RSI Divergence — Spotting Reversals Before They Happen

How to use bullish and bearish RSI divergence to find high-probability reversal setups — including what large-scale backtesting reveals about when divergence actually works.

What is Divergence?

Divergence occurs when price and a momentum indicator move in opposite directions. It is a signal that the current trend is losing strength — the price is making new highs (or lows), but momentum is not confirming it.

This gap between price and momentum often precedes reversals. It does not guarantee a reversal, but it shifts the probability in favour of the counter-trend move.

We use RSI (Relative Strength Index) for divergence because it is normalised to a 0–100 scale, making it easy to spot when it is making higher highs or lower lows alongside (or against) price.

Two Types of Divergence

Regular Bearish Divergence — Warning a Top is Forming

Condition: Price makes a higher high, but RSI makes a lower high.

What this means: Price is still rising, but the buying momentum behind each new high is getting weaker. Fewer and fewer buyers are chasing each new peak. Eventually, the buyers run out and price reverses.

How to trade it:

• Wait for bearish divergence to form over at least two swing highs on price
• RSI should ideally be above 65 at the first high — see the RSI zone section below
• Look for a bearish candlestick pattern at the second high (Shooting Star, Bearish Engulfing) as confirmation
• Enter short on the close of the confirmation candle, stop above the recent high
• Target: the prior swing low, or a 3× ATR measured move from entry

Best on: Daily charts (4H or weekly for stronger signals), at resistance levels, after extended uptrends.

Regular Bullish Divergence — Warning a Bottom is Forming

Condition: Price makes a lower low, but RSI makes a higher low.

What this means: Price is still falling, but sellers are losing conviction. Each successive low requires less selling pressure to achieve. The floor is firming up.

How to trade it:

• Wait for bullish divergence to form over at least two swing lows on price
• RSI should ideally be below 35 at the first low — see the RSI zone section below
• Look for a bullish candlestick pattern at the second low (Hammer, Bullish Engulfing) as confirmation
• Enter long on the close of the confirmation candle, stop below the recent low
• Target: the prior swing high, or a 3× ATR measured move

Best on: Daily charts at key support levels, after extended downtrends.

Hidden Divergence — Continuation Signals

There is a second type of divergence that traders use for continuation rather than reversal:

Hidden Bullish Divergence

Condition: Price makes a higher low, but RSI makes a lower low.

This occurs during an uptrend on a pullback. Price is holding up (higher low) while RSI pulls back more than price — meaning the pullback is shallow in price terms. This signals trend continuation upward.

Hidden Bearish Divergence

Condition: Price makes a lower high, but RSI makes a higher high.

Occurs during a downtrend during a bounce. Price cannot recover as high as before (lower high), but RSI recovers more — meaning the bounce is weak. Signals trend continuation downward.

Most professional divergence systems use both regular and hidden divergence. Hidden divergence is generally considered more reliable because it trades with the trend rather than against it. Charles Schwab’s market research notes that hidden divergence “precedes powerful continued moves.”

What Large-Scale Research Says About Divergence

This is where it gets interesting — and where most retail content gets it wrong.

Thomas Bulkowski — trader, author of Encyclopedia of Chart Patterns and Encyclopedia of Candlestick Charts (both published by Wiley), and founder of thepatternsite.com — systematically tested 19,294 divergence samples across 994 stocks from 1995 to 2010. His findings are sobering but actionable:

Finding 1 — Market Context Is Everything

Bullish divergence in a bull market outperforms the index at every time horizon tested (3 weeks, 1 month, 2 months, 3 months). All other combinations — bearish divergence in a bull market, bullish divergence in a bear market — either underperform or match the index.

Practical takeaway: In a sustained bull market, focus on bullish divergence setups. Bearish divergence in the same environment has a poor track record; stocks showing bearish divergence in a bull market tend to continue rising regardless.

Finding 2 — The RSI Zone Matters Enormously

This is the finding most traders miss. Including divergences where RSI’s first peak or valley is between 30 and 70 hurts results “in nearly all categories.”

In plain terms: divergence signals where RSI never reached a genuine extreme (below 30 or above 70) are significantly weaker than signals formed at true extremes.

Bulkowski’s direct quote: “It’s best to ignore divergence when the first peak or valley occurs between 30 and 70.”

What this means practically: If you see RSI divergence with RSI hovering at 45 or 55, skip it. Wait for setups where the first swing point had RSI genuinely pushed to an extreme. That is where the real edge lives.

