Knowledge Quantitative developer needed (mql5 ML+python l

I have a EA/indicator that I want built. I should say 1st off dont know how to code myself so I will be using AI to verify that the source code is complete and matches the documents spec or if better so if you can not truly do the job do not waste either of out time. This is a idea I came.up wit and used AI to produce a framework for it.... and of course AI isn't 100% accurate so I need a knowledgeable quantitative developer to work with (potentially looking for a reliable developer to work with permanently, i use AI to come up with a lot of different ideas). Also I don't have a problem paying mor than 30 of course... we can agree upon a set price but I will not add the money until the end when I know you can actually do the job smh... every time I have to cancel a job because a developer takes on a job they say they can do but then can't mql5 takes 10% regardless and im tired of wasting money.

Below isthe plan document below n a simple layout. I have documents that go much more indepth

AGI DeltaFlow Trading System — Master Plan

Objective

Build an institutional-grade AGI trading indicator for MetaTrader 5 (MQL5) integrated with a Python ML/DL/RL engine. The system must produce high-probability Buy/Sell signals by synthesizing DeltaFlow Profile, MoneyFlow Profile, Volume Profile, multi-timeframe VWAP, Price Action concepts, and a synthetic orderbook — all driven by a self-training, walk-forward-validated intelligence layer.

Architecture Overview

MQL5 Indicator Layer: Real-time market data ingestion, all profile calculations, signal visualization, and strict R:R trade mapping.

Python Intelligence Layer: ML/DL/RL model training, WFA (Walk Forward Analysis), synthetic orderbook construction, signal reinforcement, and continuous retraining loop.

Bridge: MQL5 writes feature vectors + bar data to shared CSV/JSON; Python reads, predicts, and writes signal decisions back to shared files or ZeroMQ.

Validation Loop: Historical backtest → identify weak regimes → retrain → backtest → repeat until convergence.

Stage 1 — Deep Research & Feasibility

Load: deep-research-swarm (selectively on free MQL5 equivalents, synthetic orderbook methods, free Python RL/DL frameworks). Deliverable: info.md — validated free library list, MT5-Python bridge options, synthetic orderbook techniques.

Stage 2 — Specification Design

Write SPEC.md covering: - MQL5 indicator module breakdown (all 5 profile domains + price action) - Python module breakdown (feature engineering, models, WFA, retraining) - Data bridge specification (file/ZeroMQ schema) - Signal generation & strict R:R logic - Visualization & alert specs Deliverable: /mnt/agents/output/SPEC.md

Stage 3 — MQL5 Core Indicator Implementation

Sub-agents (parallel where possible): - mql5_profiles: DeltaFlow, MoneyFlow, Volume Profile (POC/VAH/VAL), VWAPs - mql5_priceaction: Internal/swing structure, order blocks, FVGs, equal highs/lows, premium/discount - mql5_signal_engine: Signal aggregation, strict R:R mapping, synthetic orderbook approximation, alert system Deliverable: MQL5 source files committed to project/mql5/

Stage 4 — Python Intelligence Engine

Sub-agents (parallel where possible): - py_feature_eng: Historical data pipeline, feature vector construction, normalization - py_ml_core: ML/DL models (XGBoost, LSTM/GRU) using only free libs (scikit-learn, TensorFlow/PyTorch, xgboost, pandas, numpy) - py_rl_wfa: Reinforcement Learning agent (PPO/DQN) + Walk Forward Analysis engine with rolling train/test windows - py_bridge_server: File/ZeroMQ bridge listener, signal server, retraining scheduler Deliverable: Python source committed to project/python/

Stage 5 — Integration & Backtest v1

Wire MQL5 ↔ Python bridge

Pull real historical data via yahoo_finance data source

Run first backtest (Python offline engine using historical CSV)

Record metrics: Profit Factor, Sharpe, Max Drawdown, Win Rate, R:R adherence Deliverable: backtest_v1_report.md

Stage 6 — Analyze → Upgrade Loop (Iterative)

For each cycle: 1. Identify worst-performing regime/timeframe/market from backtest 2. Dispatch upgrade sub-agent targeting that exact weakness 3. Retrain/reconfigure 4. Re-backtest 5. Record delta Repeat until all regimes covered or diminishing returns.

Key Constraints

MQL5: Only free, built-in, or open-source MQL5 functions. No paid libraries.

Python: Only free external dependencies (numpy, pandas, scikit-learn, xgboost, tensorflow/pytorch, gymnasium, pyzmq, etc.). No paid APIs.

Synthetic Orderbook: Approximate via tick-volume delta, bar delta, market-sweep detection, and limit-cluster inference.

No trailing stops — strict fixed R:R per signal.

Must be self-adjusting: auto-detect regime and load corresponding WFA-validated model weights.

File Layout (per SPEC.md)

/mnt/agents/output/project/

├── mql5/

│ ├── AGI_DeltaFlow_Indicator.mq5 (main indicator)

│ ├── Profiles/

│ │ ├── DeltaFlow.mqh

│ │ ├── MoneyFlow.mqh

│ │ ├── VolumeProfile.mqh

│ │ └── VWAP.mqh

│ ├── PriceAction/

│ │ ├── Structure.mqh

│ │ ├── OrderBlocks.mqh

│ │ ├── Liquidity.mqh

│ │ └── PremiumDiscount.mqh

│ ├── SignalEngine/

│ │ ├── SignalAggregator.mqh

│ │ ├── RiskReward.mqh

│ │ └── SyntheticOrderbook.mqh

│ └── Bridge/

│ └── MT5PythonBridge.mqh

├── python/

│ ├── data/

│ │ └── fetch_yahoo.py

│ ├── features/

│ │ └── engineer.py

│ ├── models/

│ │ ├── ml_ensemble.py

│ │ ├── dl_sequence.py

│ │ └── rl_agent.py

│ ├── wfa/

│ │ └── walk_forward.py

│ ├── bridge/

│ │ └── server.py

│ ├── backtest/

│ │ └── engine.py

│ └── config/

│ └── regime_configs.yaml

└── README.md