gavindiaz/mql5
1
0
Fork 0
mirror of https://github.com/softwaredevelop/mql5.git synced 2026-08-19 15:28:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

refactor: VWAP-style Anchored Resets

Browse Source
This commit is contained in:
Toh4iem9
2026-06-18 13:25:31 +02:00
parent b56981b6ef
commit 46c1426b67
1 changed files with 41 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Indicators/MyIndicators/Quant/LLD_Pro.md
+41 -33
View File
@@ -2,14 +2,16 @@
## 1. Summary (Introduction)
The `LLD_Pro` (Lead-Lag Dominance Index) is an institutional-grade, high-frequency quantitative tool designed to detect and measure the lead-lag relationships between any two financial instruments in real-time. In financial markets, information does not flow instantaneously to all assets. Highly liquid market-representative instruments (such as indices like DXY or major assets like BTCUSD) often react to macroeconomic and sentiment shifts seconds or minutes before secondary correlated assets (like EURUSD, or altcoins like SUIUSD and ETHUSD) follow suit.
The **Lead-Lag Dominance Index Pro (LLDI Pro)** is an institutional-grade, high-frequency quantitative tool designed to detect and measure the temporal lead-lag relationships between any two financial instruments in real-time. In financial markets, information does not flow instantaneously to all assets. Highly liquid market-representative instruments (such as indices like DXY or major assets like BTCUSD) often react to macroeconomic and sentiment shifts seconds or minutes before secondary correlated assets (like EURUSD, or altcoins like SUIUSD and ETHUSD) follow suit.
By measuring the mathematical lead-lag relationship, `LLD_Pro` solves two of the most critical questions in cross-asset trading:
1. **Who is driving the current market regime?** (Which asset is leading and which is trailing?)
2. **What is the current execution window?** (By how many bars does the leader anticipate the follower?)
The indicator features a highly optimized $O(1)$ real-time mathematical engine, a high-precision 4-decimal status panel, and a bulletproof, gap-resistant bar-time synchronization module.
Featuring **VWAP-style Anchored Resets** (Session, Weekly, Monthly, and Custom Session), the indicator can completely isolate intraday/intraweek price relationships from overnight gaps and illiquidity. The entire suite features a highly optimized $O(1)$ real-time mathematical engine, a high-precision 5-decimal status panel, and a bulletproof, gap-resistant bar-time synchronization module.
---
## 2. Mathematical Foundations and Calculation Logic
@@ -25,7 +27,7 @@ $$R_{B,t} = \ln\left(\frac{P_{B,t}}{P_{B,t-1}}\right)$$
### B. Shifted Pearson Cross-Correlation (CCF)
For each bar $t$, the indicator computes the Pearson correlation coefficient ($r$) over a rolling window $W$ (`InpWindowSize`) using multiple lag offsets $k$ ($1 \le k \le \text{MaxLag}$):
For each bar $t$, the indicator computes the Pearson correlation coefficient ($r$) over a rolling or anchored window $W$ (`window_size`) using multiple lag offsets $k$ ($1 \le k \le \text{MaxLag}$):
* **Direction 1: Symbol B leads Symbol A** ($r_{B \rightarrow A}$)
Correlates the past of B with the present of A. The window of B is shifted backward by $k$ bars:
@@ -35,8 +37,6 @@ For each bar $t$, the indicator computes the Pearson correlation coefficient ($r
Correlates the past of A with the present of B. The window of A is shifted backward by $k$ bars:
$$r_{A \rightarrow B}(k) = \text{Corr}(R_{A, t-k \dots t-k-W}, R_{B, t \dots t-W})$$
### C. Dominance Index (LLDI) and Optimal Lag
The absolute peak (maximum) correlation is identified for both directions:
$$\text{Peak}_{B \rightarrow A} = \max_{k} |r_{B \rightarrow A}(k)| \quad \text{at optimal lag } L_B$$
$$\text{Peak}_{A \rightarrow B} = \max_{k} |r_{A \rightarrow B}(k)| \quad \text{at optimal lag } L_A$$
@@ -48,52 +48,60 @@ $$\text{LLDI}_t = \text{Peak}_{B \rightarrow A} - \text{Peak}_{A \rightarrow B}$
* **$\text{LLDI} < -0.02$ (Red Histogram):** Symbol A (Chart) is dominant and leads Symbol B (Secondary).
