AlgoBullion — Best Free Forex Indicator for MT5: AI Pattern Matching + Smart Money Flow Detector

Ranked among the best free forex indicators for MT5. A 45-buffer institutional-grade detector with k-NN pattern similarity engine, BOS/MSS/FVG smart money features, 5-regime market classification, adaptive sensitivity with self-learning weights, and a real-time dashboard. Free compiled ex5 download — no registration required.

Free Download — AlgoBullion v3.00

Compiled EX5 file. Works on any MetaTrader 5 platform. Licensed until 2030.
Source code available upon request — contact us for licensing.

File size: ~120 KB  |  Platform: MetaTrader 5  |  License valid until Jan 1, 2030

How AlgoBullion Looks on Chart

AlgoBullion renders a colored EMA line on the price chart with a 4-layer confidence cloud, plus match boxes showing the top historical periods that are most similar to current market conditions. A floating dashboard panel shows real-time metrics.

AlgoBullion on an MT5 chart. The colored EMA line changes between green (bullish), red (bearish), and blue (neutral). Match boxes in the background show the most similar historical patterns with similarity percentage and directional labels.

Table of Contents

1. What Is AlgoBullion?
2. The 7 Detection Engines
3. The 12-Dimensional Feature Vector
4. How Pattern Matching Works — k-NN with Mahalanobis Distance
5. Dashboard Panel Explained
6. Input Parameters — Full Reference
7. How to Read AlgoBullion Signals
8. Alerts & Notifications
9. Self-Learning: Adaptive Weights & Win Rate Tracking
10. Installation Guide
11. FAQ

1. What Is AlgoBullion?

AlgoBullion is an institutional-grade MT5 indicator built for traders who want more than just another moving average crossover. It combines a k-NN pattern similarity engine that scans up to 2,000 historical bars to find the closest matches to current market structure, with smart money detection (BOS, MSS, FVG, liquidity sweeps), 5-regime market classification, and a self-learning system that adapts its sensitivity and feature weights based on real-time outcome tracking.

The output is simple — a colored EMA line (green = bullish, red = bearish, blue = neutral) with a 4-layer confidence cloud. Match boxes show you exactly which historical periods the current market most resembles, with similarity percentages and directional labels. A docked dashboard panel gives you 37 rows of real-time analytics including probability distribution, win rates per regime, session strength, and the 12-dimensional feature breakdown.

2. The 7 Detection Engines

AlgoBullion is not a single-indicator system. It runs 7 concurrent detection engines, each feeding into the final probability-weighted output.

3. The 12-Dimensional Feature Vector

Every bar in AlgoBullion is described by a 12-dimensional feature vector. This is what makes the pattern matching statistically meaningful — it's not comparing just one indicator value. It's comparing momentum across three time horizons, volatility, volume, market structure, displacement, and institutional flow signatures all at once.

Features F1–F6 capture what most indicators track (price + volume + volatility). Features F7–F12 capture what institutional traders track (displacement, structure, imbalance, and where trapped orders sit). The combination is what separates AlgoBullion from single-indicator systems.

4. How Pattern Matching Works — k-NN with Mahalanobis Distance

The Layman Explanation

Imagine you're looking at today's market and asking: "When has the market looked exactly like this before?" AlgoBullion answers that question mathematically. It takes the current bar's 12-dimensional fingerprint (momentum, volatility, volume, structure, etc.) and scans the last 300-2,000 historical bars looking for the 5 most similar fingerprints. It then checks what happened after those similar periods — did price go up or down? — and uses that to compute a weighted probability for the current bar.

The Technical Explanation

AlgoBullion uses a Mahalanobis-style standardized distance over the 12D feature space. Each feature is normalized by its historical standard deviation so that a 1% change in ROC5 (which might be normal) is penalized differently than a 10% change in ATR % (which would be extreme). The distance formula is:

distance = sum( |current_feature[j] - historical_feature[j]| / std[j] * weight[j] ) over j=0..11

This distance is then converted to a similarity score using an exponential decay function: similarity = 100 * exp(-distance * sensitivity). The sensitivity parameter controls how strictly the engine penalizes small differences — lower values give more matches (good for ranging markets), higher values only match nearly identical conditions (good for clean trends).

