automated-trading.ch

Strategy Overview

The Random-Forest Machine Learning Strategy represents a cutting-edge approach to algorithmic trading, combining the power of artificial intelligence with robust order management and risk control. This strategy utilizes a pre-trained Random Forest machine learning model (generated for multiple instruments MES, ES, MNQ, NQ...) to predict future price movements.

Unlike traditional rule-based strategies that rely on fixed technical indicators and predefined patterns, the Machine Learning Strategy learns from historical price data to identify complex, non-linear relationships in market behavior. The model analyzes multiple features including daily price levels, support and resistance zones, market structure changes, and volatility indicators to generate highly contextualized trading signals.

The strategy is designed to be fully automatic and can scale into multiple instances on the same or on multiple accounts. It features robust order management with connection loss recovery, precise risk control with dynamic position sizing, and multiple exit strategies including ATM Strategy integration and manual management options.

How It Works

The Machine Learning Strategy operates in several stages:

1. Feature Engineering

The strategy constructs a comprehensive feature set from market data to feed into the machine learning model:

• Daily Levels: Distance from previous day's high, low, and current day's high, low, and open prices
• Higher Timeframe (HTF) Levels: Distance from 30-minute and 4-hour timeframe previous highs and lows to capture medium-term structure
• Volatility Indicators: Distance from SuperTrend upper and lower bands for adaptive volatility measurement
• Volume Dynamics: Volume differences between the current bar and previous bar to capture momentum shifts
• Price Momentum: Close price differences over the last 3 bars to capture recent directional bias
• >Time of Day: Encoded time features to capture intraday seasonality patterns

2. Model Prediction

The pre-trained Random Forest model processes the feature set to predict the close price for a specified prediction horizon (default: 20 bars ahead on 1 minute timeframe). The model generates a prediction score indicating the expected price movement magnitude and direction.

3. Signal Generation

When the predicted price movement exceeds the prediction threshold (in points), the strategy generates a trading signal:

• A long signal is generated when the model predicts upward price movement
• A short signal is generated when the model predicts downward price movement
• Signals are optionally filtered using SuperTrend indicator to reduce false signals in choppy markets

4. Entry Management

The strategy confirms and executes trades based on confirmed signals:

• Only one position per direction can be active at a time
• Market orders are used for instant execution (no pending orders)
• Entry quantity is determined based on the selected quantity strategy (Fixed or Risk Adjusted based on the stoploss level)
• Trades are only allowed within the specified trading time window

As all our strategies, this strategy implements two methods to set quantity for open positions.

• Fixed Quantity: This method will use the same quantity value for each position. If the stoploss is variable between positions (by using a dynamic stop loss value), The win/loss will be also variable between positions
• Dynamic Risk Adjusted Quantity: This method will calculate quantity of each position based on multiple factors:
  • The stoploss of the position
  • The maximum risk (in $ currency value) allowed for a position to take
  • The maximum quantity allowed for a position to take
  Using this method will produce equal win/loss amounts accross all positions. For example, a position with a tight stoploss will have a greater quantity than a wide stoploss position, so that both positions produce the same profit or loss

5. Position Management

Once a position is opened, the strategy manages it using one of several configurable methods:

• ATM Strategy: Delegates position management to your chosen ATM strategy template
• SuperTrend: Uses SuperTrend indicator to trail stop loss and manage exits dynamically

Apart from those position management methods, the strategy allows to set the stop loss based on multiple strategies and to set the take profit based on a ratio based method. Check the parameters section for a detailed explanation

6. Exit Strategies

The strategy supports multiple exit methods:

• Stop Loss: Multiple options including fixed currency amount, points, SuperTrend levels, swing prices, or prediction signal levels
• Take Profit: Can be set as fixed currency, points, stop loss ratio, or predicted price targets
• Session Closing: Automatically closes all positions at end of trading session
• Daily Limits: Stops trading when daily profit target is reached or daily loss limit is hit
• Trade Count Limits: Can stop after reaching maximum daily winning or losing trade counts

Entry Setup Examples

Example #1: Reversal Short Entry

Let's start with the first example. Running on MES 1 minute timeframe, this example shows a reversal short entry setup. The strategy generated a short signal based on the prediction of the model.Reversal signals are generated on the opposite direction following a Fair Value Gap (FVG). When supertrend entry filter is applied to reversal signals, the strategy will only take the reversal signal if the supertrend is green for bearish signals, and red for bullish signals. The same parameters are going to be used for the next examples.

