RSI divergence
RSI divergence is a disagreement between price and momentum. Price makes a new low while RSI makes a higher low (bullish), or price makes a new high while RSI makes a lower high (bearish), signalling that the move is continuing with less force behind it.
RSI is built from the size of recent up-closes against down-closes. So when a stock grinds to a new low but RSI fails to make one, the arithmetic is saying something specific: the second leg down took more sessions and smaller moves to cover the same ground. That is a regular bullish divergence, a lower low in price on a higher low in RSI. The bearish version is the mirror, a new high in price on a lower high in RSI, meaning each push up is being made with less one-sided buying than the last. Both are read as the trend running out of momentum before it runs out of price. The divergence usually appears a few bars to a few weeks before the turn it precedes, when it precedes one at all.
A textbook divergence is drawn between two swing points, which is a problem for a scan. Identifying a swing low means knowing that price rose afterwards, and by definition the current bar has no afterwards yet. The practical approximation is offsets. `rsi(14)[-20]` is RSI twenty sessions ago, which in a stock now sitting at a two-month low is a fair stand-in for the reading at the previous low. Comparing today's RSI against it, while requiring today's low to sit within a few percent of `min(low, 8w)`, captures the shape without a swing detector. The margin of three RSI points rather than any positive difference is there because a half-point divergence is noise, and the RSI cap keeps the list to stocks whose momentum is still depressed.
The approximation has honest limits. The previous low might have formed at bar 12 or bar 30 rather than bar 20, in which case the comparison is against a bar on the way into or out of it; widening the tolerance helps and costs more false positives. And the divergence itself is a condition rather than a trigger. Stocks in a genuine downtrend print bullish divergence after bullish divergence, each resolved by another lower low. The usual repair is a second condition that asks for evidence, RSI crossing back above 40 or a close above the prior bar's high, so the scan finds divergences that have started to resolve.
Hidden divergence is the continuation variety: a higher low in price on a lower low in RSI during an uptrend, read as a pullback whose momentum has reset further than its price has. The same offset construction expresses it with the inequalities swapped.
Its neighbours in the library are the other RSI reversal scans, the oversold turn and the 50 reclaim, which look at the level of RSI rather than its trajectory against price.
In Sift
Written as rsi(14) > rsi(14)[-20] — today's RSI against its reading twenty bars ago. A working scan — stocks at a two-month low whose RSI is higher than it was twenty sessions ago, a bullish divergence approximated with offsets:
where low within 3% of min(low, 8w)
and rsi(14) > rsi(14)[-20] + 3
and rsi(14) < 451
of the 500 most-traded NSE stocks match today, as of 25 Sept 2026
Scans that use it
Prebuilt scans in the library whose query reads this value — each with a hit-rate replay over the last 250 sessions.
RSI oversold turning up
RSI back above 30 within the last three sessions, in a stock still holding its 200-day average.
RSI overbought
RSI over 70 in a stock still above its 50-day average. Stretched, and worth knowing either way.
RSI reclaiming 50
RSI crossing back above its midline in a stock that never lost its long-term trend.
RSI bullish divergence
Price near a two-month low while RSI sits at least three points above where it was a month ago. The selling is running out of momentum before it runs out of price.
RSI bearish divergence
Price near a two-month high while RSI reads three points below its level of a month ago.
CCI overbought
CCI above +100 with RSI over 65. Two oscillators saying the same thing at once.
Williams %R overbought
Closing in the top fifth of its two-week range with RSI above 70.
Near a five-year low
Within 5% of the lowest low in five years with RSI under 35. The most abandoned names on the exchange, for a reversal hunt.
Fifteen sessions under the 20-day
Fifteen or more unbroken closes below the 20-day average, RSI under 35. A slump old enough to be exhausted.
Strong but not overbought
Up more than 20% in three months with RSI between 45 and 65, above the 50-day average. Room to run without chasing.
Strong directional trend
ADX above 30, +DI over −DI, the averages stacked, RSI still under 70.
Rate of change leaders
Rate of change above 10 with RSI still under 70, price above the 50-day average.
Pullback in an uptrend
RSI under 40 in a stock still above its 200-day average.
Seven up days in ten
Seven of the last ten sessions closed above their open, in a stock above its 50-day with RSI under 70.
Longest streaks above the 21 EMA
Twenty or more consecutive closes above the 21-day EMA, longest streak first.
Hammer while oversold
A long lower wick with RSI under 40.
Shooting star after a run
A long upper wick with RSI above 60. Buyers tried and could not hold it.
Bullish harami
A small up-day sitting entirely inside the previous session's big down-candle, with RSI already under 45, which is what separates a pause worth watching from a pause in the middle of nothing.
Bearish harami
A hesitation candle inside a big up-day, with RSI above 55.
Large-cap pullback
Market cap above ₹20,000 crore with RSI under 40. Size doing the quality filtering.
IT stocks in an uptrend
Information Technology names above their 200-day average with RSI over 55.
Common questions
What is the difference between regular and hidden RSI divergence?
Regular divergence is a reversal reading: price makes a new extreme and RSI does not, so the trend is making progress with less momentum. Hidden divergence is a continuation reading: in an uptrend price makes a higher low while RSI makes a lower low, so the pullback reset momentum more than it reset price. Both are the same comparison with the inequalities swapped.
How reliable is RSI divergence?
On its own, not very. A stock in a real downtrend can print one bullish divergence after another, each resolved by a further low. It works better as a change in pace that earns a stock a place on a watchlist, then confirmed by a separate trigger such as RSI reclaiming 40 or a close above the previous bar's high. The hit-rate replay on the scan page shows how the approximation has actually resolved on NSE history.
Why does the scan use rsi(14)[-20] rather than the previous swing low?
A swing low is only known in hindsight, because it needs price to have risen afterwards, and a scan is evaluated on the current bar. Comparing against a fixed offset twenty sessions ago, in a stock currently within a few percent of its two-month low, is a stand-in for the earlier low that a single-bar query can compute. It is an approximation and the page says so.