AbacuQuant Works on Any Market — Here Are the Set Files to Prove It

AbacuQuant 4.0 — Complete User Manual & Setup Guide The definitive reference for installation, configuration, prop-firm compliance, local AI setup, portfolio analysis, and live deployment. LIVE SIGNAL: CLICK HERE

Table of Contents

1. Introduction — What's New in v4.0
2. System Requirements
3. Installation and Quick Start
4. Complete Input Parameter Reference
5. Preset Configurations (Retail and Prop Firm)
6. Cloud AI Setup — OpenAI, Gemini, DeepSeek
7. Ollama Local AI Setup — Complete Guide
8. Multi-EA Portfolio Scorer
9. Native Calendar News Filter
10. Equity Trailing Drawdown and Auto Scaling
11. Strategy Reference
12. Risk Management — 12 Protection Layers
13. Backtesting and Forward Testing Protocol
14. Live Deployment Best Practices
15. Troubleshooting
16. FAQ and Glossary

1. Introduction — What's New in v4.0

AbacuQuant is an institutional-grade Expert Advisor for MetaTrader 5 designed for systematic, multi-strategy, multi-instrument algorithmic trading. The system combines ten independent technical strategies, twelve risk management layers, real artificial intelligence integration, and full compliance with major prop firm rules — all delivered as a single MQL5-compliant binary with zero DLL dependencies.

Version 4.0 is a major release that transforms AbacuQuant from a self-contained algorithmic system into a complete institutional platform. The release preserves every feature from previous versions and adds four major capabilities:

New Capability	Description	Best For
Prop Firm Lockdown Presets	8 preset profiles for FTMO, FundedNext, The5ers, FundingPips, Goat Funded Trader. Forces rule compliance through hard input overrides.	Funded account traders
Portfolio Scorer	Native correlation matrix, coverage analysis, asset diversity scoring. Composite grade A+ to F.	Multi-EA portfolio managers
Ollama Local AI	Run open-source LLMs locally for signal confirmation. No API costs, no data sharing, fully offline, zero DLL.	Privacy-conscious traders
Native Calendar Filter	Replaces external news APIs with MetaTrader 5's built-in economic calendar. Zero configuration.	News-event-aware traders
Equity Trailing DD	Tracks peak equity, blocks trading when drawdown exceeds threshold from all-time high.	Compounding accounts
Auto Lot Scaling	Lot size grows automatically as the account compounds, with broker-validated steps.	Long-term operators
ATR Stop Loss Fix	Resolves a critical edge case where ATR-mode stops could be removed under low volatility. Now always server-visible.	All ATR-mode users

---

2. System Requirements

• Platform: MetaTrader 5, build 4410 or later, Hedging or Netting account
• OS: Windows 10/11 native, macOS via Parallels, Linux via Wine
• Broker: ECN or Raw Spread strongly recommended, leverage 1:100+ ($500 min) or 1:500 ($200 min)
• Capital: $200 single EA testing, $500 single EA live, $5,000 full portfolio
• For Cloud AI: API key from OpenAI, Gemini, or DeepSeek (paid)
• For Ollama: Free Ollama install, 4-8 GB RAM, 2-5 GB disk space
• VPS: Strongly recommended for 24/7 operation, 4 GB RAM minimum (8 GB if running Ollama)

---

3. Installation and Quick Start

Installation

1. After purchase, open MetaTrader 5 logged into your MQL5 account
2. Press F4 to open MetaEditor
3. In Navigator, expand Market, right-click AbacuQuant, select Download
4. Return to terminal. In Navigator (Ctrl+N), AbacuQuant appears under Expert Advisors
5. Open an H1 chart, drag AbacuQuant onto it
6. Confirm Allow Algo Trading is active in the toolbar
7. If using AI features, add WebRequest URLs in Tools → Options → Expert Advisors

Quick Start (5 minutes)

1. Attach AbacuQuant to EURUSD H1 chart
2. Set UsePresetConfig = PRESET_CONSERVATIVE
3. Leave defaults, click OK
4. Verify Experts log shows initialization success
5. Run on demo for at least 7 days before going live

For specific instruments, load the validated set files included with purchase. Format: SYMBOL_DIRECTION_REPORTID.set (example: EUR_L_7472.set). In Inputs tab, click Load and select the .set file.

