RB Anomali input Explain

1. CORE SETTINGS GROUP (Basic Sensor)

--- Model (Basic model options for the Machine) ---

• Essence: Selecting a machine model that will be the basis for learning for the AI ​​Matrix.

--- InpPeriod (Average & Z-Score Window) --- 

• Essence: Determines how many bars back are used as a basis for calculating the current market fair (average) value.

• The Bigger (Example: 50 or 100):

  • Impact: The indicator graph becomes very smooth and immune to random movement noise .

  • Trading Effect: Signals become slower to come out ( lagging ), suitable for reading long-term trends ( swing trading ).

• The Smaller (Example: 5 or 10):

  • Impact: The indicator becomes very aggressive and sensitive to even the slightest price fluctuations.

  • Trading Effect: Signals come out lightning fast, but are prone to triggering false signals ( fakeouts ) because the market often only makes momentary corrections.

2. MACHINE LEARNING GROUP (AI Brain)

--- InpRetrainBars (Relearning Interval) ---

• Essence: Controls how often the AI ​​should “open the history book” to recycle the contents of its brain’s memory.

• The Bigger (Example: 500 bars once):

  • Impact: The AI ​​will be reluctant to update its knowledge. Your laptop's processor will be very light during backtesting.

  • Trading Effect: If the market suddenly changes character (for example, from a fast trend season to a consolidation/sideways season), the AI ​​will be late to realize it and will continue to use the old, outdated logic.

• Smaller (Example: 20 bars once):

  • Impact: AI becomes super adaptive and constantly adjusts to the latest price characteristics.

  • Trading Effect: The strategy is always fresh , but backtesting will feel heavier because the CPU is forced to open training classes repeatedly in a short time.

--- InpTrainLimit (Number of History Practice Bars) ---

• Essence: Determines how thick a history book from the past the AI ​​must read during the crash training.

• The Bigger (Example: 1000 bar):

  • Impact: AI becomes wiser because it sees very long data patterns.

  • Trading Effect: AI's guessing logic becomes more robust and stable, less prone to panic by momentary fluctuations.

• Smaller (Example: 100 bar):

  • Impact: AI only focuses on the immediate conditions.

  • Trading Effect: Great for capturing instant momentum changes, but too small a sample size makes the AI ​​prone to “misunderstanding” the larger market structure.

--- InpEpochs (Practice Round) ---

• Reality: How many times was the AI ​​forced to reread past question and answer sheets until it memorized the pattern.

• The Bigger (Example: 500 spins):

  • Impact: AI will memorize historical data in detail down to the smallest points ( Overfitting ).

  • Trading Effect: In the past (history) the accuracy will appear to be 100% perfect, but when faced with the current bar ( live running ), the AI ​​will be confused and its performance will plummet because the market never repeats with the exact same details.

• Smaller (Example: 10 turns):

  • Impact: AI is undertrained ( underfitting ).

  • Trading Effects: The matrix weights are not yet mature, the indicator's guesses tend to be haphazard and reluctant to move to reach extreme areas.

--- InpAlphaLR (Learning Rate) ---

• Essence: Controls how drastically the computer is allowed to change its brain weight values ​​every time it realizes its guess is wrong.

• The Bigger (Example: 0.1):

  • Impact: AI is very reactive and aggressively changes its mind when it guesses wrong.

  • Trading Effect: The mathematical formula in the background is prone to overshoot , causing the weight values ​​to explode into NaN (the indicator line disappears from the chart).

• The Smaller (Example: 0.0001):

  • Impact: AI is very careful, calm, and subtle when correcting its mistakes.

  • Trading Effect: The learning process becomes very stable, but requires a InpEpochsmuch larger value for the brain to reach an optimal level of intelligence.

3. TARGET & FILTER STRATEGY GROUP (Signal Filter)

--- InpTargetTP (Pips) & InpHorizonBars (Future Bars) ---

• Essential: Setting the key success benchmarks. The combination of these two parameters determines how high your trading target expectations are.

• Big TP + Small Horizon (Example: 100 Pips in 3 Bars):

  • Impact: Expectations are too high in a short period of time.

  • Trading Effect: The answer key is almost always 0 (not reached). The AI ​​will conclude there is no opportunity in that pair, so the indicator rarely or never displays an arrow at all .

• Small TP + Large Horizon (Example: 5 Pips in 20 Bars):

  • Impact: Targets are too easy to achieve.

  • Trading Effect: The AI ​​will become overly optimistic, considering almost every movement a profitable opportunity. As a result, the indicator will be flooded with arrow signals everywhere.

--- InpThresholdBull & InpThresholdBear (Z-Score Filter Limit) ---

• Essentials: Filter faucet to determine how confident the AI ​​must be before it is allowed to print a Buy or Sell arrow on the chart.

• Further Away from 0 (Example: 3.0 and -3.0):

  • Impact: The filter becomes super strict institutional standards ( High Conviction ).

  • Trading Effect: The arrow signals that come out are very few , but have a very high level of accuracy and probability certainty.

• Closer to 0 (Example: 1.0 and -1.0):

• Impact: The filter becomes very loose.

• Trading Effect: Arrow signals are continuously being released on the chart, but the signal quality is drastically reduced due to capturing a lot of market consolidation ( noise ) movements.
  
  
  

Quick Summary for Optimization (Chef Cheat Sheet)

If you want...	So the Best Parameter Solution Is...
More Accurate & Valid Signals	Increase InpThresholdBull(e.g. to 2.5 ) & Thicken InpTrainLimit(e.g. to 800 ).
Super Fast Backtest (Light CPU)	Increase InpRetrainBars(eg to 200 ) & Decrease InpEpochs(eg to 50 ).
Short Term Scalping Style	Shrink InpPeriod(to 10 ), Shrink InpTargetTP( 10.0 ), & Shrink InpHorizonBars( 3 ).