How the Confluence Score Works: Mechanics, Bands, and Update Cadence

What the Score Represents

The CRYPTINT.IO confluence score is a single number per asset that answers one question: how strongly do our five independent data pillars agree on directional bias right now? It's a summary metric. It compresses a lot of underlying data into one readout so a trader can quickly see whether multiple signal domains are aligned.

The score runs 0 to 100. High numbers mean the pillars agree on bullish bias. Low numbers mean the pillars agree on bearish bias. Middle numbers mean disagreement or insufficient signal. The number itself is not a trade instruction. It's a description of the state of the market as the five pillars read it.

Critically, confluence measures agreement, not correctness. A high confluence score means the signals agree. It does not mean they're right. Markets can still move against aligned signals because the pillars can't see everything (single-exchange liquidity events, un-reported regulatory action, flash crashes driven by technical triggers nobody predicted). The score is a probability input, not a certainty.

How Each Pillar Is Scored

Each of the five pillars runs its own scoring logic before anything is combined. Understanding the individual pillar scores matters because the confluence score is only as good as the pillar inputs.

On-Chain Pillar

The on-chain pillar weighs supply dynamics, flow metrics, and holder behavior:

• Exchange balance trend (are coins accumulating on exchanges, suggesting selling, or moving off exchanges, suggesting holding?)
• Whale flow direction (large wallets net buying or selling?)
• Stablecoin dry powder (is buying capacity building on exchanges?)
• Smart money positioning (are consistently-profitable wallets accumulating or distributing?)
• Network activity (active addresses, transaction counts, fee trends relative to baselines)

Each input contributes to a 0-100 pillar score. Our guide to exchange flows covers the flow side; smart money tracking covers the wallet-level signal.

Sentiment Pillar

The sentiment pillar combines:

• Fear and Greed Index readings (market-wide mood)
• Social volume and polarity (Twitter/X, Reddit mention counts and tone)
• Funding rates (derivatives positioning bias)
• Bot-filtered signal strength (weighting out coordinated or inauthentic activity)

The contrarian nature of sentiment extremes is baked in. Extreme greed scores produce a bearish contribution; extreme fear produces a bullish contribution. Neutral sentiment is genuinely neutral.

Technicals Pillar

The technicals pillar aggregates:

• Momentum indicators (RSI, MACD, Stochastic) with crypto-calibrated thresholds
• Trend indicators (moving averages, Ichimoku)
• Volatility context (Bollinger squeeze state, ATR-relative positioning)
• Volume confirmation (OBV divergence, volume profile levels)

Crypto-specific indicators (Bitcoin Dominance, Hash Ribbons) feed the BTC technicals score and partially influence altcoin scoring via dominance rotation logic.

News Pillar

The news pillar scores recent and ongoing events:

• Regulatory developments (weighted by severity and directness to the asset)
• Institutional actions (ETF flows, corporate treasury announcements)
• Security events (hacks, exploits, exchange failures)
• Macro surprises (unexpected Fed, CPI, or geopolitical events)

News has decay. A hack this morning weighs heavily today, modestly tomorrow, and little next week unless the aftermath continues producing news.

Macro Pillar

The macro pillar reflects the broader environment:

• Fed policy stance (current and expected)
• Dollar strength (DXY level and trend)
• Bond yields (real and nominal)
• Equity correlation regime (risk-on vs risk-off)
• Liquidity conditions (central bank balance sheets, M2 growth)

Macro moves slowly but sets the background for everything else.

How the Scores Combine

The five pillar scores don't just get averaged. They're combined with regime-aware weighting.

In a macro-dominated regime (aggressive Fed action, geopolitical shocks, strong dollar moves), the macro pillar weight increases. The signal there is higher quality in that regime because macro forces are overwhelming other drivers.

In a crypto-specific regime (major ETF news, large hacks, cycle-driven rallies), the news and on-chain pillars get higher weight. Macro still contributes but isn't the primary driver.

In technical-range regimes (consolidation periods without major news or macro catalysts), the technicals and sentiment pillars carry more weight. The other pillars aren't producing strong signals, so they shouldn't dominate the composite.

This regime-adaptive weighting is the feature that separates confluence scoring from naive multi-indicator averaging. Our guide to pillar weighting covers the adaptation logic in detail.

Score Bands and Their Meaning

Once all five pillars are scored and weighted, the composite sits on a 0-100 scale. Specific bands carry interpretive meaning.

Score bands are calibrated to historical distributions. The vast majority of readings cluster in the 30-70 range. Extreme readings (below 15 or above 85) are infrequent by design and carry higher signal value when they occur.

Update Cadence

Pillars don't all update at the same speed. The composite score reflects the most-recent data from each pillar but the underlying inputs refresh at different rates.

The composite score recalculates as any input changes. For active BTC analysis, the score might update multiple times per minute during high-volatility periods. For less-active alts, hourly or 4-hour updates are more typical.

This matters because a confluence score read five minutes ago may no longer reflect current conditions, especially during fast-moving events. The score is a snapshot of current data at the moment it's computed.

What the Score Does Not Do

Clarity on limitations matters more than hype about capabilities.

• The score is not a prediction. It describes current data agreement, not future price.
• The score does not guarantee outcomes. High-conviction setups still fail. Low-conviction setups still move. Probability shifts, certainty doesn't.
• The score is not actionable alone. Position sizing, risk management, and your own thesis still matter. A 75/100 score is an input to your decision, not the decision itself.
• The score is not equally reliable across all assets. Majors (BTC, ETH) have the deepest data; long-tail alts have thinner coverage and less-reliable scoring.
• The score cannot see black swans. A protocol exploit, a geopolitical shock, or an SEC surprise that hasn't happened yet can invalidate any current reading instantly.

Treat the score as one of several inputs to your own analytical framework. It's there to save you the work of watching five domains simultaneously, not to replace your judgment.

Why This Design

The confluence score methodology reflects specific design choices. Each one has a reason.

Why 0-100? It's a familiar scale for traders from RSI, Stochastic, and Fear and Greed. Users intuitively understand "low number bearish, high number bullish" without needing to learn a new mental model.

Why five pillars? More pillars would dilute individual contributions and obscure the "which signal is driving this?" question. Fewer pillars would create blind spots. Five is the number that covers the important failure modes without being noisy.

Why regime-aware weighting? Because market conditions determine which signals are reliable. Averaging all signals equally ignores the reality that different pillars are diagnostic in different environments.

Why separate pillar scores? Because users can drill down into why the composite is what it is. A 70/100 driven by on-chain and sentiment tells a different story than a 70/100 driven by macro and technicals. Transparency into the pillar-level detail is part of the methodology.

Related Confluence Content

For deeper detail on specific aspects of the score:

• Pillar weighting: Why weighting beats averaging, and how regime adaptation works.
• Confluence case studies: Historical examples of confluence calling moves correctly, and moments where it failed.