Black Friday·50% off your first yearEnds in
Confluence Trading: How to Stack Independent Signals
Updated 7 min read

Confluence Trading: How to Stack Independent Signals

President and co-founder, Elite Signals

Reviewed by Elie Faissal, CEO and co-founder of Elite Signals. Hypothetical examples only; no real trades or results are shown.

Short answer

Confluence trading means waiting until several independent reads point the same way before you enter: a location (a price zone), the higher-timeframe trend, momentum, participation (volume) and a trigger candle. The key word is independent. Three oscillators built from the same closing prices are one read, not three. Each layer you add filters out setups, so you take fewer trades; whether that helps you is something to test, not assume.

Most single-signal trades fail in a predictable way. A candle looks strong, or price touches a level, and the trader enters on that one piece of evidence. Confluence is the habit of asking for a second and third opinion first, from tools that look at different things. Those opinions can come from indicators on your own chart (EliteAlgo ONE, for example, waits for the trend structure to agree before it marks a signal) or from an outside signal service; our guide to trading signal platforms covers how to judge the second kind.

This guide covers what counts as an independent signal, how to build a stack layer by layer, a worked example with the risk math, and the trap most confluence traders fall into: adding filters until the backtest looks perfect.

What confluence means (and what it does not)

A confluence setup is one where separate pieces of evidence agree on direction and location. The useful way to think about it is that each layer should answer a different question about the market:

Five stacked bands. Location: is price at a level where a reaction is plausible? Context: does the higher timeframe agree? Momentum: is the move into the level slowing or speeding up? Participation: is there real activity behind the move? Trigger: is there a specific reason to enter now? A side note groups RSI, stochastic and MACD into the single Momentum band, because the same input counts as one read.
Location, context, momentum, participation and trigger each answer a different question. Three oscillators answer the same one.
The five common confluence layers and the question each one answers
LayerQuestion it answersTypical tools
LocationIs price at a level where a reaction is plausible?Support and resistance zones, prior swing highs and lows
ContextDoes the higher timeframe agree with the direction?Daily or 4-hour trend, market structure
MomentumIs the move into the level slowing or speeding up?One oscillator (RSI, stochastic or similar), divergence
ParticipationIs there real activity behind the move?Volume compared with its recent average
TriggerIs there a specific reason to enter now?A reversal candle or a break of a small lower-timeframe swing

What confluence is not: stacking tools that measure the same thing. RSI, the stochastic and a MACD histogram are all calculated from recent prices, so when one turns, the others usually turn with it. Seeing all three agree feels like strong evidence, but it is closer to one opinion said three times. Pick one tool per layer.

Why price zones are the usual base layer

Most confluence stacks start from location because it is the layer with the most outside evidence behind it. A Federal Reserve Bank of New York study of support and resistance levels published by six foreign exchange firms found "strong evidence that the levels help to predict intraday trend interruptions", while noting the predictive power varied across currencies and firms [1]. A later study of orders at a large dealing bank found that stop-loss and take-profit orders are "strongly clustered at round numbers, which are often used as support and resistance levels" [2]. In other words, zones partly work because many participants place orders around the same prices. For how to draw and grade zones, see our support and resistance zones guide.

Pattern-based reads have academic support too, with a caveat. Lo, Mamaysky and Wang tested chart patterns such as head-and-shoulders and double bottoms over a 31-year sample and concluded that "several technical indicators do provide incremental information and may have some practical value" [3]. "Incremental information" is a modest claim. It means a read adds something, not that it decides the trade.

Building the stack, one layer at a time

Start on the higher timeframe

Work top-down. Mark the nearest zone and the trend on the daily or 4-hour chart first; that is the context. Then drop to the timeframe you execute on (15-minute or 5-minute for intraday traders) and wait for price to reach the higher-timeframe level. A 5-minute signal that runs against the daily trend is a weaker stack than one that runs with it. Our guide to combining multiple timeframes covers the sequence in detail.

Add one momentum read

At the zone, compare the oscillator's swings with price's swings. If price makes a lower low while momentum makes a higher low, that is bullish divergence: the push into the level is losing force. It does not mean a reversal is coming. It means the selling looks weaker than it did.

Check participation

Volume tells you how much activity is behind a move. A push into a resistance zone on fading volume looks tired. A bounce off a support zone on rising volume suggests more participants are acting at that level. Set the rule in advance, for example volume on the reaction bar at least 125% of its 20-bar average, so you are not judging it by eye after the fact. For forex, where there is no central volume figure, broker tick volume is a rough stand-in.

Wait for the trigger

The last layer answers "why now". A reversal candle that closes back above the zone, or a break of the small swing high on the execution chart, is a trigger. Without one, you are early, and early entries at a zone are the ones most likely to be shaken out before the level holds.

Worked example: a four-layer long on EUR/USD

This is a hypothetical walk-through to show how the layers and the risk math fit together. It is not a real trade and not a prediction.

