Only MQ5 and Python: EAs and indicators for everyone

Dear aydinsarihan,
you helped me by making changes to my divergence indicator and still with a view to providing valid advice, an excellent signal is the following:
1) the divergence
2) the bar that gives rise to the divergence MUST be point 1 of the 123 of Ross

I called this indicator 4S because from the divergence low to the low of point 3, it forms an S, and the entire pattern is composed of 4 points. I'd like to point out that the 123, like the Stochastic divergence, is composed of a very specific sequence of bars, not zigzags like those found elsewhere. It took me months to create all the possible bar combinations (inside, outside, shark32, etc.), but in the end, this is the result.

EURUSD 5 min today

Dear carbonmimeti, thank you very much for the documents you have prepared. While you explain things simply on your site, it will require some effort to ensure the topic is fully understood. I'll work on it a bit today.

It's not a usable version right now. I can only get a glimpse of the view in the H1 chart, but it's not very easy. I'll look at it again at a later time.

Next-Gen PyTorch - Quant Trader

The Future of MT5: Autonomous PyTorch Trader with Sekanas-Norm 1H Integration)

nstitutional Grade PyTorch-MT5 Quant System (v2.0)
This system is an autonomous trading framework that integrates Deep Learning (Multi-Head Attention + LSTM) with the MetaTrader 5 terminal, designed for professional risk management and high-frequency execution.

1. System Requirements

1. OS: Windows 10/11 or Windows Server 2012+ (MT5 Python API is Windows-exclusive).
2. Python: 3.8 - 3.10 recommended.
3. MT5 Terminal: Latest build from your broker.
4. Hardware: Min 8GB RAM. While GPU (CUDA) is supported, the model is highly optimized for CPU-only environments to save costs on VPS.
   Attached Image (click to enlarge)

   
   
   
   
   Attached File(s)

   QuantBot.zip   12 KB | 132 downloads
   
   

2. Directory Structure & Installation
Extract the files into the following directory structure to ensure module imports work correctly:

C:/PyTorchMT5/QuantBot/
├── live_trader.py # Main execution engine
├── train.py # Model training script
├── model.py # PyTorch Architecture (Attention+LSTM)
├── data_loader.py # MT5 Data bridge & Feature engineering
├── risk_manager.py # Sk-N1H Risk protocols
├── settings.py # API & Account configuration
├── logger_setup.py # Monthly log rotation system
└── models/ # Storage for trained .pth files
3. MT5 Configuration

1. Pathing: Update MT5_TERMINAL_PATH in settings.py with your terminal64.exe location.
2. Permissions: In MT5, go to Tools > Options > Expert Advisors and check "Allow Algorithmic Trading" and "Allow DLL imports".
3. Credentials: Enter your MT5_LOGIN, PASSWORD, and SERVER in settings.py.

4. Execution Flow

1. Install Dependencies: pip install pandas numpy torch MetaTrader5 scikit-learn psutil requests
2. Train the Intelligence: Run python train.py to fetch historical data and train the local model.
3. Go Live: Run python live_trader.py to start the autonomous cycle.

5. Architecture & Learning Logic (The "Edge")

1. Attention-Driven Forecasting: Unlike standard indicators, this model uses a Multi-Head Attention mechanism to focus on the most "significant" price bars within a 50-period lookback, ignoring market noise.
2. Sk-N1H Optimization: Our unique Sekanas Norm 1H logic recalculates trading costs and volatility only once per hour (H1 Open). This prevents CPU spikes common in tick-by-tick bots and ensures ultra-stable risk calculations.
3. Triple Barrier Labeling: The model is trained using the Triple Barrier Method (Profit Take / Stop Loss / Time-out), ensuring it learns not just price direction, but the "volatility-adjusted timing" of a trade.
   Attached Image (click to enlarge)

   
   
   Currently, the profit graph is very erratic. It's not stable yet. I'm sharing this as an example for developers. I was able to progress this far with the most limited resources and the worst computer configuration available. More powerful computers could lead to more innovative solutions.
   
