Methodology
How We Generate
Every Signal
A complete, verifiable breakdown of the five scoring layers, risk gates, regime weights, and quality grade system that defines CryptoSignal Pro signals.
How does CryptoSignal Pro generate a signal?
Signal Pipeline
How a Signal is Generated
Raw Data
5 live feeds
Scoring
5 independent models
Confluence
0-100 score
6 Gates
all must clear
Grade
A+ through D
Feed
actionable signals
Overview
Signal Generation at a Glance
Every signal passes through five independent scoring layers before a grade is assigned. Weights shift automatically based on the detected market regime.
Risk gates then filter out setups that fail liquidity, event-blackout, data-integrity, trigger, R:R, or divergence checks before any signal reaches your feed.
Scoring Layers
Five Independent Layers
* Default weights for strong bull-trend regime. Weights shift automatically with detected market regime.
Technical Analysis
Trend direction, momentum, volatility, volume, support and resistance, chart patterns, and market structure form the primary layer.
Increases to 50% in strong trends. Reduces to 30% in ranging markets.
On-Chain Context
Exchange net flows, holder distribution, whale activity, valuation ratios (MVRV, SOPR), and network-health signals.
Weight stays stable across regimes — on-chain is regime-agnostic.
Sentiment Analysis
News NLP, social volume spikes, KOL signal tracking, event calendars, and natural-language sentiment classification.
Reduces to 10% in high-volatility and crash regimes where sentiment lags.
Macro Context
Dollar strength (DXY), rate expectations, token unlock calendars, FOMC/CPI/NFP schedules, and major economic events.
Increases to 20% in crash and recovery regimes where macro drives crypto.
Derivatives Data
Funding rates, open interest trends, long-short ratios, liquidation levels, and perpetual futures basis.
Increases to 20% in ranging markets where derivatives reveal positioning.
Grade System
A+ through D — What Each Grade Means
Risk Gates
Six Hard Gates
Gates run after scoring. A gate failure does not affect the confluence score — it simply prevents publication.
Liquidity Floor
Market cap or 24h volume must exceed $5M. Thin markets produce unreliable signals.
Event Blackout
Signals are suppressed during FOMC, CPI, NFP, and other high-impact macro events. Windows are defined per-event.
Data Integrity
Flash-wick detection identifies data anomalies. If a candle is flagged as corrupted, the signal is gated.
Trigger Gate
Entry confirmation must occur within the OB/FVG zone. Signals that trigger above/below the zone are invalidated.
Net R:R Floor
Net risk-reward after all fees must be >= 1.2 for B grade, >= 1.5 for A grade, >= 2.0 for A+ grade.
Divergence Veto
If the technical layer score and derivatives layer score diverge by more than 40 points, a veto fires and suppresses the signal.
63% of setups are filtered by these gates before reaching your feed. All six gates must clear independently. A single failure suppresses the signal regardless of confluence score.
Fee Deduction
Net R:R After All Fees
The gross R:R of a setup is never the headline number. All signals display net R:R after every known transaction cost.
// Gross reward-to-risk
grossRR = (TP − entry) / (entry − SL)
// Known costs
feeCost = 2 × 0.04% // taker in + taker out
slippageCost = 2 × 0.03% // in + out est.
fundingCost = current8hRate × holdDuration
netRR = grossRR − feeCost − slippageCost − fundingCost
Taker fee 0.04% + Slippage 0.03% = 0.07% per side. Round trip: 0.14% minimum deduction. 0.14% Funding rate is estimated from the current 8-hour rate and added on top.
Market Regimes
6 Detected Regimes
The engine classifies market state every 15 minutes and shifts scoring weights accordingly.
| Regime | Technical | On-Chain | Sentiment | Macro | Derivatives |
|---|---|---|---|---|---|
| Strong Trend | 50% | 20% | 15% | 5% | 10% |
| Weak Trend | 40% | 20% | 20% | 10% | 10% |
| Range | 30% | 20% | 20% | 10% | 20% |
| High Volatility | 35% | 25% | 10% | 15% | 15% |
| Crash | 30% | 20% | 10% | 20% | 20% |
| Recovery | 35% | 25% | 15% | 15% | 10% |
Regime detection uses a proprietary multi-factor classifier. Misclassification can occur during rapid transitions.
Limitations
What the Engine Cannot Do
Signals are not predictions.
The confluence score measures agreement between independent models. High agreement does not guarantee a trade will be profitable.
Data latency matters.
Source health monitoring flags stale feeds, but brief periods of degraded data quality can occur during exchange maintenance or API outages.
Regime misclassification.
Market regimes can transition quickly. The engine may lag by one or two candles when switching from trend to range mode.
Not financial advice.
The entire methodology is a systematic analytical framework. Apply your own position sizing and risk management rules.
Ready to See It in Action?
Start free. No card required. Upgrade only when you need more power.