Example #2: Trend Long Entry

This second example runs on the same instrument, timeframe and parameters as the first example. You can see on this example a first signal that was suppressed by supertrend filter. The first signal was a trend long signal, but supertrend was in Red state so the signal was ignored. Then you can see a second trand bullish signal that was generated while the supertrend was in Green state. So the signal was taken. The take profit is the predicted target by the Model, and the stoploss was the opposite of the take profit target in terms of distance from the entry price, so the risk reward ratio is 1:1 in this trade.

Features

The Random-Forest Machine Learning Strategy includes a comprehensive set of advanced features:

Artificial Intelligence-Powered Predictions

• Pre-trained Random Forest model for robust and adaptive price predictions

Advanced Order Management

• No pending orders - only market orders for guaranteed execution
• Robust in-house Order Manager resistant to connection losses
• ATM Strategy integration for user custom defined position management
• Support for both Rithmic and NinjaTrader brokerage Tradovate accounts

Entry Filtering and Confirmation

• Optional SuperTrend entry filter to confirm directional bias
• Configurable prediction threshold to filter low-confidence signals
• Trading time window restrictions

Precise Backtesting

• Compatible with NinjaTrader's official backtesting best practices
• Tick-level execution for highly accurate results
• Support for optimization across all configurable parameters

Parameters

The strategy provides comprehensive parameter control while keeping the most critical settings easily accessible:

General Settings
Execute Historical Trades	This parameter is very important and should be used carefully. This parameter will enable executing trades when the strategy is run on Historical mode you may check this parameter to test the performance of the strategy on days in the past or to simply test if the used parameters will generate trades on the current day. But when you run the strategy on Live mode, you should uncheck this parameter.
Daily TakeProfit Target	Dollar amount target for daily profit. Trading stops when reached after a position is closed (0 = disabled)
Daily StopLoss Limit	Dollar amount limit for daily losses. Trading stops when reached, the active trade will be closed if Unrealized Pnl reaches this limit(0 = disabled)
Stop @ Winning Count	Stop trading after reaching this number of winning trades per day, a trade is considered a win if its PnL is above 10$ and is considered a loss if below -10$ (0 = disabled)
Stop @ Losing Count	Stop trading after reaching this number of losing trades per day, a trade is considered a win if its PnL is above 10$ and is considered a loss if below -10$ (0 = disabled)
ML Settings
Prediction Threshold Points	Minimum predicted points movement to generate a trading signal. Filters low-confidence predictions, the recommended value for MES is 10 and for MNQ it should be higher
Model File	Path to the pre-trained Random Forest model (.zip file). Model can be selected interactively when the strategy is loaded if not specified
Trading Session
Time Zone	This parameter sets the time zone to be used for the Trade Time parameter below. Since futures markets are open and closed based on US market time zone which is EST (US Eastern, New York), this parameter allows you to avoid adapting the trade time to EST or another timezone if your execution machine is not on the same Timezone as the market. Setting this parameter value to EST will tell the strategy to ignore your local timezone and automatically use EST timezone when running. In the same manner, if you wish to run the strategy based on a CET trade time (suppose you run the strategy on DAX) and your local timezone is EST. You can set this value to CET, and the strategy will automatically figure out the correct time when to start and end trading. EST: This will use EST (Eastern Time) timezone while setting the Trade Time intervals CET: This will use CET (Centrel European Time) timezone while setting the Trade Time intervals Local Time: This will use the local machine timezone
Trade Time (hh:mm-hh:mm,...)	Trading hours in 24-hour format. Example: "09:30-16:00" for standard market hours. Multiple time windows supported (comma-separated). This multiple time windows should start and end on the same day. So if you for example want to start trading at 18:00 and end at 16:00 EST, you should split the intervals at midnight like this => "18:00-23:59,00:01-16:00", or "00:01-16:00,18:00-23:59" will also work
Entry Setup
Order Label	Label for orders placed by the strategy (useful for tracking and filtering)
Quantity Strategy	This parameter sets how the quantity should be set for a Trade Fixed Quantity: This will set a fixed quantity. Before the trade is open, the risk is evaluated and if the risk calculated is higher than the value of the parameter Target Risk Per Trade ($), the trade is skipped and will not be open Risk Adjusted: this option will make the contract size for each trade dynamic and calculated based on the StopLoss price and the Target Risk Per Trade ($) value
Max Qty	This parameter sets the maximum allowed quantity when the Risk Adjusted method is selected. This option is very useful since some propfirms has order quantity limits that can cause the account to be invalidated
Fixed Quantity	This sets the quantity when using Fixed Quantity strategy
SuperTrend Entry Filter	Enable/disable SuperTrend confirmation filter. When enabled, only enter trades in direction confirmed by SuperTrend. Notice that Reversal trades are confirmed by the inverse of SuperTrend direction
Order Management
Order Management Strategy	ATM Strategy: This method will trigger an ATM strategy to which it will entirely delegate the management of the position The ATM (Automated Trading Management) strategy is a mechanism provided by NinjaTrader that can be accessed from the Chart Panel When this method is selected, you should specifiy the template name of the ATM strategy to use and the positions opened will be managed by your custom ATM strategy SuperTrend: This method will use SuperTrend indicator as a signal to close trades. When price closes above/below the trend of the SuperTrend, the stop loss level is set one tick above/below the wick of the bar that broke the supertrend level. None: This selection means the stop loss defined by the Target Risk Per Trade ($) parameter will be used
ATM Strategy Template Name	Name of the ATM strategy template to use when Order Management Strategy is set to "ATM Strategy"
StopLoss Strategy	Multiple options available: Fixed Currency Value: fixed dollar stop loss Points: fixed points stop loss SuperTrend: dynamic stop based on SuperTrend bands Prediction Target Opposite: stop opposite to predicted target Last Swing Price: stop at most recent swing high/low
Swing Points Period	Period for identifying swing highs/lows when using "Last Swing Price" stop loss strategy
TakeProfit Strategy	Fixed Currency Value: This selection means the take profit price will be based on the fixed value of the parameter Take Profit Currency Points: The take profit price will be set as a number of Points from the entry price Stop Loss Ratio : The take profit win/loss ratio will be manually set by the user Predicted Target: The take profit price will be based on the predicted target by the Machine learning model
Stoploss/TakeProfit Ratio	When using "Stop Loss Ratio" take profit strategy, the ratio of take profit to stop loss distance. Example: 2.0 means TP = 2 x SL
Risk Per Trade ($)	This sets the risk in dollars for one position. This parameter is used before a position is open to determine if the StopLoss level calculated by the StopLoss Strategy respects this max risk, if not, the position will not be opened
Max Allowed Risk Per Trade ($)	This parameter is only used if the Quantity Strategy parameter is set to Risk Adjusted When the quantity is calculated based on a StopLoss level and a target Risk in $, the calculated quantity will be the quantity that gets as close as possible to generating the target Risk for the given stoploss, but rarely the exact target Risk. So for example if you set the Target Risk to $250, the strategy will calculate the quantity to 3 for example. This value of 3 will generate a real risk of lets say 257$ (in an ideal setting the quantity would have been 2.89 to generate exactly $250 risk, but that is impossible because futures contract size is not fractional). So this parameter will control how far you allow the real Risk to deviate from the Target Risk.
StopLoss Points	Stop loss in points when using Points-based stop loss
StopLoss Offset Points	Additional points to add/subtract to calculated stop loss for safety margin
TakeProfit Currency	Target take profit amount in dollars when using Fixed Currency take profit
TakeProfit Points	Target take profit in points when using Points-based take profit
On Opposite Signal	Action when opposite direction signal occurs: Breakeven Position | Close Position | Do Nothing
SuperTrend Settings
SuperTrend Mode	ATR: This sets the SuperTrend mode to ATR based calculation Adaptive: This sets the SuperTrend mode to an adaptative mode based on candle Close, High, Low
Period	This sets the supertrend period
Multiplier	This sets the Supertrend multiplier. The bigger this parameter the wider the superstrend bands
Moving Avg. Type	This selectes the Moving Average type for Supertrend
Smooth	This sets the smoothing period of the supertrend
Command Buttons
The strategy allows to manually execute some commands in case you want to manually intervene. The commands buttons are added to the Chart Trader panel, make sure to show it on your active chart
Arm/Disarm Buttons	The Arm Long/Disabled Long and Arm Short/Disabled Short buttons can manually activate or deactivate entering trades long or short. By default those two toggle buttons are enabled meaning the strategy is armed to enter both long and short trades. Toggling the button to the disable will restrain the strategy to enter a trade in the disabled direction. This can be useful if you have a directional bias or conviction about the direction of the market
Close Position Button	The Close position button will immediately flatten the position on the active instrument on the active Chart. This can be useful if you want to exit a trade without having to affect the strategy or restart it. After closing the position, the strategy will continue to function normally and look for new trades to take
Increase Qty Button	The increase Qty button allows to increase the quantity of the current active position by +1 contract
Decrease Qty Button	The decrease Qty button allows to decrease the quantity of the current active position by -1 contract
Set Breakeven Button	The button will set the stoploss of the open position to breakeven level if the current stoploss price is lower than breakeven price for long entries and higher than breakeven price for bearish entries.