   

Trial Phase: Smart News Scanning & AI-Powered Analysis and Monitoring System

1. This system features a corporate-grade architecture designed to monitor fundamental data impacting XAUUSD (Gold) in real-time.
2. The software is optimized to scan the global economic calendar and major financial news feeds within seconds.
3. Economic calendar events are integrated into the system 15 minutes prior to release as a "Pre-News" phase.
4. The first 5 minutes following a news release are evaluated as the "Post-News" phase to measure immediate market reaction.
5. Beyond standard calendar data, the system captures instant "Breaking News" headlines that trigger market volatility.
6. All captured raw data is transmitted to an advanced AI module with high-level Natural Language Processing (NLP) capabilities.
7. The AI module analyzes the potential impact of news on Gold prices based on historical data and real-time liquidity.
8. During this analysis, live Bid prices and Spread data from MetaTrader 5 (MT5) are utilized as critical context.
9. The system's core strength lies in its ability to not just read data, but to interpret it according to market sentiment.
10. The AI generates a clear output providing specific directions such as "BUY", "SELL", or "WAIT".
    
    Attached Image (click to enlarge)

    
    
    
    
11. These outputs include vital details such as the risk level (Low/Medium/High/Extreme) and the expected duration of the news impact.
12. To maintain trading discipline during high volatility, the system suggests a "Pause Duration" for delaying entries.
13. Technical directives, including lot size multipliers, are automatically calculated for users of Expert Advisors (EA).
14. A specialized anti-lock algorithm filters redundant news to prevent data clutter and system lag.
15. MT5 integration is handled via an isolated thread to ensure the core engine remains fast and responsive.
16. Every signal and analysis generated is relayed to the Telegram infrastructure within milliseconds for instant delivery.
17. This framework enables traders to follow a data-driven, AI-supported strategy rather than making emotional decisions.
18. Operating fully asynchronously, this engine monitors the market 24/7 to catch opportunities that the human eye might miss.
19. All live analyses and instant signals produced by the system are shared with full transparency.
20. You can follow all these processes and live-streamed signals instantly via the @For_X_News_Gold Telegram channel.

Examples of messages transferred to Telegram (@For_X_News_Gold)

[20.03.2026 10:03] X_Gold_News: *BREAKING NEWS !*
Germany February PPI -0.5% vs +0.3% m/m expected
*Analysis:* Germany's February PPI unexpectedly fell significantly (-0.5% vs +0.3% expected m/m), missing expectations by a wide margin. This indicates a stronger-than-expected deceleration in producer prices and weakening inflationary pressures in the Eurozone's largest economy. Such data strengthens the case for a more dovish European Central Bank (ECB) stance, potentially leading to earlier or more aggressive interest rate cuts. A more dovish ECB would likely weaken the Euro (EUR) against the US Dollar (USD). A stronger USD typically pressures XAUUSD (Gold) downwards due to their inverse correlation. Furthermore, lower inflation reduces Gold's appeal as an inflation hedge. Expect potential downward pressure on Gold as the market processes the implications for ECB policy and USD strength.
*Direction:* SELL | *Risk:* HIGH
*Lot Size Multiplier:* 0.6x (Impact: 0.75)
*Action Delay:* 25 mins (volatility-adjusted).


[20.03.2026 10:03] X_Gold_News: *BREAKING NEWS !*
SNB holds policy rate at zero, signals increased FX intervention readiness
*Analysis:* The SNB's decision to hold its policy rate at zero, coupled with a signal for increased FX intervention readiness, is a clear dovish signal for the Swiss Franc. This stance indicates the SNB is prepared to weaken the CHF, which can contribute to broader USD strength against other major currencies or foster risk-on sentiment globally. Both scenarios are generally bearish for gold (XAUUSD). Expect moderate downward pressure as markets digest the implications of a dovish SNB and potential USD strength.
*Direction:* SELL | *Risk:* MEDIUM
*Lot Size Multiplier:* 0.6x (Impact: 0.65)
*Action Delay:* 20 mins (volatility-adjusted).

https://t.me/For_X_News_Gold

EA Virtual MartiVir : Smart Lot Filtering with Precision & Dynamic Grid


This Expert Advisor implements a sophisticated hybrid execution model that tracks initial martingale levels virtually and only triggers real market orders once a specific lot threshold is reached. By utilizing a core EMA cross-over strategy with 5 and 16 periods on the M5 timeframe, the system ensures that every grid sequence begins with high-probability trend momentum.
The EA dynamically adapts to market volatility using a unique price-normalized scaling factor, ensuring that Take Profit and Stop Loss levels remain proportional to current price movements. A disciplined grid management system is strictly enforced through a customizable maximum step limit, preventing the common pitfalls of uncontrolled martingale growth. The logic incorporates an intelligent lot multiplier that translates virtual volume into optimized real-world positions, significantly reducing the overall drawdown risk.
Every trading basket is managed collectively, where the EA calculates a weighted average price to close all active positions simultaneously at a unified profit or loss target. To ensure maximum reliability during high volatility, the system includes a robust retry mechanism for both opening and closing orders to mitigate slippage and execution errors.

