AS-KBN-Futures — Futures Signal Engine (v4.6)

Python-based futures and crypto signal engine
Closed-bar, confluence-driven, ATR-managed alert model
Outputs: Discord, email, and SMS support in development

Overview

AS-KBN-Futures is a deterministic signal engine designed for futures and crypto market monitoring. It evaluates confirmed bars, computes a multi-factor indicator stack, scores long and short confluence, and emits structured alerts when a qualified setup or exit condition is detected. utrechtuniversity.github

The design emphasizes closed-bar validation, adaptive risk control, and consistent signal interpretation across symbols and timeframes. Its role is signal generation and state-aware alerting rather than brokerage execution. andreabergia

Signal Architecture

The engine follows a closed-bar event model:

1. Market OHLCV data is ingested from futures and crypto feeds.
2. Indicator values are computed from the confirmed bar set.
3. Long and short confluence scores are evaluated independently.
4. ATR-based risk levels are derived for stop, target, break-even, and trailing logic.
5. A per-symbol state model evaluates whether entry, exit, cap, or reset conditions are active.
6. Alert payloads are emitted to the enabled delivery channels. tilburgsciencehub

This structure keeps signal generation stable by avoiding intrabar repaint behavior and by separating indicator evaluation from downstream alert formatting. andreabergia

Indicator Stack

The engine computes its indicator set internally and converts each enabled factor into a directional vote or gating condition.

VWAP

VWAP provides institutional bias context.

• Long bias: close above VWAP.
• Short bias: close below VWAP.

MACD

MACD provides momentum confirmation.

• Long bias: histogram above zero.
• Short bias: histogram below zero.

SMC / FVG

Market structure logic identifies bullish or bearish fair-value gaps within the active lookback window.

Keltner or Bollinger Channels

Channel position provides relative location and mean-reversion context.

• Long bias: price at or below the midline.
• Short bias: price at or above the midline.

CCI

CCI supplies timing confirmation around oversold and overbought transitions.

• Long bias: CCI crosses above the oversold threshold.
• Short bias: CCI crosses below the overbought threshold.

SuperTrend

SuperTrend acts as a regime filter.

• Long bias: bullish regime.
• Short bias: bearish regime.

VIDYA

VIDYA contributes adaptive trend confirmation.

• Long bias: price above VIDYA with a rising VIDYA slope.
• Short bias: price below VIDYA with a falling VIDYA slope.

ADX / DI

ADX is used as an optional trend-strength gate.

• Entry permission is conditioned on ADX meeting or exceeding the configured threshold.

Confluence Model

Each enabled factor contributes one point to the active long or short score. Long and short eligibility are evaluated separately, allowing the engine to measure directional agreement without collapsing all logic into a single composite state.

A setup becomes eligible when its directional score meets or exceeds the minimum score threshold. Because evaluation is based on the most recently closed bar, the resulting signal state is non-repainting by design. tilburgsciencehub

Risk Engine

Risk logic is ATR-based throughout, allowing the model to adapt to changing volatility conditions.

Core formulas

• Stop-loss: SL = entry ± ATR × sl_mult
• Take-profit: TP = entry ± ATR × tp_mult
• Break-even trigger: entry ± ATR × be_mult
• Trailing distance: ATR × trail_mult

Exit sequence

The exit stack is evaluated in a defined order:

1. Hard stop-loss.
2. Partial profit event, when enabled.
3. Break-even activation and trailing transition.
4. Signal-based exit, including channel re-entry or opposite confluence.

Exit checks use bar high and low values to detect threshold interaction with the active position state.

State Model

Each symbol maintains its own signal state, including:

• Position side: long, short, or flat.
• Entry price.
• Entry quantity snapshot.
• Stop, target, and trailing values.
• Daily P&L buckets for 1x, 2x, and 3x tiers.
• Daily win-rate state.
• Daily cap status.
• Last bar timestamp.
• Order sequence reference.

This state model preserves signal continuity across bars and supports deterministic interpretation of entries, exits, daily caps, and reset conditions.

Daily Cap Logic

The engine tracks separate daily P&L buckets by contract tier:

• 1x bucket.
• 2x bucket.
• 3x bucket.

When a configured cap is reached, the engine emits a CAP_HIT event, disables new entries for the remainder of the session, and can optionally apply cascade blocking across higher tiers. This keeps the signal model aligned with tier-specific daily risk governance.

Alert Model

The engine emits structured alerts for the following event classes:

• Long entry.
• Short entry.
• Stop-loss exit.
• Take-profit exit.
• Break-even or trailing exit.
• Signal-based exit.
• Daily cap event.
• Session reset.

Output channels

Alert payloads can include symbol, timeframe, entry or exit price, realized P&L, tier size, win-rate context, exit reason, timestamp, and strategy version.

Design Boundaries

What the engine is

AS-KBN-Futures is a closed-bar, multi-factor signal engine built around:

• Institutional bias.
• Momentum confirmation.
• Market structure context.
• Trend and volatility filters.
• ATR-based risk control.
• Daily P&L governance.
• Multi-channel alert delivery.

What the engine is not

The engine is not positioned as a brokerage execution layer or account-management service. Its primary role is signal generation, stateful event evaluation, alert publishing, and research-grade strategy expression.

Positioning

AS-KBN-Futures is designed for:

• Human-in-the-loop decision support.
• Alert-driven market monitoring.
• Structured downstream ingestion.
• Strategy research and historical evaluation.