---

4. Complete Input Parameter Reference

4.1 Main Settings

Input	Description
UsePresetConfig	Operating profile. 11 options: Custom, 3 retail profiles, 8 prop-firm lockdowns. See Section 5.
MagicNumberBase	Default 1234567. Use different base per instance to separate positions.
USABrokerCompliance	Activates FIFO, hedging restrictions, and leverage caps for US-regulated brokers.
MaxOpenTradesPerSymbol	Hard cap per symbol across strategies. 0 = disabled. Prop firm presets force 1.
MaxOpenTradesPerMagic	Hard cap per magic. 0 = disabled. Prop firm presets force 1.
SignalTimeframe	Default H1. All validated set files use H1.
TradeDirectionAllowed	Long only, short only, or both.
ShowDashboard	On-chart performance dashboard with live metrics.
PauseEA	Soft pause: stops new trades, manages existing.
EmergencyCloseAll	Closes all EA positions and removes itself.

4.2 Cloud AI Integration

Input	Description
EnableExternalAI	Master switch for cloud AI.
AI_Provider	OpenAI, Gemini, DeepSeek, Custom, or None.
AI_API_Key	Your provider API key.
AI_Endpoint	Provider endpoint URL.
AI_Model	Model name (gpt-4o-mini, gemini-2.0-flash-lite-001, deepseek-chat).
AI_MaxTokens	Default 500. Lower reduces cost.
AI_ConfirmSignals	True = AI gates every trade. False = audit only.
AI_ShowAnalysisInHTML	Generates HTML report in MQL5/Files.
AI_ConfirmationTimeout	Milliseconds to wait, default 15000.
AI_ConfirmationPrompt	Base instruction sent with each signal.

4.3 Ollama Local LLM

Input	Description
EnableOllama	Master switch for local AI.
Ollama_Endpoint	Default http://127.0.0.1:11434/api/generate
Ollama_Model	Must match exactly a pulled model (e.g., llama3.2:3b).
Ollama_TimeoutMs	Default 30000. Increase for CPU-only systems.
Ollama_MaxTokens	Default 200.
Ollama_Temperature	Default 0.2 for deterministic decisions.
Ollama_ConfirmSignals	True = blocks each trade until response. False = audit.
Ollama_ShowPopup	On-chart popup with analysis result.
Ollama_PreferOverCloud	When both AIs active, prioritizes Ollama.

4.4 Portfolio Scorer

Input	Description
EnablePortfolioScorer	Activates CSV export and cross-EA correlation analysis.
PortfolioCSVPrefix	Default ABQ. Must match across all instances.
PortfolioCorrThreshold	Default 0.35. Pairs above this are flagged as redundant.
PortfolioShowPopup	On-chart score popup.
PortfolioRefreshHours	Default 6. How often to recompute live.

4.5 News Filter, Trailing DD, Auto Scaling

Input	Description
EnableNativeNewsFilter	MT5 native economic calendar filter.
News_BlockMinutesBefore/After	Window around high-impact events. Default 15/15.
News_BlockHighImpactOnly	True = only red events block.
News_CloseOpenTrades	Close positions before event.
News_FilteredCurrencies	Default USD,EUR,GBP,JPY. Empty = all.
EnableEquityTrailingDD	Activates trailing DD lock.
TrailingDD_Percent	Default 5%. Distance from peak that triggers.
TrailingDD_CloseAll / PauseEA	Response when triggered.
EnableAutoScaling	Grows lots as account compounds.
AutoScaling_StepPercent	Default 10%. Each step growth triggers multiplier increase.
AutoScaling_LotIncrement	Default 0.10. Multiplier added per step.
AutoScaling_MaxMultiplier	Default 3.0. Hard cap.