Traders can benefit from the built-in CSV logging feature, which records every virtual and real transaction with precise timestamps and execution notes for post-trade analysis. The Expert Advisor remains idle during consolidation phases thanks to its bar-based filtering, which prevents multiple grid orders from being opened within the same candle. Ultimately, this tool offers a unique balance between aggressive recovery logic and institutional-grade risk management by keeping the riskiest parts of the sequence off the real balance sheet.

MartiVir EA's live demo account only took 2 trades on the season's most volatile day.
Does anyone have any suggestions regarding optimization?

I have not tried yet, but is it possible to add ADX and VWAP for the trend filter?

Currently, MartiVir EA handles 1-3 trades per day. Today's trades:


I added ADX and VWAP filters as optional. They're having the opposite effect with the default settings. Now is the time to relax the rules and give the filters a chance. I think it's backfiring.
General experience shows that using indicators with negative predictive power is more useful than using positive predictive power because you have to set traps in the market with indicators.

Forex Gold News Api
I've improved it based on your suggestions; let's see if you like it.

@For_X_News_Gold
*BREAKING NEWS*
Economic and event calendar in Asia - Japan wholesale services inflation indicator
*AI Analysis:*
This headline indicates the upcoming or recent release of Japan's wholesale services inflation indicator. While significant for the Bank of Japan's (BoJ) monetary policy outlook, its direct impact on XAUUSD is generally indirect and moderate.
1. Impact on JPY & USDJPY: A higher-than-expected inflation figure could strengthen the JPY, as it might signal a greater likelihood of the BoJ tightening monetary policy (e.g., exiting negative interest rates or adjusting Yield Curve Control) sooner. This would typically lead to a fall in USDJPY. Conversely, a lower-than-expected figure would likely weaken the JPY, pushing USDJPY higher.
2. USD Correlation: The primary channel of influence on XAUUSD will be through its impact on the US Dollar. If a strong JPY (due to high inflation) contributes to overall USD weakness (via USDJPY and broader DXY sentiment), it could provide some support for XAUUSD, as gold becomes cheaper for non-USD holders. Conversely, a weaker JPY leading to USD strength would typically exert downward pressure on gold.
3. Safe-Haven Flows: While Japan is a major economy, its wholesale services inflation data is unlikely to trigger significant global safe-haven flows into gold unless it signals a major systemic economic shock or contributes to widespread market instability. Investors typically look to gold as a safe haven during broader geopolitical tensions or significant global economic crises, or specifically during periods of high US inflation/economic uncertainty.
Conclusion: The market's reaction to this specific Japanese inflation data will primarily influence JPY and USDJPY. The impact on XAUUSD will be a secondary effect, mainly mediated through changes in overall USD strength/weakness. Given that the actual indicator data is not provided, a 'WAIT' stance is prudent. Observe the released data, the immediate reaction in JPY pairs, and subsequently the broader USD market before considering any XAUUSD trade. The impact is likely to be contained unless the figures are vastly different from expectations, causing a significant shift in global monetary policy outlooks.
*Signal:*
├ Direction: *WAIT* | Risk: MEDIUM
├ Lot Multiplier: 0.1x (Impact: 0.78)
└ Delay: 30 mins

Martivir EA
EA's Martivir is free. It seems better than many paid EA games. You can now see the actions it performs online with its default settings.

Which MatiVir Version?

