AUTHOR: Tony Mudau

Trading Concepts and Definitions Guide

This document explains the core trading concepts used by the Agent Force 1 pipeline, how they are applied in this codebase, and how they affect entry and exit decisions.

1) End-to-End Trading Flow (Conceptual)

The system follows a layered decision path:

1. Symbol evaluation (orchestration + technical + market)
2. Entry quality gating (RR, EV, range/chasing, confidence, volatility regime)
3. Cross-symbol ranking (score + diversification penalty)
4. Portfolio/risk shaping (lot sizing, correlation, exposure, hard caps)
5. Execution guard checks (stale bars, spread, slippage, volatility spikes)
6. Order execution or scheduling
7. Exit management (time-stop, lifecycle, technical/market exit logic, optional hard TP)
8. Learning/adaptation (session/symbol/pattern feedback loops)

Each layer is intentionally defensive. A trade must survive all relevant layers.

2) Core Strategy Concepts

2.1 Directional Context

• Trend: broad directional state inferred from indicators (e.g., bullish/bearish/neutral).
• Momentum (MACD signal): confirms whether direction has momentum support.
• RSI state: helps detect exhaustion or stretch (e.g., overbought/oversold).

Direction is not purely one-indicator driven. The system combines these to avoid weak signals.

2.2 Multi-Timeframe (MTF) Alignment

• Entry timeframe context is cross-checked against higher timeframes (H1/D1).
• Alignment is scored, not only blocked.
• Misalignment can reduce confidence and ranking quality.

This reduces trades that look good on one timeframe but conflict with broader structure.

2.3 Volatility Regimes

• Compression: low-volatility/squeezed environment.
• Normal: typical environment.
• Expansion: high-volatility regime.

Regime affects strategy validity:

• Trend momentum trades are usually poor in compression.
• Expansion can increase noise/slippage risk if not handled carefully.

2.4 Strategy Mode

The technical layer can classify setups as:

• trend
• htf_pullback
• mean_reversion

Modes are filtered by session, volatility regime, and context quality.

3) Entry Quality Concepts

3.1 Risk-Reward Ratio (RR)

RR = reward / risk, where:

• risk = distance from entry to stop-loss
• reward = distance from entry to take-profit

Higher RR improves expectancy, but only if win rate remains viable.

3.2 Expected Value (EV) in R Units

The system approximates expectancy as:

EV_per_R = w * RR - (1 - w)

Where:

• w = estimated win rate from pattern memory
• RR = reward-to-risk ratio

Interpretation:

• EV > 0: favorable on average
• EV < 0: unfavorable on average

EV is used both as a hard gate and as a scoring/penalty input.

3.3 Confidence (Raw vs Effective)

• Raw confidence: initial signal confidence from technical synthesis.
• Effective confidence: raw confidence adjusted by:
  • MTF conflict penalties
  • reversal/exhaustion signals
  • EV blending
  • dynamic confidence floors (session adaptation)

Effective confidence is the more realistic confidence used in stricter gating.

3.4 Entry Microstructure Filters

Before execution, entries are validated against:

• minimum RR
• too close to resistance/support
• chasing near range extremes
• excessive deviation from current price

These filters remove common bad entries caused by late chasing or stale proposals.

4) Execution Quality Concepts

Execution guards are final “last-mile” quality checks:

• Stale bars: reject if market context is too old for timeframe.
• Technical disagreement: optional micro-trend mismatch rejection.
• Excessive slippage vs proposal entry: reject if execution drift is too high.
• Spread too wide: reject poor liquidity moments.
• Volatility spike: reject abnormal single-bar extension conditions.

Even good ideas can become bad executions if this layer is ignored.

5) Symbol-Specific Profiles

Global thresholds are often wrong across instruments (e.g., EURUSD vs XAUUSD). To address that, the system supports symbol-specific trade profiles with fallback to DEFAULT.

Supported fields:

• entry_min_rr
• entry_max_deviation_pct
• ev_min_per_r
• exec_max_slippage_pct
• exec_max_spread_ratio

Conceptually:

• FX majors: tighter slippage/deviation and tighter spreads.
• Metals/volatile symbols: wider execution tolerances but often stricter EV/RR requirements.

6) Diversification and Concentration Control

6.1 Concentration Bias Problem

A pure “highest score wins” selector can over-focus one symbol/theme repeatedly.

6.2 Diversified Ranking

The selector now adjusts candidate score by a concentration penalty based on:

• same symbol already open
• shared currency legs with existing open positions
• recent repetition in trade history

This reduces over-clustering while still favoring high-quality setups.

7) Risk Management Concepts

7.1 Per-Trade Risk Budget

Position size is derived from:

• account equity
• max risk % per trade
• stop-loss distance in account-currency terms

7.2 Risk per Lot Estimation

Primary method:

• broker-native profit/risk calculation (order_calc_profit)

Fallback method:

• symbol metadata aware estimate (tick value/size, point/contract size)

This avoids major sizing errors on non-FX products.

7.3 Correlation and Exposure Scaling

Lot size is scaled down when:

• correlated same-theme positions accumulate
• book concentration in relevant currencies is high

This helps prevent hidden portfolio-level risk stacking.

8) Exit Management Concepts

The system does not only optimize entries; it manages exits intentionally:

• Time-stop logic: closes stale/underperforming positions after age thresholds.
• Lifecycle actions: breakeven and trailing logic as position evolves.
• Hard take-profit (optional): force-close when account-currency profit threshold is hit.
• LLM/technical/market exit context: additional close guidance.

Good exits are essential for realized expectancy, not just theoretical setup quality.

9) Adaptation and Learning Concepts

Post-trade learning updates behavior over time:

• Symbol adaptation: pause/scale down weak symbols.
• Pattern cooldown: stop repeating failing pattern fingerprints for a cooldown period.
• Session confidence floor: raise required confidence when a session underperforms.

This creates a closed-loop system that reacts to recent regime/performance changes.

10) Key Definitions (Quick Reference)

• SL (Stop-Loss): predefined invalidation price to cap downside.
• TP (Take-Profit): predefined target price to realize gains.
• R: one unit of risk (distance from entry to SL).
• RR: reward-to-risk ratio in price terms.
• EV per R: expected return measured in risk units.
• MTF alignment: agreement between entry and higher timeframe directions.
• Compression: low-volatility regime; often poor for momentum continuation.
• Expansion: elevated-volatility regime; can increase opportunity and execution risk.
• Slippage: difference between intended and actual executable price.
• Spread ratio: spread relative to instrument mid-price (liquidity quality proxy).
• Concentration bias: repeated overexposure to one symbol/theme.
• Diversification penalty: score reduction to avoid clustered exposure.
• Execution guard: final pre-order safety filters.

11) Practical Configuration Principles

1. Keep DEFAULT conservative and broadly safe.
2. Override only symbols with materially different behavior.
3. Tighten thresholds first for frequent slippage/spread offenders.
4. Raise EV/RR thresholds for noisy or whipsaw-prone symbols.
5. Reassess profiles after every meaningful sample window (e.g., 50-100 closed trades per symbol).

12) What “Good Trade Quality” Means in This System

A high-quality trade in this architecture is one that:

• has valid directional and MTF context,
• passes EV and RR constraints,
• is not a chase near local extremes,
• is executable under acceptable spread/slippage conditions,
• does not overly concentrate the current book,
• is sized to a controlled risk budget,
• and has a managed exit path.

This is the operational definition of disciplined, intentional trading used here.