Finding 3 — Win Rate Is Lower Than You Think

Even the best combination — bullish divergence in a bull market — only beats the S&P 500 45 to 48 percent of the time. The strategy fails more often than it works.

This is not a bug; it is a feature. Divergence strategies work through asymmetric reward — losers are cut quickly at the stop, winners are held to a significant target. A 40% win rate with 3:1 reward-to-risk still produces strong positive expectancy. Trying to improve win rate by adding more and more filters often destroys the very trades that drive profitability.

Finding 4 — Pivot Spacing Affects Quality

Divergence tends to work best when the two swing points being compared are 1 to 2 months apart. Pivots that are only a week or two apart produce noisier signals. A divergence formed over a longer structural period represents a more meaningful shift in momentum than one formed over a few days.

The RSI Zone Rule in Practice

Based on Bulkowski’s analysis and validated in our own backtesting (see below), here is a practical checklist for RSI zone filtering:

For bullish divergence:

• First swing low: RSI ideally at or below 35 (the more oversold, the better)
• Second swing low: RSI higher than the first, showing momentum improvement
• Signals where the first RSI low is between 40–60 should be skipped or treated with reduced confidence

For bearish divergence:

• First swing high: RSI ideally at or above 65 (the more overbought, the better)
• Second swing high: RSI lower than the first
• Signals where RSI never exceeded 60 should be treated with scepticism

A Step-by-Step Divergence Scan

When looking at any chart for divergence, follow this checklist:

1. Identify the market regime — Is the broader market in a bull or bear trend? Divergence only has a proven edge when trading with the market regime.
2. Identify the stock’s trend — For bullish divergence, look for stocks in a short-term downtrend within a longer-term uptrend.
3. Mark the swing points — At least two clear swing highs or lows. Ideally 1–2 months apart.
4. Check RSI zone at the FIRST swing point — Was RSI genuinely at an extreme (below 35 or above 65)? If RSI was between 40–60, skip the setup.
5. Compare RSI at both swing points — Is RSI moving in the opposite direction to price?
6. Look for confirmation — A reversal candlestick pattern (Hammer, Engulfing), a break of a short-term trendline, or a MACD cross.
7. Check timeframe alignment — Daily chart divergence carries more weight than intraday. Weekly divergence is the strongest signal.

Our Own Backtesting Numbers

We ran a systematic backtest of a multi-indicator divergence system (RSI + Stochastic + MACD confluence required, all three must agree on the same pivot) across 54 NASDAQ 100 stocks over a 5-year period. Here is what the data produced:

The 37.6% win rate aligns directly with Bulkowski’s finding of 45–48% best case for single-indicator divergence — our stricter 3-indicator confluence gate naturally reduces win rate slightly while improving quality per signal.

The profit factor of 3.34× is the more important number: for every $1 lost across all losing trades, the system returned $3.34 from winning trades. This asymmetry is what makes the low win rate sustainable and profitable.

Exit breakdown across 197 trades: stop loss (58.9%), time stop at bar 3 (26.9%), trailing stop (7.6%), full target (6.6%). The full target was hit only 6.6% of the time — yet those trades, combined with the trailing stop exits, drove the bulk of the system’s profit.

Why RSI Settings Matter

The standard RSI is 14 periods. Some traders use shorter settings (9 or 7) for more sensitivity, or longer (21 or 25) for fewer false signals. For swing trading on the daily chart, 14 is a solid default.

In Indian markets (NSE/BSE), many prop traders use RSI(14) on the daily chart for overnight swing trades, and RSI(9) on the 15-minute chart for intraday.

Pitfalls to Avoid

Divergence in a strong trend can persist far longer than you expect. In a strong bull run, bearish divergence can appear on RSI and remain unresolved for weeks while price continues rising. Never fight the primary trend based on divergence alone.

Divergence in the RSI neutral zone is noise, not signal. This is the single most common mistake — acting on divergence when RSI never reached a genuine extreme. Apply the zone filter consistently.

Not all swing points are equal. A swing high formed on 3× average volume is more significant than one formed on thin volume. High-volume swing points create more reliable divergence signals.

Divergence is a filter, not a trigger. The divergence tells you to be on alert. The candlestick pattern or trendline break gives you the actual trade entry. Never enter solely because divergence exists.

Win rate is not the metric to optimise. A 40% win rate with 3R targets is more profitable than a 60% win rate with 1R targets. Evaluate divergence systems on profit factor and average R-per-trade, not win rate alone.

This post is for educational purposes only and does not constitute financial advice.