4.6 Risk and Lot Management

Input	Description
LotSizingMethod	Proportional balance or fixed risk percent.
RiskPercentPerTrade	Default 0.5. Prop firm presets force 0.25-0.5.
InitialBalanceForLotCalc	Reference balance for proportional method.
LotAtInitialBalance	Lot at reference balance.
ProfitManagementMode	Pips, group percent, or ATR-based.
TakeProfit / StopLoss	Pips, used in pip mode. Default 50/25.
AtrTakeProfit/StopLossMultiplier	Multipliers in ATR mode. Default 2.0/1.5.
EnableTrailingStop	Activates trailing SL.
TrailingStopPips / TrailingStartPips	Default 10/15.
EnableBreakEven	Moves SL to entry plus buffer at threshold.
BreakEvenPipsProfit / SecurePips	Default 15/2.
EnablePartialClose	Closes portion at profit threshold.
MaxConsecutiveLosses	Pause after N losses. 0 = disabled.
EquityStopPercentage	Default 8%. Hard stop on equity loss.
MaxDrawdownPercent	Max DD from initial balance. Default 8%.
CloseAllOnMaxDrawdown	Closes all positions on DD limit.
UseVirtualStopLoss	Hidden internal SL.
EnableDailyPLLimit	Daily caps on profit/loss.
DailyProfitTargetPercent / DailyLossLimitPercent	Daily caps. Default 0/3.
EnableMaxSpreadFilter / MaxSpreadPips	Default 3 pips. Adjust per instrument.
EnableMaxLoadFilter	Caps margin exposure.
MaxEquityLoadPercent / MaxDepositLoadPercent	Default 5/0. Margin limits.

4.7 Strategy Inputs

Each of the 10 strategies has its own input group with at least an enable toggle and signal group ID. Strategies sharing the same group ID combine into confluence — a trade only opens when all strategies in the group agree. Groups: Imbalance, Fibonacci, Candle Patterns, Support/Resistance, Bollinger Bands, RSI, EMA, MACD, ADX, ATR Volatility, Stochastic. See Section 11 for details.

---

5. Preset Configurations

5.1 Retail Presets

Preset	Risk	Max DD	Grid	Recovery	Best For
PRESET_CUSTOM	Manual	Manual	Manual	Manual	Power users with own .set files
PRESET_CONSERVATIVE	0.25%	10%	Off	Off	Validation, beginners
PRESET_BALANCED	0.5%	15%	Off	On (1.1x)	Standard retail
PRESET_AGGRESSIVE	1.0%	20%	On	On (1.2x)	Experienced traders

5.2 Prop Firm Presets — Hard Lockdowns

Prop firm presets enforce strict compliance with the rules of each evaluation program. When activated, the EA forcibly disables grid trading, recovery mode, and volatility adaptation; sets one position per symbol and one per magic; enforces visible server-side stop losses; and applies a 30% safety buffer below the firm's published drawdown limits.

Preset	Firm Daily DD	Firm Total DD	EA Daily Cap	EA Total Cap	Risk/Trade
PRESET_FTMO_CHALLENGE	5%	10%	3.5%	7%	0.5%
PRESET_FTMO_FUNDED	5%	10%	3.5%	7%	0.5%
PRESET_FUNDEDNEXT_2STEP	5%	10%	3.5%	7%	0.5%
PRESET_FUNDEDNEXT_1STEP	3%	6%	2.1%	4.2%	0.3%
PRESET_THE5ERS_HIGHSTAKES	4%	5%	2.8%	3.5%	0.25%
PRESET_FUNDINGPIPS_2STEP	5%	10%	3.5%	7%	0.5%
PRESET_FUNDINGPIPS_2STEP_PRO	3%	6%	2.1%	4.2%	0.3%
PRESET_GOAT_FUNDED	5%	10%	3.5%	7%	0.5%

When a prop firm preset activates, the Experts tab prints a banner confirming all enforced rules. Always verify the banner matches your account before going live. Rules verified against each firm's official documentation as of April 2026.

Important: Prop firm presets cannot be partially overridden. Manual changes to locked parameters are reset on every initialization. This is by design to prevent accidental rule breaches.

---

6. Cloud AI Setup — OpenAI, Gemini, DeepSeek

6.1 General WebRequest Configuration

Before using any cloud AI, allow MetaTrader 5 to make web requests:

1. Tools → Options → Expert Advisors
2. Check Allow WebRequest for listed URL
3. Add the URLs you intend to use:
   • OpenAI: https://api.openai.com
   • Gemini: https://generativelanguage.googleapis.com
   • DeepSeek: https://api.deepseek.com
4. Click OK

6.2 OpenAI Setup

1. Visit platform.openai.com, create account
2. Add payment under Settings → Billing (minimum $5 credit)
3. Go to API Keys → Create new secret key, copy it
4. Configure: EnableExternalAI=true, AI_Provider=AI_PROVIDER_OPENAI, paste key, AI_Endpoint=https://api.openai.com/v1/chat/completions, AI_Model=gpt-4o-mini (cheap) or gpt-4o (better)

Cost estimate: ~$0.001-0.005 per signal with gpt-4o-mini. 50 signals/day costs $1.50-$7.50/month.