Backtesting and Preferred Settings

Here I will try to periodically list backtests with different risk appetit. But before that here are some remarks regarding backtesting

• Past results are not guarantee of future results. All backtesting results are for indicative purposes only
• The period on which the backtest was run was not handpicked to show the best results
• Commissions are included in the backtests to get as realistic results as possible. The commission schema used is NinjaTrade Lifetime license commissions schema
• Since the AI model requires data of previous day, the signals will only be triggered starting from the second trading day on the selected interval. So if you run backtest on only one day you will not see any trades. Also make sure to enable the first parameter : Execute Historical Trades
• At automated-trading.ch we run very precise backtests due to the following 2 factors
• We are following the NinjaTrader official guide on how to achieve precise intrabar granularity backtesting by using additional tick level DataSeries See here for more info
• We are only using Market Orders instead of Pending Orders, in both real-time and becktest modes.

Backtest - MES - 1 Minute - Risk Profile - Defensive with $400 risk per trade

This template has the lead on performance on MES instrument. Good for 50k evaluation accounts.

• Instrument MES
• Timeframe 1 Minute
• Period From 01 January 2026 to 10 April 2026
• Run Mode Backtest
• Total Profit 12 138 $
• Max Drawdown (-1 841) $
• Run Release Version 1.0.0.0

Download Parameters Template

---

Righ-Click on the button -> Save Link As... into NinjaTrader templates folder ...\Documents\NinjaTrader 8\templates\Strategy\ATCHMachineLearningPredictor

• Analysis
• Monthly Analysis
• Summary
• Parameters
• Performance Data

Please make sure, after importing the template file to check that all parameters are properly imported by comparing the values of the parameters on your NinjaTrader to the ones one this image. The reason this is needed is that sometimes NinjaTrader does a glitch when importing Template files

Entry Time and Exit Time are on UTC+1 timezone

Show Order Details File on New Tab

Download & Installation Instructions

Step 1 : Downloading Random-Forest model files for each instrument

Training a Random-Forest model is specific to each instrument and timeframe. This is why we provide a model for each instrument on the 1 Minute timeframe

Step 2 : Downloading and Installing Microsoft.ML Dependencies

The Machine Learning Strategy requires Microsoft.ML runtime dependencies to function properly. These dependencies need to be installed in your NinjaTrader Custom bin directory.

Step 3 : Ninjatrader Strategy

Frequently Asked Questions

Release Notes

User Comments & Feedback

You can find feedback of our users and ask questions about this strategy by joining our discord community by following this invite link or clicking on the Discord logo image. Joining is completely free