Divergence: regular and hidden, bullish and bearish, with RSI and MACD

A divergence is a disagreement between price and a momentum indicator. Price makes a new high, but the RSI or MACD does not. Or price makes a new low, but momentum holds above its previous low. The indicator is saying that the latest push was weaker than the one before, even though price went further.

Divergence is one of the most quoted ideas in technical analysis and one of the most abused. It is easy to see on a finished chart, hard to define before the fact, and it can keep building for months while a trend continues. This guide defines the four kinds of divergence precisely, shows two real examples on Tesla with both RSI and MACD, reports a small mechanical test on seven large US symbols, and explains how traders confirm a divergence before acting.

What divergence measures

Momentum indicators condense the speed of recent price changes. RSI compares the size of recent gains with recent losses. The MACD line is the gap between a 12-period and a 26-period exponential moving average, so it widens when price accelerates and narrows when it slows.

When price climbs to a higher high on a smaller burst of buying than the previous high, momentum makes a lower high. That is the whole idea: the move is still going, but it is covering less ground per bar, or it took a few large days to get there rather than steady gains. A trend can survive that for a long time. Divergence tells you about the quality of the move, not its direction tomorrow.

The four kinds of divergence

Regular divergence looks at the extremes that extend a trend: the highs in an uptrend and the lows in a downtrend. Hidden divergence looks at the corrective swings: the pullback lows in an uptrend and the rally highs in a downtrend. A hidden bullish divergence means the latest pullback pushed the oscillator lower than the last one did, yet price held a higher low. Many traders read that as a sharp but shallow shake-out within a trend that is still intact.

Andrew Cardwell, who studied RSI extensively in the 1990s, called the hidden patterns "positive and negative reversals" and treated them as trend-continuation signals. Some books use "positive divergence" and "negative divergence" for the regular kinds. The labels vary. The geometry above does not.

How to draw a divergence correctly

Most bad divergence calls come from loose definitions. A consistent procedure:

1. Find two swing points in price. A swing high is a bar whose high or close exceeds the bars on each side of it, for example the highest close of the five sessions before and after. Use the same rule every time.
2. Read the indicator at those exact bars. Not at the indicator's own peak a few bars away. If price peaked on October 1 and November 3, compare the RSI values on October 1 and November 3.
3. Keep the swings a sensible distance apart. Two highs three bars apart are one high. Two highs a year apart describe different markets. Between about two weeks and three months covers most daily-chart divergences.
4. Require a meaningful difference. An RSI of 71.2 against 70.8 is noise. Many traders want several points of disagreement.
5. Stay on one timeframe. A daily price high compared with a weekly RSI reading is not a divergence. See Multi-timeframe analysis: weekly to daily to intraday, top down for how to combine timeframes properly.

Two real examples on Tesla

Bullish divergence, March–April 2025

On March 10, 2025, Tesla closed at $222.15 after a long slide, with RSI at 20.6, deeply oversold, and the MACD line at −33.9. Four weeks later, on April 8, it closed at $221.86: a marginally lower low. But RSI was now 35.7 and MACD had risen to −13.6. Price had returned to the same place on far less downside momentum. Both indicators agreed.

Twenty sessions later, on May 7, Tesla was 24.5% higher. That looks like a textbook success, and partly it was. But the second low came one day before the April 9 tariff pause that lifted the whole market, so most stocks rallied hard from there whether or not they showed a divergence. One example can't separate the signal from the backdrop.

Bearish divergence, October–November 2025

By October 1, 2025, Tesla had more than doubled from its April low, closing at $459.46 with RSI at 73.8, overbought, and MACD at 27.8. On November 3 it closed higher, at $468.37. RSI, however, peaked at only 60.5 and MACD at 11.8. The new high came with far weaker momentum. Tesla closed as low as $395.23 on November 20, down 15.6%, and twenty sessions after the second high, on December 2, it was still 8.4% lower.

Look closely at the chart, though: after that dip, Tesla resumed its climb and closed at a new high of $489.88 on December 16. A divergence can mark a pause rather than a top, and a trader who read it as a top would have needed a clear exit plan in both directions.

What a small mechanical test shows

To get beyond hand-picked charts, we scanned two years of daily data (October 30, 2024 to October 29, 2026) for SPY, AAPL, NVDA, MSFT, TSLA, AMZN and META with a fixed rule: swing points are closes that are the highest or lowest of the five sessions on each side, the two swings must be 10 to 60 sessions apart, and RSI (14) must disagree with price by more than 4 points. We then measured the return over the next 20 sessions and compared it with every 20-session window in the same data.