6.3 Google Gemini Setup

1. Visit aistudio.google.com, sign in with Google account
2. Click Get API key → Create API key
3. Configure: EnableExternalAI=true, AI_Provider=AI_PROVIDER_GEMINI, paste key, AI_Model=gemini-2.0-flash-lite-001 (fast/cheap) or gemini-2.0-flash-001 (better)

Cost estimate: Generous free tier. Most retail traders never exceed it.

6.4 DeepSeek Setup

1. Visit platform.deepseek.com, create account
2. Add payment or use free credits, create API key
3. Configure: EnableExternalAI=true, AI_Provider=AI_PROVIDER_DEEPSEEK, paste key, AI_Endpoint=https://api.deepseek.com/v1/chat/completions, AI_Model=deepseek-chat

Cost estimate: Among the cheapest providers. Off-peak pricing offers significant additional discounts.

6.5 Operating Modes

• Audit mode (AI_ConfirmSignals=false): EA places trades, AI analyzes after for review
• Filter mode (AI_ConfirmSignals=true): AI must CONFIRM each trade before execution. Fail-safe: API failure allows trade.

---

7. Ollama Local AI Setup — Complete Guide

Ollama is an open-source local LLM server that runs language models on your own hardware. Using Ollama means every signal analysis happens locally — no API keys, no usage costs, no rate limits, no data leaves your machine.

7.1 Hardware Requirements

• RAM: 8 GB minimum (4 for model + 4 for OS/MT5), 16 GB recommended
• Disk: 2-8 GB free for model files
• CPU: Modern Intel i5 or AMD Ryzen 5 (2018+) sufficient for 3B models
• GPU (optional): NVIDIA with 4+ GB VRAM provides 5-10x speedup
• OS: Windows 10/11, macOS 11+, Linux

7.2 Step 1 — Install Ollama

1. Visit ollama.com
2. Click Download, select your OS
3. Run the installer:
   • Windows: Double-click .exe, follow prompts. Installs as background service.
   • macOS: Open .dmg, drag to Applications, launch once
   • Linux: Run the curl install command, installs as systemd service
4. Ollama auto-starts on port 11434

7.3 Step 2 — Verify Ollama Is Running

Open command prompt (Windows: Win+R, cmd) or terminal:

ollama --version

Should show ollama version 0.5.x. Then test the server:

curl http://127.0.0.1:11434

Should return Ollama is running. If "connection refused", start it: open Ollama app (Win/Mac) or sudo systemctl start ollama (Linux).

7.4 Step 3 — Pull a Model

Model	Size	RAM	Speed	Quality	Best For
llama3.2:3b	2 GB	4 GB	Fast	Good	Most users (recommended)
llama3.2:1b	1.3 GB	2 GB	Very fast	Acceptable	Low-RAM systems
llama3.1:8b	4.9 GB	8 GB	Medium	Excellent	Higher-end systems
qwen2.5:7b	4.7 GB	8 GB	Medium	Excellent	Strong reasoning
mistral:7b	4.1 GB	8 GB	Medium	Excellent	General-purpose

Pull the model:

ollama pull llama3.2:3b

Verify with ollama list .

7.5 Step 4 — Test Manually

ollama run llama3.2:3b

Type a test: Should I buy EURUSD if RSI is at 25 and trend is bullish? Answer CONFIRM or REJECT and brief reason.

The model should respond within seconds. Type /bye to exit.

7.6 Step 5 — Allow MT5 WebRequest

1. MetaTrader 5 → Tools → Options → Expert Advisors
2. Check Allow WebRequest for listed URL
3. Add http://127.0.0.1:11434
4. Optionally also add http://localhost:11434
5. Click OK

Important: Format must be exact — protocol, host, port. No trailing slashes or paths.

