How It Works

Six-stage intelligence pipeline

Every signal passes through a rigorous multi-stage analysis pipeline before it reaches your bot. No single indicator trades alone. Confluence is mandatory.

  ┌─╱╲─┐
  │╱  ╲│
  ╱ ◉◉ ╲
  └────┘

Stage 01 — Detect

Multi-Indicator Scan

Four independent indicator families — Nadaraya-Watson Envelope, Traditional Pivots, Multi-Divergence Analysis, and Funding Rate Intelligence — continuously scan every monitored pair for actionable setups. Each indicator operates on its own mathematical basis, eliminating confirmation bias.

  ╲  │  ╱
   ╲ │ ╱
    ╳╳╳
   ╱ │ ╲

Stage 02 — Converge

Confluence Evaluation

When two or more indicators independently agree on direction and timing, a confluence event fires. The system calculates a confidence score based on indicator agreement strength, divergence count, NW envelope position, and pivot proximity — requiring a minimum threshold before any signal is generated.

  ╔══════╗
  ║ GATE ║
  ║ ▓▓░░ ║
  ╚══════╝

Stage 03 — Gate

Optimization Gate

Every confluence signal passes through a dynamic optimization gate trained on the bot's own historical performance data. Regime analysis, time-of-day filters, and strategy-specific win rate thresholds determine whether a signal strong enough to generate meets the conditions to be worth acting on.

  ▶ ─── ▶
  │ API │
  ▶ ─── ▶
  │ BOT │

Stage 04 — Deliver

Real-Time Signal API

Qualifying signals are published to the Northstar Signals API within milliseconds. Each signal includes entry price, direction, confidence score, stop loss level, take-profit targets, leverage recommendation, and the full confluence breakdown. Your bot receives everything it needs to execute.

  ◄──────►
  │REPLAY│
  │ ↺↻↺↻ │
  ◄──────►

Stage 05 — Replay

Trade Replay Engine

Every closed trade is reconstructed candle-by-candle in the Trade Replay system — visualizing entry timing, TP/SL placement against actual price action, and MFE/MAE metrics. Backtesting against historical data validates strategy changes before they go live.

  ◈══════◈
  ║AURORA║
  ║ ✦ AI ║
  ◈══════◈

Stage 06 — Analyze

Aurora Deep Analysis

Completed trades feed into Aurora — our dual-AI deep analysis engine. Two LLMs combine to identify regime weaknesses, hourly patterns, and optimization opportunities that continuously sharpen the signal engine.

Signal Intelligence

Four independent signal families

Northstar's crypto trading signals are generated from four mathematically independent indicator families. Each detects opportunity through a fundamentally different lens — and they must agree before a signal fires.

📐

Nadaraya-Watson Envelope

Kernel regression envelope that dynamically calculates fair-value price channels. When price breaches the upper or lower envelope boundary, it signals statistical overextension — a potential mean-reversion opportunity. Width adapts to volatility in real time.

◆

Traditional Pivots

Classic support and resistance framework computed from prior-period high, low, and close. Seven levels (S3 through R3) act as magnetic price targets and reversal zones. Northstar detects when price approaches or touches these levels as potential entry triggers.

⚡

Multi-Divergence Analysis

Scans for divergences across RSI, MACD, and Stochastic simultaneously. When price makes a new low but momentum indicators don't confirm it, it signals exhaustion. Multiple divergence indicators agreeing compounds the signal strength exponentially.

💰

Funding Rate Intelligence

Monitors perpetual futures funding rates across exchanges to detect market positioning shifts. Funding flips — when rates cross from negative to positive or vice versa — reveal when the crowd's directional bet is changing, providing powerful confirmation for technical setups.

Contextual Intelligence

Regime-aware market analysis

Northstar doesn't just read indicators — it understands the broader market context. A proprietary regime classification engine identifies the current state and adapts signal generation accordingly.

/\ /🌙\ ‾‾‾

MOON BULL

Strong sustained uptrend

^ (o> ~

BABY BULL

Early bullish momentum

▲ ╱╲ ╱══╲

TURBO BULL

Explosive parabolic move

ʕ°ᴥ°ʔ

POLAR BEAR

Cold, steady downtrend

ʕ·ᴥ·ʔ

GRIZZLY BEAR

Aggressive bearish momentum

ʕ×ᴥ×ʔ

BLACK BEAR

Deep bearish capitulation

~≈~≈~

LIGHT CHOP

Low-volatility sideways

≋≋≋≋

ACCUMULATION

Smart money positioning

The optimization gate cross-references regime × direction × strategy to filter signals. For example, LONG signals in GRIZZLY BEAR regimes with historically low win rates are automatically suppressed — preventing the engine from fighting the trend.

