mirror of
https://github.com/softwaredevelop/mql5.git
synced 2026-07-27 20:47:44 +00:00
51 lines
3.1 KiB
Markdown
51 lines
3.1 KiB
Markdown
|
# Hurst Exponent Pro (Indicator)
|
||
|
|
|
||
|
|
## 1. Summary
|
||
|
|
|
||
|
|
**Hurst Exponent Pro** is an advanced quantitative indicator derived from Fractal Geometry and Chaos Theory. It measures the **Long-Term Memory** of a time series.
|
||
|
|
|
||
|
|
Unlike standard technical indicators that measure price or momentum, the Hurst Exponent measures the **predictability** and **persistence** of the market structure itself. It answers the fundamental question: *"Is the current price movement trending (persistent), mean-reverting (anti-persistent), or completely random?"*
|
||
|
|
|
||
|
|
## 2. Methodology
|
||
|
|
|
||
|
|
The indicator offers two distinct calculation engines:
|
||
|
|
|
||
|
|
### A. Detrended Fluctuation Analysis (DFA) - *Recommended*
|
||
|
|
|
||
|
|
* **Method:** A robust algorithm that removes local polynomial trends from the data before analyzing the fluctuations.
|
||
|
|
* **Strength:** It is highly resistant to "non-stationary" noise (e.g., sudden volatility spikes or drifts), making it the preferred choice for modern financial markets (Forex, Crypto).
|
||
|
|
* **Output:** Provides a smoother, more reliable estimate of the fractal dimension.
|
||
|
|
|
||
|
|
### B. Rescaled Range Analysis (R/S) - *Classic*
|
||
|
|
|
||
|
|
* **Method:** The original method developed by H.E. Hurst. It scales the range of price deviations by their standard deviation.
|
||
|
|
* **Use Case:** Useful for historical comparison or analyzing simpler datasets, but tends to overestimate trend strength in noisy markets.
|
||
|
|
|
||
|
|
**Calculation Core:** Both methods utilize **Log-Log Regression** over multiple time scales (e.g., 8, 16, 32, ..., 128 bars) to determine the slope ($H$), ensuring statistical validity.
|
||
|
|
|
||
|
|
## 3. Interpretation (The Hurst Values)
|
||
|
|
|
||
|
|
The indicator oscillates between **0.0** and **1.0**.
|
||
|
|
|
||
|
|
| Hurst Value ($H$) | Market Regime | Interpretation & Strategy |
|
||
|
|
| :--- | :--- | :--- |
|
||
|
|
| **$0.5$** | **Random Walk** (Brownian Motion) | The market is unpredictable (50/50). Price changes are independent. **Avoid trading.** |
|
||
|
|
| **$> 0.5$** (0.6 - 1.0) | **Persistent** (Trending) | The market has "memory". A positive move is likely to be followed by another positive move. **Trend Following strategies work best here.** |
|
||
|
|
| **$< 0.5$** (0.0 - 0.4) | **Anti-Persistent** (Mean Reverting) | The market "fights" the move. A positive move is likely to be followed by a reversal. **Oscillator/Reversion strategies work best here.** |
|
||
|
|
|
||
|
|
## 4. Parameters
|
||
|
|
|
||
|
|
* `InpPeriod`: The lookback window for the analysis (Default: `256`).
|
||
|
|
* *Note:* The Hurst Exponent is a statistical measure that requires a large sample size. Values below 100 are statistically unstable. We recommend using 256 or 512 bars for reliable results.
|
||
|
|
* `InpMethod`: Switch between `METHOD_DFA` (default) and `METHOD_RS_CLASSIC`.
|
||
|
|
* `InpPrice`: The price source (Close, High, Low, etc.).
|
||
|
|
|
||
|
|
## 5. Usage in Trading
|
||
|
|
|
||
|
|
1. **Trend Filter:**
|
||
|
|
Only enter trend-following trades (e.g., MA crossovers) when **Hurst > 0.6**. This filters out "whipsaw" losses in random markets.
|
||
|
|
2. **Mean Reversion Filter:**
|
||
|
|
Only look for tops/bottoms (e.g., Bollinger Band bounces) when **Hurst < 0.4**.
|
||
|
|
3. **Regime Change:**
|
||
|
|
Watch for the Hurst line crossing the **0.5** level. This signals a fundamental shift in market structure (e.g., from a trending state into a random consolidation).
|