Divergence is the most maligned concept in technical analysis, and the criticism is fair. Put RSI on a chart, look for price making a higher high while RSI makes a lower high, and you will find dozens of instances a week. A small fraction lead anywhere. The rest are what people mean when they say divergence doesn't work.
But the problem was never the concept. It is that a single oscillator is one narrow read of momentum, and one narrow read produces a lot of disagreement with price that carries no information. The concept starts working when you stop asking one indicator and start requiring several to agree.
This post covers the two divergence types properly — they are frequently confused, and confusing them is expensive — and then the confluence mechanism that All Divergences is built around.
Regular divergence: the reversal read
Regular divergence occurs when price makes a new extreme and the oscillator doesn't.
- Bullish regular divergence: price makes a lower low; the oscillator makes a higher low. Price is still falling but the momentum behind the fall is weakening.
- Bearish regular divergence: price makes a higher high; the oscillator makes a lower high. The rally is extending but with less force behind each push.
The interpretation is exhaustion. The move is continuing on price but decaying underneath, and a reversal becomes more plausible. This is the divergence everyone knows, and the one every tool detects.
Hidden divergence: the continuation read
Hidden divergence — also called reverse or continuation divergence — is the mirror construction, and it means close to the opposite thing.
- Bullish hidden divergence: price makes a higher low; the oscillator makes a lower low. Price refused to give back as much as momentum did.
- Bearish hidden divergence: price makes a lower high; the oscillator makes a higher high. Momentum recovered further than price managed to.
The interpretation is trend strength, not exhaustion. In an uptrend, a bullish hidden divergence on a pullback says the pullback was shallow relative to how much momentum reset — the trend absorbed the correction. It is a continuation signal.
This is where a great deal of damage gets done. A trader who knows only regular divergence looks at a bullish hidden divergence, sees price and oscillator disagreeing, and reads it as a reversal warning in the wrong direction. Most single-oscillator tools don't detect hidden divergence at all, so the trader never learns the distinction — they just experience divergence as unreliable.
Why one oscillator isn't enough
Every oscillator is a specific formula applied to a specific input over a specific lookback. RSI measures the ratio of average gains to average losses. MACD measures the spread between two moving averages. OBV accumulates volume by direction. MFI weights price by volume. These are different questions, and at any given pivot they will frequently give different answers.
That's the useful part. If RSI diverges and nothing else does, RSI is telling you about its own lookback window. If RSI, MACD histogram, stochastics and CCI all diverge at the same pivot, the disagreement between price and momentum is broad enough that it isn't an artefact of one calculation.
All Divergences runs detection across ten built-in indicators simultaneously: MACD, MACD histogram, RSI, stochastics, CCI, Momentum, OBV, VW-MACD, Chaikin Money Flow and MFI. That mix is deliberate — some are pure price momentum, some are volume-weighted, some are volume-only. Agreement across categories is stronger evidence than agreement across four flavours of the same idea.
The minimum count filter
The mechanism that makes this practical is a simple threshold: only flag a signal when at least N indicators diverge at the same pivot. The indicator reports the count, so you can see how much agreement any given signal has.
In use, this changes the tool's character completely. At a threshold of one, you get every divergence any oscillator produces — which is the noisy experience described at the top of this post. At three or four, the signal count drops hard and what remains is qualitatively different: pivots where momentum broadly failed to confirm price.
The right threshold depends on how many indicators you have selected and how often you want to trade. Selecting all ten and requiring three is loose. Selecting four and requiring three is strict. It's worth being explicit with yourself about which you're doing.
External indicators
The tool accepts any external indicator series as an input and runs full divergence detection against it. That includes third-party indicators and your own custom NinjaScript.
This matters if you have a momentum or volume measure you already trust. Rather than eyeballing it for divergence alongside the built-ins, you plug it in and it participates in the count like any other. Your proprietary read becomes one of the votes.
Signal timing
Divergence detection is pivot-based, which creates an inherent lag: you cannot confirm a pivot until enough bars have formed either side of it. Traditional divergence tools inherit the full pivot delay, which is why divergence signals often arrive after the reaction has already begun.
Two modes address this. Same-bar detection flags the divergence as it becomes true, accepting that the pivot is still provisional. Next-bar confirmation waits one bar. Neither eliminates the underlying constraint — a pivot needs bars around it to be a pivot — but the same-bar mode gets you the read at the earliest point it is knowable.
Pivot sensitivity and historical depth are both adjustable, which is the other half of the timing question. A tight pivot definition finds more, earlier, weaker pivots; a loose one finds fewer, later, more significant ones.
Using it alongside the smart-money stack
Divergence is momentum information, and the rest of the toolkit is structural information. They complement each other precisely because they measure different things:
- At a structural level. Price arrives at a bullish Order Block or an inverted FVG. A four-indicator bullish regular divergence at that pivot says momentum is exhausted exactly where structure says a reaction is plausible.
- On a liquidity event. A swing failure pattern with regular divergence at the failed swing is a stop run into decaying momentum.
- Hidden divergence with the trend. Under bullish ICT Bias, bullish hidden divergence on a pullback is a continuation read that agrees with the higher timeframe. This is the pairing most traders miss entirely.
- Distinct from SMT divergence. Worth stating plainly: SMT compares price on one instrument against price on another. This tool compares price against oscillators on the same instrument. Different mechanisms, different information, and they can be used together.
Common mistakes
Reading hidden divergence as a reversal. The single most costly error, because it puts you against a trend at the moment the trend just demonstrated strength. Know which type you're looking at before you interpret it.
Running with a minimum count of one. This is the default temptation and it recreates the exact problem the tool exists to solve. If you are going to use one indicator's divergence, you don't need ten indicators loaded.
Trading divergence as an entry signal. Divergence says momentum and price disagree. It does not say where to enter, where the stop goes, or that the reversal happens now. Divergence can persist for a long time while price continues — that is normal, not a failure of the signal.
Stacking ten correlated indicators and calling it confluence. If your selected set is RSI, stochastics, CCI and Momentum, you have four variations on the same measurement. They will agree often, and the agreement is not independent evidence. Mix price momentum with volume-based measures if you want the count to mean something.
Ignoring the pivot sensitivity setting. Everything about divergence rests on what counts as a pivot. Leaving that setting at default while changing timeframes means the tool is answering a different question than you think it is.
The reframe
Divergence isn't unreliable. Single-oscillator divergence is unreliable, and that is a different claim. Once the signal has to survive a vote across ten independent momentum and volume measures, and once you can tell the reversal read from the continuation read, what's left is a genuine piece of information about whether a move has force behind it.
Disclaimer: Trading futures and other leveraged instruments involves substantial risk of loss and is not suitable for all investors. Past performance and indicator signals are not indicative of future results.
