RSI Divergence: Bullish, Bearish, and Hidden Patterns
RSI divergence compares the direction of selected price swings with selected RSI swings. It describes a disagreement between those series. Whether that disagreement predicts a useful trade depends on the pivot rule, timing, exits, costs, and data used to evaluate it.
Four patterns, four explicit comparisons
| Pattern | Price comparison | RSI comparison | Hypothesis to evaluate |
|---|---|---|---|
| Regular bullish | Lower low | Higher low | Potential upward reversal |
| Regular bearish | Higher high | Lower high | Potential downward reversal |
| Hidden bullish | Higher low | Lower low | Potential continuation of a defined uptrend |
| Hidden bearish | Lower high | Higher high | Potential continuation of a defined downtrend |
The words bullish and bearish describe the interpretation commonly attached to the pattern, not a guaranteed outcome. Hidden divergence does not prove a pullback is temporary. Regular divergence can persist while the existing price trend continues.
RSI itself compares Wilder-smoothed gains and losses. The TradingView RSI reference explains the calculation and reference levels. A threshold such as 30 or 70 can be part of a rule, but this article does not claim that adding it improves accuracy.
Make pivot selection reproducible
Before marking a divergence, decide whether price pivots use highs/lows or closes. Switching to a close-only line chart changes the underlying comparison and may remove a wick-based pattern. Use the same choice in research and execution.
Then specify the left and right bars required to confirm a pivot, the allowed distance between two pivots, and how RSI is paired with them. One approach samples RSI at price-pivot bars; another compares independently selected oscillator pivots. Those are different algorithms and can produce different signals.
A right-side confirmation requirement introduces delay. If a low at bar A is confirmed only after later bars, the alert becomes available later, even if a line is drawn back to A. TradingView's repainting documentation explains why plotting into the past can mislead a viewer about signal timing.
Turn the pattern into a trading hypothesis
Choose one pattern and define the trend context for it. State whether an additional price break is required, which bar confirms it, and when the order can execute. Specify exits and sizing before inspecting performance. A visual divergence by itself supplies none of these choices.
Compare a baseline with any extra filter on the same sample. Report changes in trade count and costs, not only win rate. RSI, moving averages, and MACD are all price-derived; agreement does not automatically create independent confirmation.
No universal length or timeframe is established here. Length 14 is a common starting configuration, while shorter and longer lengths alter sensitivity. Testing many configurations and reporting only the best creates selection bias. Freeze the rule before evaluating an unused period.
TradingView and thinkorswim workflows
For TradingView, inspect the source or documented rules of a divergence script. Check whether markers wait for confirmed bars, whether alerts match the markers, and whether historical lines can appear after the fact. Pineify's AI Coding Agent can help draft a custom rule; the result still needs compilation and behavior review.
Schwab documents RSI and Stock Hacker in thinkorswim. An RSI threshold scan is not automatically a divergence scan. A custom study must implement the pivot comparisons and comply with the scanner's constraints. Pine Script and thinkScript are different languages; a TradingView script cannot be pasted unchanged into thinkorswim.
Check current Pineify pricing for the feature and usage terms that apply. No universal lifetime, unrestricted-data, or user-count claim is needed to explain this workflow.
Record failures as well as successes
Log all qualifying patterns with their detection time, settings, hypothetical entry and exit, costs, and outcome. Keep screenshots from the decision time rather than only later charts where the outcome is visible. Include signals that never receive the extra confirmation and distinguish them from executed trades.
The useful question is whether a fully specified rule adds value after costs on independent observations. A selected chart or anecdote cannot establish that result, and a fixed number of paper trades does not automatically provide statistical confidence.
Does bullish RSI divergence mean I should buy?
It identifies a price/oscillator relationship. A trade decision needs additional explicit execution and risk rules, and the pattern can fail.
Is hidden divergence more reliable than regular divergence?
This article has no reproducible comparison establishing that. They express different hypotheses and should be evaluated separately under defined rules.
Why can a divergence look earlier on a historical chart?
A script may confirm a pivot using later bars and then draw the marker at the earlier pivot. Test the time the signal became available, not just its plotted location.