Read this table carefully, because it mostly teaches restraint.

• Divergences are rare under a strict definition. Twenty signals in about 3,500 symbol-days. Most charts you see with "obvious" divergences use looser rules, or were picked because they worked.
• The two bullish divergences are really one event. Both fell in April 2025, at the bottom of a market-wide sell-off. They tell you about April 2025, not about divergence.
• Hidden bullish signals looked best, 8 of 9 up, but they occurred in stocks that were in uptrends during a strong bull market, when the baseline was already positive 56% of the time.
• Regular bearish divergence was a coin flip. In a rising market, one in three was followed by a gain anyway, and the average decline was small.

With samples this small, none of these differences is statistically meaningful. The honest conclusion is that divergence on its own is a weak, infrequent signal and needs a second piece of evidence.

Confirmation: what turns a warning into a signal

Because divergence can persist, most practitioners wait for price to confirm a change before acting:

• A break of structure. For a bearish divergence, the first close below the last swing low between the two highs. For a bullish one, the first close above the last swing high between the two lows.
• A trendline or moving-average break. A close through a rising trendline (see Trends and trendlines: how to identify a trend and draw it correctly) or the 20- or 50-day moving average.
• An indicator trigger. A MACD signal-line crossover, or RSI falling back below 70 or rising above 30. Wilder's failure swings, described on the Relative Strength Index (RSI): how to read and use it page, are a stricter version of this idea.
• Volume. A divergence in On-Balance Volume (OBV): reading volume flow and divergence or falling volume on the new high adds independent evidence, since it measures participation rather than speed.

Confirmation costs part of the move. That is the price of avoiding the many divergences that never turn into reversals.

RSI versus MACD for divergence

Because the MACD line is measured in dollars, a stock that has doubled will show larger MACD swings simply because its price is higher. Over long periods that can create or hide divergences. RSI's fixed scale avoids the problem, which is one reason many traders prefer it for divergence work. Both indicators measure similar things, so a divergence that shows up in both is less independent confirmation than it looks.

The Stochastic oscillator (slow %K and %D): how to read it can also show divergence and reacts even faster than RSI. Faster indicators show more divergences, and more of them fail.

How Finabase helps you look for divergence

Finabase does not detect divergences automatically. You can see them by adding the indicators to a chart:

• Open any symbol's chart, for example a two-year Tesla chart with RSI and MACD. Finabase computes RSI with Wilder's smoothing and MACD with TA-Lib's standard routine, both on closing prices. Indicator panels share the price chart's time axis.
• Use the trend line drawing tool to connect two price swings and the matching indicator readings, as in the chart above.
• Be aware of one naming quirk. Finabase's MACD panel labels the histogram "Divergence", a name inherited from the original Stock Master app. That histogram is the MACD line minus its signal line. It is not the price-versus- indicator divergence this page describes.
• The Scanner lists today's RSI overbought and oversold readings and MACD histogram crossovers across an exchange. Those are useful candidates to open and inspect for divergence, and a MACD crossover can serve as the confirmation step.

To test a divergence-plus-confirmation idea over time, set it up in Backtesting, keeping in mind the cautions in Backtesting basics: biases, costs and an honest SPY test.

Frequently asked questions

Is a bearish divergence a sell signal?

Not by itself. It says the latest high came with weaker momentum. Trends often continue through one or more divergences. Many traders wait for price to break a prior swing low or a moving average before treating it as a signal.

What is hidden divergence?

Hidden divergence compares the corrective swings of a trend. In an uptrend, price makes a higher low while the oscillator makes a lower low. It is usually read as a sign that the trend is intact and the pullback is ending, the opposite of what a regular divergence suggests.

Which indicator is best for spotting divergence?

RSI is the most common because its 0–100 scale makes readings comparable across time. MACD works too, but its values grow with price, which can distort long comparisons. The Stochastic oscillator (slow %K and %D): how to read it is faster and noisier. Using two momentum indicators adds less than it seems, because they measure similar things.

What timeframe should I use?

Any, as long as price and the indicator use the same one. Divergences on weekly charts are rarer and describe bigger moves. Divergences on 5-minute charts are frequent and short-lived.

Summary

A divergence is a disagreement between price and momentum. Regular divergences warn that a trend is tiring; hidden divergences suggest a pullback within a trend is ending. They are easy to see after the fact and much harder to define in advance. Drawn with a strict, consistent rule, they turn out to be rare, and in our small test their record was mixed and heavily shaped by the market backdrop. Use divergence as a reason to pay attention, then wait for price itself to confirm before you act.