Aurora Insights

Dual-AI deep analysis engine

Aurora is Northstar's proprietary deep analysis system — a dual-LLM architecture combining two AI engines to continuously evaluate and optimize signal performance. The engine discovers why wins and losses happen.

Edison Kosmos

Edison Scientific's Kosmos model ingests the complete trade history — entries, exits, regimes, time-of-day, indicator states, PNL outcomes — and produces a structured optimization report. It identifies regime × side toxicity, hourly performance patterns, loss attribution breakdowns, and quantified expected impact of proposed changes.

Kosmos operates as the data scientist: rigorous, quantitative, discovering patterns that human analysis would miss across thousands of data points.

Claude AI

Anthropic's Claude serves as the strategic analyst. It receives Kosmos's quantitative findings alongside raw trade data and produces actionable strategy recommendations — translating statistical patterns into concrete parameter changes, gate adjustments, and risk management improvements.

The dual-engine architecture prevents either model's blind spots from dominating. One finds the patterns. The other interprets them.

Aurora deep analysis — regime performance, hourly patterns, loss attribution, optimization impact

Win rate trajectory — self-optimization in action

As Aurora identifies and suppresses losing patterns, the engine's win rate trends upward over time.

API Preview

One endpoint. Everything you need.

Each crypto perpetual futures trading signal from Northstar arrives as a single JSON payload via REST API or WebSocket. Complete context for immediate execution — no interpretation required.

northstar-api — signal payload

// GET /api/v1/signals/latest { "signal_id": "ns-2026-0224-1847", "timestamp": "2026-02-24T18:47:12Z", "symbol": "BTCUSDT", "direction": "long", "timeframe": "15m", // Confidence & Confluence "confidence": 78, "confluence_count": 3, "indicators": ["nw_breach", "pivot_touch", "rsi_divergence"], // Execution Parameters "entry_price": 64436.60, "stop_loss": 63890.20, "take_profit_1": 64980.00, "take_profit_2": 65380.73, "leverage_suggested": 9, // Context "regime": "accumulation", "nw_position": "lower_breach", "pivot_proximity": "S1_touch", "div_score": 4, "risk_reward": 2.4 }

Why Northstar

Signals that understand markets,
not just price

Most crypto signal providers watch one indicator and call it analysis. Northstar requires independent mathematical agreement across multiple signal families before generating a single trading signal.

▸

Multi-Indicator Confluence

No single indicator trades alone. Every crypto trading signal requires two or more independent indicator families to agree on direction and timing before it fires. This eliminates false signals and dramatically improves win rate.

▸

Dual Timeframe Analysis

Simultaneous monitoring across 15-minute scalp and 1-hour swing timeframes. Cross-timeframe regime analysis ensures short-term signals align with broader market structure — preventing trades against the macro trend.

▸

Self-Optimizing Gate

The optimization gate continuously learns from its own trading history. Strategy-specific win rates, regime performance, and time-of-day analysis dynamically filter signals — the engine gets smarter with every trade it takes.

▸

Complete Execution Context

Each signal includes entry, stop-loss, multi-stage take-profit targets, leverage recommendation, confluence breakdown, and regime context. Your bot or trading system receives everything needed for immediate execution.

▸

Built for Machines First

Northstar's crypto trading signals API is designed for autonomous bot consumption — structured JSON payloads, sub-100ms latency, WebSocket streaming, and zero ambiguity. Humans benefit from the same precision with a clean dashboard interface.

▸

24/7 Autonomous Operation

The signal engine runs continuously with no human intervention. Automated health monitoring, exchange connectivity verification, and self-healing architecture ensure signals never stop flowing — even when you're asleep.

Crypto trading signals
built for machines

Six-stage intelligence pipeline

Four independent signal families

Regime-aware market analysis

Dual-AI deep analysis engine

Edison Kosmos

Claude AI

Win rate trajectory — self-optimization in action

One endpoint. Everything you need.

Signals that understand markets,
not just price

Multi-Indicator Confluence

Dual Timeframe Analysis

Self-Optimizing Gate

Complete Execution Context

Built for Machines First

24/7 Autonomous Operation

15+ perpetual futures pairs

The signal engine is live.
The API is coming.

Six-stage intelligence pipeline

Four independent signal families

Regime-aware market analysis

Dual-AI deep analysis engine

Edison Kosmos

Claude AI

Win rate trajectory — self-optimization in action

One endpoint. Everything you need.

Signals that understand markets,not just price

Multi-Indicator Confluence

Dual Timeframe Analysis

Self-Optimizing Gate

Complete Execution Context

Built for Machines First

24/7 Autonomous Operation

15+ perpetual futures pairs

The signal engine is live.The API is coming.

Signals that understand markets,
not just price

The signal engine is live.
The API is coming.