* **$\text{LLDI} \in [-0.02, 0.02]$ (Gray Histogram):** Symmetrical relationship (no clear leader).
The **Optimal Lag** represents the exact shift $L$ (expressed in bars) at which the absolute correlation is maximized. It is signed positive if Symbol B leads, and negative if Symbol A leads.
---
## 3. MQL5 Implementation Details
## 3. MQL5 UI & Architecture
* **Decoupled Architecture:**
The mathematical logic is kept completely separate in `LLD_Calculator.mqh` (encapsulated in `CLeadLagDominanceCalculator`), while the visual representation, buffers, and object drawing are handled by the lightweight `LLD_Pro.mq5` wrapper.
* **Decoupled Math Engine (`LLD_Calculator.mqh`):**
All wave tracking and correlation sweep calculations are encapsulated inside the highly optimized `CLeadLagDominanceCalculator` include class.
* **Real-Time $O(1)$ Tick-by-Tick Engine:**
In standard MT5 implementations, indicators are calculated once per bar, making real-time readings static inside the current candle. `LLD_Pro` solves this by modifying the incremental `start_index` passed to the calculator. Instead of using `prev_calculated` directly, it passes `prev_calculated - 1`. This forces the calculator to recalculate the returns and correlation metrics for the **current forming bar** (index `rates_total - 1`) on **every single tick**, maintaining high performance ($O(1)$) while ensuring the visual display and label are dynamically updating in real-time.
* **Single-Bar $O(1)$ Optimization (Anti-Freeze Guard):**
Running full historical cross-correlations inside nested loops in `OnCalculate` results in an $O(N^2)$ computational complexity, forcing up to 4.5 billion operations on startup, which freezes the MT5 terminal.
`LLD_Pro` solves this by modifying the calculator to process **only a single index (`current_index`)** per call. It computes the log-returns and Pearson sweeps exclusively for the requested bar, maintaining a strict $O(1)$ complexity on live ticks, ensuring instant and fluid execution.
* **Bulletproof Bar-Time Synchronization:**
`LLD_Pro` employs a high-performance alignment loop that maps and aligns the prices of Symbol B directly to the timestamp of Symbol A's bars using the native `iBarShift(..., false)` and `iClose` combination. It guarantees perfect chronological alignment on any timeframe and resolves data-gap blankness.
* **Unique Subwindow Labeling (Anti-Collision Guard):**
When multiple instances of the LLD indicator are applied to the same chart (e.g., comparing different symbols), standard global labels collide and overwrite each other. `LLD_Pro` resolves this by calling `ChartWindowFind()` and appending the specific subwindow index to the object prefix (e.g. `WYC_LLD_1234_Status_Sub_2`). This allows multiple status panels to coexist and render perfectly in their respective windows.
* **Timer-Driven Weekend Refresh:**
Asynchronous loading of historical data on weekends often causes standard indicators to get stuck on blank screens. `LLD_Pro` registers a 1-second system timer (`OnTimer`). If data synchronization fails initially, the timer continuously checks for data availability and triggers a `ChartRedraw()` once loaded.
* **Data Window Mapping with `DRAW_NONE`:**
To allow the trader to hover over any historical bar and see the precise `Optimal Lag` value inside the MT5 Data Window, the lag buffer is mapped to `INDICATOR_DATA` instead of `INDICATOR_CALCULATIONS`. However, its draw style is set to `DRAW_NONE` in `OnInit` so that it remains completely invisible on the chart, preventing any visual distortion of the separate window's vertical scale.
* **Symbol Selection Guard:**
Typing incorrect symbol names or mismatched broker suffixes (e.g., `BTCUSD.m` instead of `BTCUSD`) can cause indicators to fail silently. `LLD_Pro` implements an aggressive `SymbolExist` check in `OnInit()`. If the comparison symbol is invalid, it throws a pop-up **MT5 Alert** and gracefully stops execution (`INIT_FAILED`), preventing silent crashes.
