Lamtem Bot
An AI-powered Smart Money Concepts trading bot for Deriv synthetic indices.
Lamtem Bot is a pre-development trading system designed to analyse synthetic indices, detect Smart Money Concepts structure, score trade quality with machine learning, and execute directly through MetaTrader 5.
The big idea is not just automation, but transparent automation: a bot that shows its market structure, predicted path, risk logic, and live reasoning clearly enough that a human can follow every decision.
Project Summary
Manual trading demands constant attention, strong emotional control, and the ability to monitor multiple instruments and timeframes at once. For most traders, that level of consistency is hard to maintain every day.
Lamtem Bot is designed to take over that burden by combining rule-based structure detection, multi-timeframe confirmation, AI-assisted setup filtering, and controlled execution so the trader can supervise results instead of manually trading every move.
System Architecture
Data Pipeline
Pulls live and historical OHLCV candle data from MT5 across 1H, 15M, and 5M timeframes.
SMC Engine
Detects swing structure, BOS, CHoCH, order blocks, fair value gaps, and liquidity zones.
ML Scoring Layer
Ranks setups using trained historical pattern data so low-probability entries can be filtered out.
Risk Manager
Controls sizing, drawdown limits, stop loss and take profit logic, and pause conditions.
Execution Engine
Places and manages trades through the MT5 Python API with logging and monitoring built in.
Visual Dashboard
Explains the bot live through chart overlays, prediction paths, checklist logic, and performance panels.
What Makes It Strong
Multi-Timeframe Logic
Uses 1H for directional bias, 15M for point of interest detection, and 5M for confirmation entry timing.
Prediction Overlay
Draws the expected path, retest, provisional stop loss, and take profit before a trade even triggers.
Live Reasoning Panel
Shows which conditions are confirmed, pending, or informational so every setup remains interpretable.
State Persistence
Stores market structure, active zones, pending setups, and trade state so the bot can recover after interruptions.
Telegram Alerts
Sends setup, entry, exit, daily summary, and error notifications directly to mobile.
Narration Engine
Turns each setup and result into plain-English reasoning, making the system easier to trust and review.
Build Direction
Risk and Execution Controls
| Control | Purpose |
|---|---|
| 1% Position Risk | Sizes trades relative to current balance so risk stays controlled as the account changes. |
| Minimum 1:2 RR | Skips setups that do not offer enough upside to justify the downside. |
| Daily Loss Limit | Halts trading when drawdown crosses the session threshold, protecting the account on bad days. |
| Consecutive Loss Pause | Forces a cooldown after repeated losses so the bot does not keep trading through poor conditions. |
| ATR and Volatility Filters | Avoids dead markets, unstable spikes, and structurally messy conditions. |
| Kill Switch Intelligence | Lets the bot shift into monitoring-only mode when overall market quality becomes unreliable. |
Advanced Intelligence Layer
Market Regime Detection
Changes behaviour based on whether the market is trending, ranging, dead, or highly volatile.
Confidence Breakdown
Scores structure, liquidity, timing, volatility, and candlestick confirmation separately instead of using one opaque number.
Zone Memory
Tracks tap count, mitigation, age, and historical importance so stale zones are treated differently from fresh ones.
Execution Intelligence
Evaluates fakeouts, spread quality, entry depth, and momentum behaviour before sending an order.
Candlestick Pattern Layer
Uses entry-timeframe confirmation such as engulfings, pinbars, stars, and tweezers to add structure-aware confidence.
Offline ML Training
Uses labelled historical SMC setups to train a classifier that sits on top of the core rule engine.
Technical Direction
| Layer | Technology | Role |
|---|---|---|
| Core Language | Python 3.10+ | Main implementation language for analysis, execution, and orchestration. |
| Broker Integration | MetaTrader5 Python library | Handles chart connection, candle retrieval, and order execution through MT5. |
| Data Layer | pandas, numpy | Processes historical and live candle data for structure analysis and training. |
| Indicators | ta | Supports ATR, EMA, RSI, and other context measurements used in filters. |
| Machine Learning | scikit-learn | Trains setup-quality classifiers such as Random Forest and Gradient Boosting. |
| Visualization | Plotly, matplotlib | Powers the browser dashboard and supporting visual review outputs. |
| Notifications | python-telegram-bot | Delivers trade events, summaries, and fault alerts. |
| Runtime | Local PC to VPS | Starts in local development mode with a clear path to 24/7 hosted operation. |
Why This Project Matters
Lamtem Bot stands out in the portfolio because it is not only a product idea. It is also a systems-engineering challenge that touches algorithmic logic, AI-assisted filtering, financial discipline, state recovery, visual explanation, and real-time automation in one coordinated build.
It shows an interest in building tools that do more than operate quietly in the background. The system is intentionally designed to explain itself, making it both technically ambitious and unusually transparent.
Want to Talk About It?
Reach out if you want to discuss the trading logic, MT5 architecture, AI scoring layer, or how the live reasoning dashboard could be built.
Get in Touch