Illustrated EUR/USD 15-minute chart, not real data. Price declines into a shaded demand zone from 1.0845 to 1.0855 and makes a lower low at 1.0848 while the oscillator pane below makes a higher low. The volume bar on the push into the zone is highlighted, about 40% above its 20-bar average. An engulfing candle closes inside the zone; a long entry at 1.0852 is drawn with a stop at 1.0840 and a target at 1.0882.
Price makes a lower low inside the demand zone while the oscillator makes a higher low, volume rises on the push in, and a bullish engulfing candle closes inside the zone.
Hypothetical four-layer stack, EUR/USD 15-minute chart
LayerWhat the chart shows (hypothetical)Passes?
LocationDaily demand zone around 1.0850, marked three days earlier after a sharp bounce; price reaches it at 9:42 AM ETYes
MomentumPrice makes a lower low at 1.0848; the oscillator makes a higher lowYes
ParticipationVolume on the push into the zone is about 40% above its 20-bar averageYes
TriggerA 5-minute bullish engulfing candle closes inside the zone at 9:45 AM ETYes

Hypothetical trade plan

Entry 1.0852. Stop 1.0840, about 10 pips below the zone. Target 1.0882, the previous swing high. Risk is 12 pips and the planned reward is 30 pips, a reward-to-risk ratio of 2.5 to 1. On a $10,000 account risking 1% ($100), 12 pips at about $10 per pip per standard lot gives a size of 0.83 lots, rounded down to 0.8.

If the stop is hit, the loss is roughly $96, a number you knew before you entered. If any one of the four layers had failed, the plan was to stay flat.

The sizing step can be worked out with our position size calculator, and why a 2.5 to 1 plan can still lose money over time depends on how often it works, which is the subject of our guide to hit rate versus reward-to-risk.

The trap: more filters is not always better

Every layer you add removes setups. That is the point, but it has two costs.

Funnel of hypothetical setup counts for one market over one month: zone touch only, 40; plus higher-timeframe trend, 18; plus momentum, 8; plus volume and trigger, 3. Counts of setups only, no outcomes.
Adding layers cuts 40 zone touches down to 3 setups. Fewer setups means a few weeks of bad luck can dominate the sample.
How a stack thins out setups (hypothetical counts for one market over one month)
Rule setSetups that qualifyWhat you give up
Zone touch only40Nothing filtered; many touches break straight through
Zone + higher-timeframe trend18Counter-trend bounces
Zone + trend + momentum8Fast V-shaped reversals that never show divergence
Zone + trend + momentum + volume + trigger3Most moves; a few weeks of bad luck can dominate the sample

The first cost is sample size. With three trades a month, it takes years to learn whether the stack does anything at all. The second cost is overfitting. If you try enough combinations of filters on past charts, one of them will look excellent by chance. Finance researchers have made this point formally: Harvey, Liu and Zhu argue that after so much data mining, the usual statistical bar for a "discovery" is far too low, and that new findings need a higher hurdle [4]. The same logic applies to a trader testing 30 indicator combinations on one chart.

A practical defence: write the stack down before you test it, test it on charts you did not use to design it, and keep the number of layers small enough that you still see a reasonable number of setups. Three to four independent layers is a common stopping point.

When the layers disagree

Disagreement is information. If the zone says long but the daily trend is down and volume is drying up, the market is telling you it has not decided. Stay flat. This is also why confluence traders do poorly in choppy, directionless stretches if they lower their standards to stay busy; see filtering false signals in choppy markets.

  1. Write the layers and the pass rule for each before the session, ideally while you scan pre-market.
  2. Check that each layer is independent. One tool per question.
  3. Require every layer to pass. "Three out of four" quietly turns into "whatever I see today".
  4. Set the stop from the location layer (beyond the zone), then size from the stop.
  5. Log every setup, taken or skipped, so you can review whether each layer earns its place.

Common questions

How many confluence factors should a setup have?

Three to four independent layers is a common range: location, higher-timeframe context, and one or two of momentum, volume or a trigger candle. Beyond that, setups get so rare that you cannot tell whether the rules work.

Do RSI and MACD count as two confluence factors?

Usually not. Both are calculated from recent closing prices, so they tend to agree with each other by construction. Treat them as one momentum read and look for your second factor in a different layer, such as location or volume.

Does confluence trading work on 1-minute charts?

The logic is the same on any timeframe, but the window between the layers lining up and the entry is very short, and spreads and fees take a bigger share of a small move. Many traders mark the higher-timeframe zones before the session and use the 1-minute chart only for the trigger.

Sources

  1. Federal Reserve Bank of New York, Support for Resistance: Technical Analysis and Intraday Exchange Rates, 2000.
  2. Federal Reserve Bank of New York, Currency Orders and Exchange-Rate Dynamics: An Explanation for the Predictive Success of Technical Analysis, 2001.
  3. National Bureau of Economic Research, Foundations of Technical Analysis: Computational Algorithms, Statistical Inference, and Empirical Implementation, 2000.
  4. National Bureau of Economic Research, ... and the Cross-Section of Expected Returns, 2014.

Written by

President and co-founder, Elite Signals

Joe Zaylah is co-founder and President of Elite Signals. He has traded forex, stocks and options for 11+ years, runs the company's software and operations, and oversees the editorial direction of everything published on the site.