AI Trader Configuration Guide
Complete configuration templates, expected outcomes, constraints, and Operator Context notes for deploying the AI Trader under different market conditions and account sizes.
The AI Trader requires a model with strong reasoning, instruction adherence, and JSON output reliability. A single hallucinated Stop Loss or ignored constraint translates directly to lost capital.
Model Tier Matrix
| Tier | Model Type | Examples | Suitability | Notes |
|---|---|---|---|---|
| 🟢 Tier 1 | Pro / Flagship | GLM 5.1, Mimo v2 Pro, MiniMax 2.7 | ✅ Excellent | Best reasoning, longest attention span, lowest hallucination rate |
| 🟡 Tier 2 | Coder | GPT Codex 5.4 | ⚠️ Good (with caveats) | Perfect JSON output, but may lack deep market intuition for SMC/ICT concepts |
| 🔴 Tier 3 | Flash / Lite | Gemini 3.1 Flash Lite | ❌ Avoid | Too fast/hasty. Skips negative constraints. Struggles with R:R math |
Best practice: Use a Tier 1 model for the core AI Decision loop. If you want to reduce API costs, use a Tier 3 model only for background tasks like Sentinel summarization — never for buy/sell decisions.
Configuration Presets
Hyper-Sniper Capital Preservation
Goal: Protect a micro account at all costs. Only trade when the setup is structurally perfect.
Best for: Accounts under $50, beginners, or extremely risk-averse operators.
Market & Scanning
| Timeframe | H1 or H4 |
| Scan Mode | Smart Scan |
| Scan Interval | 900 sec (15 min) |
| Re-entry Cooldown | 240 min (4 hours) |
Trade Identity
| Trading Style | Swing / Position |
| Entry Profile | Conservative |
Risk Management
| Risk Per Trade | 2.0% – 5.0% |
| Max Positions | 1 |
| Max Per Symbol | 1 |
| Min Confidence | 90 - 95 |
Trade Validation
| Validate R:R | ON |
| Min R:R Ratio | 3.0 |
| Max Daily Trades | 2 |
| Max Daily Loss | 10% |
Trailing Stop
| Enabled | ON |
| Mode | ATR Dynamic |
| ATR Multiplier | 2.0 |
Services
| Sentinel | ON |
Smart Balanced Recommended Default
Goal: Balanced growth with adaptive style selection. The AI picks the best approach per trade.
Best for: Accounts $50-$5,000, intermediate operators who want hands-off automation.
Market & Scanning
| Timeframe | M15 or H1 |
| Scan Mode | Smart Scan |
| Scan Interval | 120 - 300 sec |
| Re-entry Cooldown | 60 min |
Trade Identity
| Trading Style | Auto / Smart |
| Entry Profile | Balanced |
Risk Management
| Risk Per Trade | 1.0% - 2.0% |
| Max Positions | 3 |
| Max Per Symbol | 1 |
| Min Confidence | 75 |
Trade Validation
| Validate R:R | ON |
| Min R:R Ratio | 2.0 |
| Max Daily Trades | 10 |
| Max Daily Loss | 5% |
Trailing Stop
| Enabled | ON |
| Mode | ATR Dynamic |
| ATR Multiplier | 1.5 |
Services
| Sentinel | ON |
Aggressive Compounding High Risk
Goal: Rapid account multiplication through high-leverage percentage-based compounding.
Best for: Accounts under $50 where the operator accepts full loss as a possibility.
This preset has a high probability of account wipeout in choppy markets. The upside is exponential growth if the AI catches a strong trend early. Use on expendable micro accounts only.
Market & Scanning
| Timeframe | M15 or H1 |
| Scan Mode | Smart Scan |
| Scan Interval | 120 - 300 sec |
| Re-entry Cooldown | 0 or 15 min |
Trade Identity
| Trading Style | Intraday |
| Entry Profile | Aggressive |
Risk Management
| Risk Per Trade | 10% - 20% |
| Max Positions | 2 - 3 |
| Max Per Symbol | 2 |
| Min Confidence | 65 - 70 |
Trade Validation
| Validate R:R | ON |
| Min R:R Ratio | 2.0 |
| Max Daily Trades | 15 |
| Max Daily Loss | 30% |
Trailing Stop
| Enabled | ON |
| Mode | ATR Dynamic |
| ATR Multiplier | 1.5 |
Services
| Sentinel | ON |
Operator Context Templates
The Operator Context is injected directly into the AI prompt on every scan cycle. It overrides default behavior without changing code. Always keep Trading Bias = HTF Bias Only to prevent trading against the macro trend.
