mirror of
https://github.com/softwaredevelop/mql5.git
synced 2026-08-06 00:57:45 +00:00
Toh4iem9
78 lines
5.5 KiB
Markdown
78 lines
5.5 KiB
Markdown
# Holt Double Exponential Smoothing Professional
|
|
|
|
## 1. Summary (Introduction)
|
|
|
|
The Holt Double Exponential Smoothing (DES), or Holt's Linear Trend Method, is a powerful statistical technique for smoothing and forecasting time series data. Unlike traditional moving averages, Holt's method explicitly models and smooths two distinct components: the **Level** (average value) and the **Trend** (rate of change).
|
|
|
|
By separating these components, the Holt method produces an exceptionally smooth moving average that is also a **one-bar-ahead forecast** of the price level.
|
|
|
|
Our professional MQL5 suite provides a complete, unified family of indicators based on this model:
|
|
|
|
* **`Holt_Pro.mq5`**: A versatile chart indicator that can display either the main **Holt MA** line or a full **Holt Forecast Channel**.
|
|
* **`Holt_Oscillator_Pro.mq5`**: A separate-window indicator that plots the **Trend** component as a histogram, visualizing the trend's velocity.
|
|
|
|
Both indicators can be calculated using either **standard** or **Heikin Ashi** price data, selectable from a single input parameter.
|
|
|
|
## 2. Mathematical Foundations and Calculation Logic
|
|
|
|
The Holt method is a recursive system defined by two interconnected smoothing equations, controlled by two smoothing factors, `alpha` and `beta`.
|
|
|
|
### Required Components
|
|
|
|
* **Alpha ($\alpha$):** The smoothing factor for the **Level** component (0 < $\alpha$ < 1).
|
|
* **Beta ($\beta$):** The smoothing factor for the **Trend** component (0 < $\beta$ < 1).
|
|
* **Price Data:** The source price series (e.g., `PRICE_CLOSE`).
|
|
|
|
### Calculation Steps (Algorithm)
|
|
|
|
1. **Initialization:** The model requires initial values for the first Level ($L_0$) and Trend ($T_0$).
|
|
* $L_0 = \text{Price}_0$
|
|
* $T_0 = \text{Price}_1 - \text{Price}_0$
|
|
|
|
2. **Recursive Calculation:** For each subsequent bar `t`:
|
|
* **Level Equation:** $L_t = \alpha \cdot \text{Price}_t + (1 - \alpha) \cdot (L_{t-1} + T_{t-1})$
|
|
* **Trend Equation:** $T_t = \beta \cdot (L_t - L_{t-1}) + (1 - \beta) \cdot T_{t-1}$
|
|
|
|
3. **Forecasting:** The forecast for `m` periods ahead is:
|
|
* $\text{Forecast}_{t+m} = L_t + m \cdot T_t$
|
|
* The main **Holt MA** line is the one-bar-ahead forecast ($m=1$).
|
|
* The **Holt Channel** uses this formula with a user-defined `m`.
|
|
* The **Holt Oscillator** directly plots the `Trend` component ($T_t$).
|
|
|
|
## 3. MQL5 Implementation Details
|
|
|
|
Our MQL5 implementation is a cohesive and reusable indicator family, built upon a single, robust calculation engine.
|
|
|
|
* **Centralized Calculation Engine (`Holt_Engine.mqh`):**
|
|
The core of our implementation is a single, powerful calculation engine. This include file contains the complete logic for calculating all Holt components (Level, Trend, Forecast, and Channel bands). It supports both standard and Heikin Ashi data sources through class inheritance (`CHoltEngine` and `CHoltEngine_HA`), eliminating code duplication.
|
|
|
|
* **Specialized Wrappers (`Holt_Calculator.mqh`, `Holt_Oscillator_Calculator.mqh`):**
|
|
The final indicators use thin "wrapper" classes that utilize the central engine.
|
|
* `Holt_Calculator` calls the engine and provides the outputs needed for the `Holt_Pro` indicator (MA and Channel bands).
|
|
* `Holt_Oscillator_Calculator` calls the same engine but only extracts the `Trend` component for the `Holt_Oscillator_Pro` indicator.
|
|
|
|
* **Stability via Full Recalculation:** The Holt method is highly recursive. To ensure perfect accuracy and prevent calculation errors, all our Holt indicators employ a "brute-force" **full recalculation** on every tick.
|
|
|
|
## 4. Parameters
|
|
|
|
* **Period (`InpPeriod`):** Used for the robust initialization of the model. Default is `20`.
|
|
* **Alpha (`InpAlpha`):** The smoothing factor for the Level. Lower values create a smoother level. Default is `0.1`.
|
|
* **Beta (`InpBeta`):** The smoothing factor for the Trend. Lower values create a more stable trend component. Default is `0.05`.
|
|
* **Source Price (`InpSourcePrice`):** The price data used for the calculation. This unified dropdown menu allows you to select from all standard and Heikin Ashi price types.
|
|
* **Display Mode (`InpDisplayMode`):** (Only for `Holt_Pro`) Toggles between displaying only the MA line or the full MA + Channel.
|
|
* **Forecast Period (`InpForecastPeriod`):** (Only for `Holt_Pro` in Channel mode) The number of bars (`m`) to project into the future for the channel bands. Default is `5`.
|
|
|
|
## 5. Usage and Interpretation
|
|
|
|
### `Holt_Pro` (The Main Line and Channel)
|
|
|
|
* **Trend Filter:** The slope of the Holt MA line indicates the smoothed, forecasted trend direction. A flattening line is an early warning that the trend is losing momentum.
|
|
* **Dynamic Support/Resistance:** The line acts as an intelligent S/R level that projects where the "fair value" should be in the next bar.
|
|
* **Channel as Forecast Path:** The channel visualizes the path the price is expected to follow if the current trend persists. When price touches or exceeds the outer channel bands, it signals that the move is potentially over-extended.
|
|
|
|
### `Holt_Oscillator_Pro` (The Trend Engine)
|
|
|
|
* **Trend Direction:** Values above the zero line indicate a bullish trend; values below indicate a bearish trend. A crossover of the zero line is a strong trend-change signal.
|
|
* **Trend Velocity:** The magnitude of the oscillator's bars indicates the speed of the trend. Rising bars show acceleration; falling bars (while still on the same side of zero) show deceleration.
|
|
* **Divergence:** Divergence between the oscillator and price can signal trend exhaustion and an impending reversal.
|