7.7 Step 6 — Configure AbacuQuant

1. EnableOllama = true
2. Ollama_Endpoint = http://127.0.0.1:11434/api/generate (default)
3. Ollama_Model = llama3.2:3b (must match exactly what you pulled)
4. Ollama_TimeoutMs = 30000 for CPU, 15000 for GPU
5. Ollama_Temperature = 0.2 for deterministic decisions
6. Choose mode: audit (Ollama_ConfirmSignals=false) or filter (true)

7.8 Step 7 — Verify Integration

On EA initialization with Ollama enabled, you should see:

• Green popup "OLLAMA OFFLINE LLM ONLINE" with endpoint and model
• "ok.wav" sound
• Experts tab confirmation log

7.9 Common Errors

Error	Cause	Solution
WebRequest 4014	URL not allowed	Add to Tools → Options → Expert Advisors
5203 / Connection refused	Ollama not running	Start Ollama app or sudo systemctl start ollama
HTTP 404	Wrong model name	Run ollama list , match exactly
Timeout	Model slow on CPU	Increase Ollama_TimeoutMs to 60000, use smaller model
Empty response	JSON parsing or model output issue	Lower temperature, try different model

7.10 Performance Tuning

• Keep Ollama running: Auto-start with OS
• Pre-load model: First query loads from disk; subsequent are faster
• Use GPU if available: NVIDIA detected automatically
• Audit mode for production: Doesn't delay trades
• VPS: Install Ollama on same VPS as MT5

7.11 Remote Ollama Server

To run Ollama on a different machine:

1. On Ollama machine, set environment variable OLLAMA_HOST=0.0.0.0:11434 before starting
2. Note the LAN IP (e.g., 192.168.1.50)
3. Set Ollama_Endpoint to http://192.168.1.50:11434/api/generate
4. Add this URL to MT5 allowed WebRequest list
5. Verify firewall allows the connection

---

8. Multi-EA Portfolio Scorer

The Portfolio Scorer evaluates the diversification quality of multi-EA setups using three weighted dimensions: correlation (50%), coverage (25%), and asset diversity (25%).

8.1 Why It Matters

Two EAs with 0.85 correlation give roughly the same risk/return as one EA with double the lots. The Portfolio Scorer makes this measurable. Industry-standard threshold: 0.35 absolute correlation.

8.2 Setup

1. On every AbacuQuant instance: EnablePortfolioScorer=true, same PortfolioCSVPrefix (default "ABQ")
2. Run individual backtests. Each backtest exports daily P&L to CSV in MQL5/Files/
3. CSV format: PREFIX_SYMBOL_MAGIC.csv (example: ABQ_EURUSD_1234567.csv)
4. After 2+ backtests, attach any AbacuQuant to a chart with EnablePortfolioScorer=true. On init, EA discovers all matching CSVs and computes analysis.

8.3 Score Interpretation

Score	Grade	Interpretation
90-100	A+	Excellent diversification, low correlations
80-89	A	Very good portfolio quality
70-79	B	Acceptable, some redundancy
60-69	C	Notable redundancy — consider replacing
50-59	D	Poor diversification
Below 50	F	Effectively running same strategy multiple times

8.4 Output Locations

• Experts tab: Full correlation matrix, coverage by hour and weekday, diversity count, composite score, grade
• On-chart popup: Composite score, grade, EA count, average correlation

8.5 Live Mode

In live trading, every EA recomputes the score every PortfolioRefreshHours (default 6). This catches portfolio drift as market conditions change.

---

9. Native Calendar News Filter

Uses MetaTrader 5's built-in economic calendar (CalendarValueHistory and CalendarEventById functions). No external API setup required.

9.1 Configuration Recommendations

Trading Style	Before	After	Currencies	Close Trades
Funded Account / Conservative	30 min	30 min	USD,EUR,GBP,JPY	true
Standard Retail	15 min	15 min	USD,EUR,GBP,JPY	false
Aggressive News Trader	5 min	0 min	USD only	false

9.2 Limitations

• Calendar unavailable in Strategy Tester (auto-disabled in backtests)
• Data depends on broker's calendar feed
• Throttled to one query per minute for performance

---

10. Equity Trailing Drawdown and Auto Scaling

10.1 Equity Trailing Drawdown

Mirrors trailing drawdown rules used by FundingPips Zero accounts and similar instant-funding programs. Tracks peak account equity. When equity falls more than the configured percent below peak, response triggers.

