSI itself at 5,377, which is worth noticing: the sub-topic outperforms the topic it belongs to.
The corpus carries description text for 29 of those 40, and across those 29, 2 mention invalidation, failure, or what a bad read looks like.
When it fails
In a trend it fires repeatedly and is wrong every time
Seventeen on a single advance.
A strong trend produces divergence constantly, because RSI cannot keep making new highs indefinitely — it is bounded at 100 and price is not. The bound is the cause, and it has nothing to do with the market losing strength.
That is the structural reason this signal fails where it costs most.
It is only visible afterwards
You cannot see a divergence until the second high is in, which is at or after the price you would have wanted to sell.
The two definitions cover everything
Covered above. Regular and hidden divergence between them name almost every disagreement, in opposite directions.
You found the ones that worked
Every top has a divergence before it. So do eight bars in fifty-four of a chart that did nothing — and at the moment the second high prints, those are the same picture.
Related
RSI is the indicator underneath, and why a bounded 0–100 series behaves this way in a trend.
Reversals is what divergence claims to predict, and how often that claim lands.
And market structure is the read that has to come first — in a confirmed uptrend, a bearish divergence is a reason to manage a position, not to take the other side.
I count these rather than rely on the ones that stick in the mind. Write the rule down, let a computer find every instance, and the hit rate stops looking impressive very quickly. Where I still glance at it is as a reason to tighten a stop on something I am already in, which is a much smaller claim than the one usually made for it.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money. Some links on this page earn a commission if you buy through them. It costs you nothing and it does not decide what appears here or in what order — how these pages are made is set out in our methodology.