What it is

Confluence is the situation where several distinct signals point to the same conclusion at the same time. A breakout above resistance that also coincides with rising volume, a bounce off a support level that lines up with an oversold RSI and a higher low in the broader trend - these are confluence. The intuition is old and sound: a conclusion supported by multiple independent witnesses is more trustworthy than one resting on a single clue.

Combining signals is the practice of building that confluence deliberately - defining a setup that requires two or three conditions to agree before you act, rather than trading on any one indicator alone. Done well, it raises the quality of your trades. Done badly, it quietly strangles the strategy. The difference hinges almost entirely on one concept: independence.

It helps to be precise about what a single signal is and is not. A signal is one condition that is either true or false at a given moment: "price is above its 200-day average", "RSI is below 30", "today's volume exceeds twice its average". Each on its own is a thin reed - any single indicator fires on plenty of occasions that go nowhere. The hope behind confluence is that requiring several reeds to bend the same way at once filters out the false alarms, leaving the rarer moments when genuinely different forces line up. Whether that hope is fulfilled or betrayed depends entirely on whether the reeds are truly different reeds or three copies of one.

How it works

The statistical engine behind confluence is the same one behind diversification: combining sources of information reduces the influence of any single source's noise - but only if the sources are independent.

• Independent signals measure genuinely different things. A momentum signal (RSI), a volume signal (volume versus its average), and a trend signal (price versus a long moving average) capture three distinct aspects of the market. When all three agree, that agreement carries real weight, because each could have disagreed for its own reasons and did not. Independent confirmations multiply confidence.
• Redundant signals measure the same thing in different costumes. RSI, the stochastic oscillator, and the MACD are all momentum indicators built from price; stacking all three and calling their agreement "confluence" is self-deception. They will almost always agree because they are computing variations of one underlying quantity. Three redundant signals are barely more informative than one - they are a single witness who has changed clothes three times.

The practical test is simple: ask what each signal actually measures. If two signals would tend to fire together by construction, they are redundant and their agreement proves little. If they measure different forces - price momentum versus participation versus valuation - their agreement is informative. Real confluence comes from combining signals across different families: a technical signal plus a volume signal plus a regime or trend filter, not three flavours of the same oscillator.

The diversification analogy is worth making concrete, because it is exactly the same mathematics. Holding ten stocks that are all banks gives you almost no diversification - they rise and fall together, so you effectively hold one big bank position. Holding a bank, a utility, and a technology firm spreads you across genuinely different drivers, and the combined risk is lower than the sum of its parts. Signals behave identically. Three momentum indicators are ten bank stocks: correlated, redundant, falsely reassuring. A momentum signal, a volume signal, and a trend filter are the diversified basket: when they agree, the agreement is meaningful precisely because each could independently have disagreed. The whole value of confluence comes from this independence, and nothing else.

How to use it

A disciplined way to combine signals:

1. Pick signals from different families. For example: one trend filter (price above a rising 200-day average), one entry trigger (a pullback to support), and one participation check (volume confirming the move). Each measures something the others do not.
2. Define agreement precisely. State the exact condition for each signal and require them to hold together. Vague "it looks good on a few indicators" is not a rule.
3. Use a trend filter as a regime gate, not as a third entry trigger. A common, robust structure is one filter that decides whether to trade at all and one or two triggers that decide when.
4. Count your conditions. Each added filter both raises selectivity and shrinks the number of trades. Stop adding when the marginal filter measures nothing new.
5. Prefer agreement across timeframes to agreement across more indicators on one timeframe. A daily trend agreeing with a weekly trend is genuine multi-scale confluence; two daily oscillators agreeing is usually redundancy. Different timeframes are a natural source of independence because they respond to different horizons of buying and selling.

Worked example. Suppose you trade pullbacks in an uptrend. A sound three-signal setup: (1) price above a rising 200-day moving average (regime), (2) a pullback to the 20-day average that holds (entry location), (3) the down-move on shrinking volume while the rebound comes on rising volume (participation). These three measure different things - long trend, short location, and conviction - so their agreement is real confluence. Replacing (3) with "and the stochastic is also oversold" would add a redundant momentum signal that mostly echoes the pullback already captured in (2).

There is also a useful distinction between a gating signal and a triggering signal, and confusing them is a common error. A gate is a slow condition that simply permits or forbids trading - the 200-day trend filter says "only look for longs while the regime is up". A trigger is the fast condition that actually fires the entry - the pullback holding and rebounding. A gate that you treat as a trigger floods you with entries; a trigger that you treat as a gate paralyses you. The clean structure is one or two gates that decide whether you are allowed to act, and one trigger that decides the precise when. When you count your conditions, count gates and triggers separately: stacking three triggers is usually redundancy, whereas one extra well-chosen gate genuinely raises quality without much thinning the opportunity flow.

Strengths & limits

The strength of confluence is selectivity with confidence: by requiring independent confirmations, you trade less often but with a higher-quality reason, which can lift win rate and expectancy. It also imposes a healthy patience. A single-indicator trader is tempted into the market constantly, because some indicator is always flashing something; a trader who waits for genuine confluence sits on their hands through most of the noise and acts only when several independent forces align. That enforced selectivity is often worth more than any individual signal, because the largest, most avoidable losses come not from bad signals but from trading too much, on thin reasons, in conditions that did not warrant a position at all.

The limits are two-sided, and the second is the silent killer. First, redundancy gives false comfort: three correlated signals feel like strong agreement but add almost no information, so you take on the costs of complexity with none of the benefit. Second, over-filtering. Every condition you add makes the setup rarer. Pile on enough filters and the perfect, fully-confirmed setup appears two or three times a year - too few trades to ever build a meaningful sample, to diversify across, or to let an edge express itself. Worse, heavy filtering is a direct path to overfitting: each extra condition is another knob you can twist to fit the past, producing a backtest that looks immaculate precisely because it has been carved to match history and will not survive new data. The art is balance: enough independent confluence to raise quality, few enough conditions that real, repeatable opportunities still occur often enough to matter. A reasonable rule of thumb is to stop at two or three genuinely independent conditions; beyond that, each new filter is far more likely to be curve-fitting than insight, and the dwindling trade count quietly destroys your ability to ever know whether the setup works at all. The hardest discipline is deleting a filter that you added in good faith but that, on honest inspection, only ever fired alongside the others and therefore added nothing but the comforting illusion of extra confirmation.

Key takeaway: Confluence means several signals agreeing, but the value comes only from independent signals measuring different things - redundant indicators give false confidence, and over-filtering with too many conditions makes setups too rare to trade and invites overfitting.