Example: Account starts $10,000, grows to $12,000 (peak), falls to $11,400. Drawdown from peak = 5%. With TrailingDD_Percent=5.0 and TrailingDD_CloseAll=true, EA closes all positions immediately.

10.2 Auto Lot Scaling

Grows lots automatically as account compounds. Math:

• Growth percent = (current equity − reference) / reference × 100
• Steps = floor(growth percent / AutoScaling_StepPercent)
• Multiplier = 1 + steps × AutoScaling_LotIncrement (capped at AutoScaling_MaxMultiplier)
• Final lot = base lot × multiplier (re-validated against broker min/max/step)

Equity Growth	Multiplier	Effective Lot (base 0.10)
0%	1.00x	0.10
+10%	1.10x	0.11
+50%	1.50x	0.15
+100%	2.00x	0.20
+200%+	3.00x (capped)	0.30

---

11. Strategy Reference

AbacuQuant includes ten independent strategies. Each has its own enable toggle and signal group ID. Strategies sharing a group ID combine into confluence — trade only opens when all agree.

• Imbalance Strategy (ATR-Based): Detects supply/demand imbalances using ATR-relative candle body size. Flagship strategy for catching strong directional moves.
• Fibonacci Strategy: Identifies retracements to key Fibonacci levels (default 50%) with configurable lookback.
• Candle Pattern Strategy: Recognizes 10 patterns: Bullish/Bearish Engulfing, Hammer, Shooting Star, Doji at Level, Morning/Evening Star, Piercing/Dark Cloud, Harami, Tweezer Top/Bottom, Three Soldiers/Crows.
• Support/Resistance Strategy: Identifies dynamic levels from recent highs/lows. Triggers on confirmed breakouts.
• Bollinger Bands Strategy: Mean-reversion using configurable bands. Signals on outer band touches.
• RSI Strategy: Standard overbought/oversold detection. Default 70/30. Best as confluence filter.
• EMA Strategy: Trend-following based on price relationship with moving average. Configurable period, method, applied price.
• MACD Strategy: Momentum-based on signal line crossovers. Standard 12/26/9 default.
• ADX Strategy: Trend strength filter. Triggers when ADX exceeds threshold (default 25).
• ATR Volatility Strategy: Regime filter. Only triggers within configured ATR range.
• Stochastic Strategy: Classic crossover signals with configurable K/D periods.
• MTF Trend Filter (overlay): Requires alignment with higher-timeframe EMA before any entry.

---

12. Risk Management — 12 Protection Layers

Twelve independent risk layers operate simultaneously. Each can be enabled/disabled independently:

1. Static Maximum Drawdown — Hard cap from initial balance. Triggers global stop.
2. Equity Trailing Drawdown — Moves up with each peak. FTMO/FundingPips Zero compatible.
3. Equity Stop Percentage — Absolute equity loss limit.
4. Daily Profit/Loss Limit — Separate caps for daily gain/loss.
5. Max Consecutive Losses Pause — Suspends trading after losing streak.
6. Maximum Spread Filter — Rejects entries when spread exceeds threshold.
7. Account Load Protection — Caps total margin exposure.
8. Volume Limit Validation — Article 2555 broker volume compliance.
9. Stops Level/Freeze Level Validation — Every SL validated against broker minimums. v4.0 ATR fix expands stops to broker minimum instead of removing them.
10. Correlation Filter — Rejects positions when correlated exposure exists.
11. Max Open Trades per Symbol/Magic — Prevents position stacking.
12. Virtual Stop Loss — Optional hidden stops for instruments with wide minimums.