Key Selection Optimizations

• Heap-based Top-K: O(N log K) complexity instead of O(N^2) — 2,000 bars scanned in milliseconds.
• Regime Gating: Matches are only pulled from bars in the same market regime (Trending Up/Down, Breakout, Ranging, Low Vol). You don't match a trending bar to a consolidation bar — that would produce false signals.
• Two-Pass Fallback: If pass 1 (strict regime gate) produces fewer than 3 matches, pass 2 widens to any regime to ensure the dashboard always has data.
• Temporal Decay: Older historical matches are penalized by an exponential decay factor. A match from 10 bars ago carries more weight than one from 500 bars ago.
• SMA-Seeded EMA: The resulting weighted momentum line is smoothed with an EMA that is seeded using an SMA of the first N values — eliminating the cold-start bias that plagues standard EMAs.

5. Dashboard Panel Explained

The right-side dashboard panel shows 37 rows of real-time data updated every second via the OnTimer event. It is fully draggable — click and drag the dark panel to reposition it anywhere on your chart.

6. Input Parameters — Full Reference

AlgoBullion exposes 40+ customizable input parameters grouped into 10 logical sections. Below is the parameters dialog showing every option you can configure.

7. How to Read AlgoBullion Signals

8. Alerts & Notifications

AlgoBullion supports 8 independent alert triggers, each configurable independently with popup (Alert), mobile push notification, and sound output.

9. Self-Learning: Adaptive Weights & Win Rate Tracking

AlgoBullion contains a persistent learning system that stores running statistics across MT5 sessions using GlobalVariables. This means the indicator gets smarter the longer you use it on a specific symbol.

• Per-Regime Win Rates: For each of the 5 market regimes (Trend Up/Down, Breakout, Ranging, Low Vol), the engine tracks total predictions and correct predictions. A correct prediction is one where the weighted momentum direction matches the actual price movement X bars forward.
• Adaptive Feature Weights: Each of the 12 features starts with a default weight (see Section 3 table). After each prediction, the engine checks if the prediction was correct. If yes — features that were closely matched get their weights increased. If no — features with large discrepancies get penalized. The learning rate is 0.05, so weights adapt gradually over hundreds of bars.
• Per-Symbol Persistence: All learning data is keyed by symbol + regime. EURUSD Trend Up has its own weights and win rate. XAUUSD Breakout has its own. GBPUSD Ranging has its own. This prevents cross-symbol contamination.
• Dashboard Transparency: All 5 regime win rates are displayed in real-time on the dashboard (green if above 60%, blue if 45-60%, red if below 45%, dim if no data). You can see which market conditions AlgoBullion performs best in — and avoid the ones where it struggles.

10. Installation Guide

1. Download AlgoBullion.ex5 from the download button at the top of this page.
2. Copy the file to: MQL5\Indicators\ (Open MT5 → File → Open Data Folder → MQL5 → Indicators).
3. Restart MT5 or right-click the Navigator panel and click Refresh.
4. Find AlgoBullion under Navigator → Indicators.
5. Drag it onto any chart (works on all instruments: XAUUSD, EURUSD, GBPUSD, US30, NASDAQ, etc.).
6. For alerts to work, enable push notifications in MT5: Tools → Options → Notifications → Enable Push Notifications → enter your MetaQuotes ID.
7. For news filter to work, the economic calendar must be available in your MT5 terminal (Tools → Calendar → enable).

The indicator is valid on all timeframes. The scan depth auto-scales: H1 uses 300 bars × 2 = 600. H4 uses 300 bars × 3 = 900. Higher timeframes get proportionally deeper scans.

11. FAQ