Monetears Neural Network AI
- 专家
- 版本: 3.0
- 激活: 5
The robot watches every candle and studies four things: the Open, High, Low, and Close prices. It converts this data into patterns and teaches itself, bar by bar, which patterns lead to rising prices and which lead to falling prices.
It doesn't just use one brain — it uses a team of 3 independent AI agents. Each agent learns separately, and they vote together. A trade only happens when the team agrees with strong confidence. This makes the signals more reliable than a single network.
Learning never stops. On every new candle, the agents retrain themselves using the latest market data, so the robot adapts automatically when market conditions change — no manual updates needed.
Built on a proven, safe foundation
The AI signal is layered on top of a battle-tested risk management engine:
- ✅ ATR-based stops — stop loss and take profit automatically adjust to each instrument's volatility
- ✅ Smart spread protection — skips trading when the spread is too wide (news, rollover)
- ✅ SMA trend filter — optional moving average filter keeps trades in the direction of the main trend
- ✅ Broker-safe orders — respects stop levels, tick size, volume steps, and margin requirements
- ✅ One position at a time — no over-trading, no grid, no martingale
Key Advantages
🧠 Self-learning — adapts to any symbol: forex, gold, indices, crypto
⚡ Fully automatic — set it on the chart and it trades, learns, and manages itself
🛡️ Risk first — every trade has a stop loss; position sizing follows your settings
🔧 No coding needed — all settings adjustable from the inputs panel
💻 Lightweight — pure MQL5, no Python, no DLLs, no external tools
Input Parameters
| Parameter | Default | Description |
|---|---|---|
| **InpLots** | 0.01 | Trade volume in lots. Automatically snapped to your broker's minimum, maximum, and step size. |
| **InpMAPeriod** | 100 | SMA period used for the trend filter. |
| **InpUseMAFilter** | true | If enabled, the EA only buys above the SMA and only sells below it. |
| **InpDeviation** | 20 | Maximum allowed slippage in points. |
| **InpMagic** | 193982 | Unique ID for the EA's trades, so it never confuses them with your manual trades or other EAs. |
Stop Loss & Take Profit
| Parameter | Default | Description |
|---|---|---|
| **InpATRPeriod** | 14 | ATR period used to measure market volatility. |
| **InpMinSLATR** | 0.5 | Minimum stop distance, in ATR. Stops will never be tighter than this. |
| **InpMaxSLATR** | 3.0 | Maximum stop distance, in ATR. Prevents stops from becoming too wide. |
| **InpRiskReward** | 2.0 | Take profit = stop distance × this value. 2.0 means reward is twice the risk. |
| **InpMaxSpreadFrac** | 0.30 | Skips the trade if the spread is more than 30% of the stop distance. |
Neural Network Settings
| Parameter | Default | Description |
|---|---|---|
| **InpHiddenNeurons** | 8 | Number of neurons in the hidden layer. More neurons = more learning capacity (8 is a good balance). |
| **InpAgents** | 3 | Number of AI agents in the ensemble. They vote together — more agents = more stable signals. |
| **InpTrainBars** | 500 | How many past candles the agents learn from. |
| **InpTrainEpochs** | 150 | How many training passes each agent performs. |
| **InpLearnRate** | 0.1 | How fast the agents learn. Lower = slower but steadier learning. |
| **InpSignalThreshold** | 0.60 | Minimum confidence (60%) required from the AI team before opening a trade. Raise it for fewer, safer trades. |
| **InpRetrainOnNewBar** | true | If enabled, agents retrain on every new candle to stay in sync with the market. |