---

13. Backtesting and Forward Testing Protocol

13.1 Stage 1 — Genetic Optimization (Open Prices Only)

1. View → Strategy Tester (Ctrl+R)
2. Select AbacuQuant, set symbol and H1 timeframe
3. Date range: 2+ years forex, 1+ year indices/stocks
4. Modeling: Open Prices Only
5. Optimization: Genetic Algorithm, Forward 50%
6. Optimization criterion: Custom Max (uses anti-overfitting OnTester score)
7. Computing: Local Agents Only (required for forward test tab)
8. Optimize 3-5 key parameters at a time (conservative approach)
9. Click Start

13.2 Stage 2 — Forward Test with Tick Story

1. Select best parameter set from Stage 1
2. Re-run with same date range, Every Tick Based on Real Ticks modeling
3. Use Tick Story custom instrument matching live broker data quality
4. If equity curves match closely → result is robust
5. If significant divergence → overfitted, discard and retune

13.3 Acceptance Criteria

• Profit Factor ≥ 1.5
• Linear Regression R² ≥ 0.85 (equity smoothness)
• Maximum Drawdown < 15%
• Recovery mode disabled (no grid, no martingale)

The OnTester score also disqualifies: fewer than 50 trades, drawdown above 8%, profit factor below 1.2, negative expectancy. Penalizes win rates above 70% (overfit) or below 30% (weak), and unbalanced payoff ratios.

13.4 Demo Forward Testing

1. Open demo with intended live broker
2. Deploy validated configuration on demo
3. Run 30-90 days minimum
4. Compare daily P&L to backtest expectations
5. Acceptable variance: ±20% monthly P&L, ±2% on max drawdown
6. Investigate larger deviations: spread, slippage, execution

13.5 Common Divergence Causes

Symptom	Cause	Action
Demo loses more than backtest	Wider broker spread	Tighten MaxSpreadPips, switch to ECN
Slippage on every trade	Slow execution	VPS in same datacenter as broker
Trades not opening on demo	Larger broker stop level	v4.0 ATR fix handles this; verify
Symbol mismatch	Different naming conventions	Re-run tester with broker's symbols

---

14. Live Deployment Best Practices

14.1 VPS Selection

• RAM: 4 GB minimum, 8 GB recommended (16 if running Ollama)
• CPU: 2 cores minimum, 4 for multi-EA
• Storage: 50 GB SSD
• Latency: Under 50 ms to broker servers
• OS: Windows Server 2019 or 2022
• Providers: Vultr, ForexVPS, Beeks, Contabo, AlexHost

14.2 Broker Selection

• ECN or Raw Spread with commission
• EURUSD spread under 0.3 pips raw + commission
• Execution under 100 ms average
• Symmetric slippage (positive and negative)
• Stop level as low as possible (0-5 points on majors)
• Full symbol coverage: forex, gold, oil, NDX, NJ225, US stocks
• Tier-1 regulation (FCA, ASIC, CySEC, FSCA)

14.3 Pre-Launch Checklist

1. Algo trading enabled in toolbar
2. WebRequest URLs added if using AI
3. Smiley face on chart is happy
4. Initialization log shows correct preset and lockdown banner
5. Spread filter configured for instrument
6. Sufficient free margin
7. Notifications configured if desired
8. Performance log and journal enabled
9. VPS auto-restarts MT5 if it crashes

14.4 Ongoing Monitoring

• Daily: Glance at dashboard for unusual drawdown
• Weekly: Review trade journal CSV
• Monthly: Compare actual to backtest expectations
• Quarterly: Re-run Portfolio Scorer if managing multiple EAs
• After news events: Verify filter blocked trades as expected

---

15. Troubleshooting

15.1 EA Not Opening Trades

Cause	Verify	Solution
Algo trading disabled	Toolbar red, smiley X	Click Algo Trading button
Spread too wide	Dashboard shows spread > max	Increase MaxSpreadPips or wait
Daily loss limit hit	Status: "Daily Loss Limit"	Wait until next day
No strategy enabled	Init log shows none active	Enable at least one strategy
Insufficient margin	"Not enough money" in log	Reduce risk or deposit more
Outside session	Status: "Session Not Active"	Verify session inputs
News block	Status: "News Block"	Wait for window to clear
Trailing DD locked	Status: "Trailing DD - Locked"	Manual reset

15.2 Other Issues

• "Cannot load custom indicator": Insufficient bars in history. Wait or load more history.
• "WebRequest is not allowed": Add URL to MT5 → Tools → Options → Expert Advisors
• EA disappears on chart: Compile errors. Recompile in MetaEditor or contact support.
• Slow execution: Disable AI for testing. Reduce dashboard rate. Close unused charts.
• High CPU: Likely Ollama on CPU. Use smaller model or disable.