Since I started with the first version(MartiVir_v1_6e), I didn't change it. The logic is the same as the default settings in the version.
I gave all the codes clearly. you can check again.



https://assets.faireconomy.media/images/attach/ex5.gif MartiVir_v1_6e.ex5
MartiVir_v1_6e.mq5


https://assets.faireconomy.media/images/attach/ex5.gif Marti_Vir_v6_7_Filtered.ex5
https://assets.faireconomy.media/images/attach/mq5.gif Marti_Vir_v6_7_Filtered.mq5

Thanks

Maybe it works on crypto, but on fiat where there is no increased volatility, it is not possible:
The reference price is constantly reset to the current Bid
That means: every second (OpenTime = 1) the reference price is set to the current Bid
Therefore:
BuyTriggerLine = Bid + step
SellTriggerLine = Bid – step
And because the reference price is updated before the price has a chance to breakout…
the lines move with the price and the price can never cross them.
So the entry condition is only active in an extremely short window.
SekanasStep is usually too big for a 1 second move
For example:
EURUSD H1 open price ~1.09000
g_Sekanas_Price_Move = 1.09000 × 0.0001 = 0.000109
SekanasStep = 5 → 5 × 0.000109 = 0.000545 (5.4 pips)
So the price must make:
5.4 pips move in one direction in 1 second
This practically doesn't happen.
The MT4 version is adjusted so that the lines are static until the trade is opened and the EA has a real chance to break through them.
I have OpenTime for EU set to 25 (feel free to give it 10–30 seconds) so that the price has time to breakout.
I have SekanasStep for EU set to 5 (you can reduce it for testing (e.g. 2–3) so that you see the first entries faster).
In any case, the original version for MT5 will help you find suitable parameters faster.

Hi Lopuch,
Thank you so much for the deep analysis and for taking the time to test the EA logically. Your observation is absolutely correct and spot on!
The default OpenTime = 1 combined with SekanasStep = 5 acts essentially as an extreme high-frequency/momentum filter. It was initially conceptualized to catch violent volatility spikes (like sudden crypto dumps/pumps or heavy news events), which is why the reference price trails the current Bid so aggressively.
As you perfectly pointed out, in standard fiat pairs like EURUSD, expecting a ~5.4 pip move in a single second is unrealistic. The reference price resets before the price action has a chance to break the trigger lines.
Your adjustment is exactly how the EA should be optimized for Forex pairs. By increasing the OpenTime to 10-30 seconds, we create a stable "trap" or channel, giving the normal market volatility enough time to cross the dynamic Sk-N1H trigger lines. Lowering the SekanasStep to 2 or 3 for pairs like EURUSD is also a great optimization to increase the frequency of valid entry signals.
I really appreciate you sharing these parameter adjustments with the community. This kind of collaborative testing is exactly what helps refine the logic. I will add a note about adjusting the OpenTime for fiat currencies in the next update's documentation.
Thanks again for the great feedback!
Best regards,

Machine Learning System BTC_Grid_Oracle Onnx EA



GridOracle V3 BTC Edition is a hybrid trading system powered by a Python-based machine learning engine designed to accommodate Bitcoin's high volatility. At the heart of the system is the Gradient Boosting Regressor algorithm, which analyzes historical data to predict the expected range of movement (volatility) of the next candle .The machine learning process simultaneously processes eight different technical features, including RSI, ATR, EMA trend strengths, and time-based (hourly/daily) data . Unlike traditional EAs, this system features an "Auto-Loop" capability, constantly updating itself as the market structure changes.
The model, trained in Python, is converted to ONNX format, which MetaTrader 5 can read with high performance, and integrated directly into the terminal.
The system dynamically optimizes grid intervals based on a mathematical prediction, targeting the absolute price change of the next bar as the "Target".
To install, you first BTC_GridOracle_Trainer.pyneed to place the Python script ( ) in a working folder on your computer.
GridOracle_BTCUSD_H1.csvThe data file exported via MT5 must be located in the same main directory as the Python script .
Once the training is complete, you must copy the generated file to the MQL5\FilesGridOracle_BTCUSD.onnx folder in the MetaTrader 5 terminal .
This architecture combines Python's advanced data processing power with MT5's transaction speed to deliver a professional, data-driven grid strategy for the BTCUSD pair.

Very interested, any chance to make it with XAUUSD and USOIL?

Of course, we can use Sekanas in every currency pair according to the standard. To reiterate, I use Sekanas in my work to eliminate the confusion surrounding pip points. 1 Sekanas is one ten-thousandth of an asset. I have standardized TP and SL values using these units, so it works with all currency pairs. You only need to change the pair names in the PY code and the MT5 admin panel. If you have any further questions, you can contact me.

Thats impressive! Any chances to share code of BTC_GridOracle_V4.ex5 ?