Order Flow

Overview

The Order Flow strategy reads the market's buying and selling pressure to identify divergences between price action and actual order flow. When price moves one direction while actual money flow moves the other, it signals an imminent reversal.

Category: Order Flow Timeframe: Single (5m default) Best Regime: All regimes Typical Hold Time: 1-2 hours Min Confidence: 0.70

Core Concepts

Cumulative Volume Delta (CVD)

CVD tracks the cumulative difference between buying and selling volume. VECTRA calculates CVD using candle analysis (close position within high-low range) and, when available, real WebSocket trade data that identifies actual buyer-initiated vs. seller-initiated trades.

CVD Divergence Detection

The most powerful signal in this strategy:

  • Bullish Divergence: Price makes lower lows, but CVD makes higher lows → Hidden buying pressure (accumulation)

  • Bearish Divergence: Price makes higher highs, but CVD makes lower highs → Hidden selling pressure (distribution)

VECTRA grades divergences as mild, moderate, or strong based on the magnitude of disagreement.

Volume Imbalance

Beyond CVD, the strategy analyzes:

  • Buy vs. Sell volume ratio

  • Volume profile (where most volume occurred)

  • Large institutional order detection ($50K+ for BTC via WebSocket)

  • Volume spikes relative to moving average

Signal Generation

1

Long setup — Bullish conditions

  • Bullish CVD divergence detected (moderate or strong)

  • Volume spike confirms accumulation

  • Order flow imbalance shows buyer dominance

  • Price is not overbought (RSI < dynamic threshold)

  • Regime is not strongly bearish

Integration with Real Order Flow

When the WebSocket connection to Hyperliquid is active, VECTRA collects real-time trade data including:

  • Actual taker buy vs. sell classification

  • Large order flagging (institutional-size trades)

  • Liquidation trade identification

  • Aggregated flow metrics over configurable time windows (default: 300 seconds)

This real data supplements the OHLCV-derived flow estimates, providing a much more accurate picture of who is actually buying and selling.