---

16. FAQ and Glossary

16.1 Frequently Asked Questions

Can I use AbacuQuant on multiple charts at the same time? Yes. Use different MagicNumberBase per instance. Portfolio Scorer is designed for this.

Does AbacuQuant work with all brokers? Yes (Article 2555 compliant), but ECN/Raw Spread brokers strongly recommended for performance.

Minimum recommended deposit? $200 with 1:500 leverage for single EA testing. $5,000+ for full portfolio diversification.

Prop firm compatible? Yes. v4.0 includes lockdowns for FTMO, FundedNext, The5ers, FundingPips, Goat Funded Trader.

Need to pay for AI? No, AI is optional. EA works fully without it. Cloud AI requires API keys (paid). Ollama is free and local.

What if Ollama goes down during trading? Audit mode (false): trades continue normally, analysis logged as failed. Filter mode (true): fail-safe — trades proceed without AI confirmation.

Mac/Linux compatible? Yes via Parallels (Mac) or Wine (Linux). Performance comparable to Windows native.

How often is it updated? Active development. Updates delivered automatically through MQL5 Market.

Need to recompile or modify code? No. Delivered as compiled .ex5. All configuration via input parameters.

Can I share my license? No. License tied to your MQL5 account per terms of service.

16.2 Glossary

Term	Definition
ATR	Average True Range — volatility measure for dynamic stop loss/take profit.
Confluence	Agreement of multiple strategies on same direction. Higher signal quality.
Drawdown	Percentage decline from previous equity peak.
Forward Test	Validation on out-of-sample data not used in optimization.
Genetic Optimization	Heuristic search using evolutionary algorithms for parameter tuning.
Hedging Account	Allows simultaneous long and short on same instrument.
Linear Regression (R²)	Equity curve smoothness measure. Higher = smoother growth.
Magic Number	Unique trade identifier for EA position management.
Netting Account	Aggregates positions on same instrument into single net position.
OnTester Score	Custom optimization criterion that filters overfitted parameter sets.
Pearson Correlation	Linear correlation between return series. Used by Portfolio Scorer.
Profit Factor	Gross profit divided by gross loss. 1.5+ generally viable.
Recovery Factor	Net profit divided by max drawdown. Higher = better risk-adjusted return.
Sharpe Ratio	Risk-adjusted return. 1+ good, 2+ very good, 3+ excellent.
Slippage	Difference between expected and actual execution price.
Stop Level	Minimum allowed distance between price and stop order, set by broker.
Tick Story	Tool to create realistic Strategy Tester instruments from real historical ticks.
VPS	Virtual Private Server for 24/7 MT5 hosting.

---

End of Manual

For setup assistance, configuration recommendations, or to report any issue, contact through MQL5 private message. Every question is answered personally. Updates and additional documentation are published in the MQL5 blog associated with this product.

View AbacuQuant on the MQL5 Market
View Live Trading Signal

---

Keywords: AbacuQuant 4.0, MetaTrader 5 Expert Advisor, MQL5 EA, Algorithmic Trading, Automated Forex Trading, Multi-Strategy EA, AI Trading System, Local LLM Trading, Ollama Trading Bot, Llama 3 Forex, OpenAI Trading EA, Gemini Trading EA, DeepSeek Trading, Prop Firm EA, FTMO Compatible EA, FundedNext EA, The5ers EA, FundingPips EA, Goat Funded Trader EA, Funded Account Trading, Portfolio EA, Multi-EA Portfolio, Pearson Correlation EA, Risk Management EA, ATR Stop Loss, Trailing Drawdown, Auto Lot Scaling, MetaTrader 5 News Filter, Economic Calendar EA, Backtesting Protocol, Strategy Tester Optimization, Genetic Algorithm, Forward Testing, Tick Story Validation, Darwinex Zero, EURUSD EA, XAUUSD EA, Gold Trading EA, NASDAQ EA, NDX EA, GBPJPY EA, Crude Oil EA, XLV ETF EA, Hedging Account EA, ECN Broker EA, Institutional Grade EA, Systematic Trading, Quantitative Trading, Sharpe Ratio Trading, Low Drawdown EA, MQL5 Market