Sniper Mode Note
CRITICAL DIRECTIVE: You are managing a hyper-micro account with extremely limited margin. Your absolute priority is capital preservation. You must act as a Tier-1 Sniper. Do NOT trade chop. Do NOT trade the middle of ranges. You are ONLY permitted to execute OPEN_TRADE if price has swept major H4/D1 liquidity and reached an extreme Premium/Discount structural zone with a confirmed reversal pattern. If the setup is not 99% perfect, you MUST return SKIP. Accuracy over frequency.Balanced Mode Note
Standard autonomous mode. Follow all system constraints. Prioritize high-quality setups aligned with the Sentinel HTF bias. Prefer trades where multiple confluences align (structure + momentum + liquidity). Do not force entries during low-volatility sessions.Aggressive Compounding Note
CRITICAL DIRECTIVE: You are executing a high-leverage compounding protocol on a highly constrained margin account. Your objective is aggressive, asymmetric capital growth. Do NOT wait for perfect multi-day confirmations; attack strong H1/H4 momentum immediately. If you catch a strong trend, you are encouraged to scale into the winner on shallow pullbacks. HOWEVER, your defense must be ruthless: if momentum stalls or rejects near a take-profit target, do NOT hesitate to use MODIFY_TRADE to drag your Stop Loss entirely into profit or CLOSE_TRADE to secure the bag. Never let a green trade turn red.News Event Lockdown
HIGH-IMPACT NEWS EVENT ACTIVE. Do NOT open any new trades. Focus exclusively on managing existing positions — tighten SL to breakeven or lock in profit if positions are green. If a position is at risk from the news spike, close it preemptively. Resume normal operations after this context expires.Hard Constraints & Safety Rails
These constraints are hard-coded and cannot be overridden by Operator Context or configuration. They act as the final safety net.
If unsure, AI must return SKIP
If any trade has SL=0 and TP=0, AI must immediately MODIFY_TRADE
BUY: SL < Entry < TP. SELL: TP < Entry < SL. Violating = auto rejection
If bid, ask, ATR, or S/R data missing → SKIP with DATA_INVALID
Must not trade against strong H4/D1/W1 Sentinel Bias
Wait ≥3 full candles before re-entry
Wait ≥2 full candles after trailing-stop close
Wait ≥1 full candle after TP close
After 2+ losses on same symbol: requires HIGH quality + confidence ≥ min + 10
Must close existing position first, then consider OPEN_TRADE on later scan
Must not panic-close swing/intraday trades during minor pullbacks if H1/H4 structure intact
Avoid entries directly into opposite-side fresh FVG/sweep rejection unless structure reclaims
Self-Improvement Behavior Matrix
The AI's memory feedback loop dynamically adjusts behavior based on tracked performance. This operates independently of your configuration.
| Trigger | AI Injects | Behavior Change |
|---|---|---|
| 3+ Loss Streak | [CAUTION] LOSING STREAK | Raises confidence bar. Overrides even aggressive Operator Notes |
| 5+ Win Streak | [GOOD] HOT STREAK | Prevents reckless scaling. Keeps discipline intact |
| Style P/L < 0 | [CAUTION] AVOID style(s) | Deprioritizes failing styles |
| Style P/L > 0 | [GOOD] PREFER style(s) | Leans into what is currently working |
| Recent Close (same symbol) | RE-ENTRY GUARD | Prevents emotional revenge re-entry |
| Cross-Symbol Losses | Shows recent losses from other instruments | Gives portfolio-wide awareness |
Critical insight: The self-improvement loop acts as an autonomous "circuit breaker." Even in the most aggressive compounding setup, if the AI detects consistent losses, it will override the Operator Context and start returning SKIP to preserve remaining capital.
Expected Results & Trade Frequency
These are estimates based on architectural behavior, not guaranteed financial outcomes. Past performance does not guarantee future results. Trading involves substantial risk of loss.
| Metric | Sniper (A) | Balanced (B) | Compounding (C) |
|---|---|---|---|
| Trades/Day | 0-1 (often 0) | 2-6 | 5-15 |
| Days Without Trade | 3-7 days common | Rare | Almost never |
| Win Rate | 60-75% | 50-65% | 40-55% |
| Avg R:R | 3.0+ | 2.0-2.5 | 2.0 |
| Drawdown Risk | Very Low (5-15%) | Moderate (10-25%) | Very High (30-100%) |
| Growth Curve | Slow, steady, linear | Moderate, steady | Exponential or bust |
| Time to Double | Weeks to months | 1-4 weeks | Days (if survives) |
| Wipeout Probability | Very Low (<5%) | Low (5-15%) | High (40-60%) |
| Ideal Account | Any (especially micro) | $50-$5,000+ | Under $50 (expendable) |
| API Cost/Day | $0.05-$0.20 | $0.20-$1.00 | $0.50-$2.00 |
Micro Account Considerations
When trading with equity under $50 (e.g., Rp 125.000):
reasoning field to understand why it entered.