---
## 4. Parameters
* **Comparison Symbol (`InpSecondSymbol`):** The secondary symbol to correlate with (e.g., `USDX` or `BTCUSD`). Default is `BTCUSD`.
* **Rolling Window (`InpWindowSize`):** The number of historical bars used to calculate each correlation point. Default is `50`.
* **Maximum Lag (`InpMaxLag`):** The maximum number of bar shifts tested. Determines the limit of the search window for leader/follower. Default is `10`.
* **Comparison Symbol (`InpSecondSymbol`):** The secondary symbol to correlate with (Default: `"BTCUSD"`).
* **Anchor Reset (`InpAnchor`):** The reset anchor period (None, Session, Week, Month, Custom Session).
* **Rolling Window (`InpWindowSize`):** The number of historical bars used to calculate each correlation point (Used if Anchor = None). Default is `50`.
* **Maximum Lag (`InpMaxLag`):** The maximum number of bar shifts tested. Default is `10`.
* **Custom Start (`InpCustomStart`):** Session start time in format "HH:MM" (Used if Anchor = Custom).
* **Custom End (`InpCustomEnd`):** Session end time in format "HH:MM" (Used if Anchor = Custom).
## 5. Usage and Interpretation
---
### A. Lag Scaling Across Timeframes
## 5. Advanced Statistical Arbitrage Strategies
The physical lag time is absolute, but its representation in bars depends on the chart timeframe. If the physical delay is 15 minutes:
### A. Dynamic Intraday Lead-Lag Transitions
* On the **M1** chart, the lag is represented as **15 bars** (`Optimal Lag = 15`).
* On the **M5** chart, the lag is represented as **3 bars** (`Optimal Lag = 3`).
* On the **M15** chart, the lag is represented as **1 bar** (`Optimal Lag = 1`).
* On the **H1** chart, the delay is too small to be represented as a phase shift. The indicator will show a **Symmetrical / Gray** regime (`Optimal Lag = 0`) because the sub-bar offset is compressed inside a single 1-hour candle.
Lead-lag relationships in financial markets are highly dynamic and rotate throughout the trading day based on session liquidity.
* **The Concept:** During the European session, `EURUSD` may lead `GBPUSD`. However, during the US session, `GBPUSD` may take the lead.
* **The Solution:** By applying **`ANCHOR_SESSION`** or **`ANCHOR_CUSTOM_SESSION`**, the indicator resets its data pool at the start of each active session. This filters out overnight noise and provides an absolute, real-time speedometer of who is leading the market *today*.
### B. The Contemporaneous Lockstep Phenomenon (Symmetry vs. Lead-Lag)
A significant quantitative insight occurs when pairing highly cointegrated assets within the same asset class (e.g. **BTCUSD vs. ETHUSD** or **BTCUSD vs. SUIUSD**):
A significant quantitative insight occurs when pairing highly cointegrated assets within the same asset class (e.g. **BTCUSD vs. ETHUSD**):
* Since major cryptocurrencies are highly integrated, their real-time correlations occur almost entirely at lag $k = 0$ (Contemporaneous Correlation).
* At shifted lags ($k \ge 1$), the predictive power of BTCUSD over ETHUSD is roughly equivalent to the predictive power of ETHUSD over BTCUSD ($\text{Peak}_{B \rightarrow A} \approx \text{Peak}_{A \rightarrow B}$).
* This mathematically yields $\text{LLDI} \approx 0.00000$ (Symmetrical/Gray). This flat reading is not a bug; it is a **mathematical proof of instantaneous market integration**.
* At shifted lags ($k \ge 1$), the predictive power of BTCUSD over ETHUSD is roughly equivalent to the predictive power of ETHUSD over BTCUSD ($\text{Peak}_{B \rightarrow A} \approx \text{Peak}_{A \rightarrow B}$). This mathematically yields $\text{LLDI} \approx 0.00000$ (Symmetrical/Gray). This flat reading is not a bug; it is a **mathematical proof of instantaneous market integration**.
* Conversely, when pairing uncorrelated or macro-driven assets (e.g. **DXY vs. BTCUSD**), clear asymmetric lead-lag relationships appear, producing non-zero, beautifully colored LLDI values with 5-decimal precision.
### C. Multi-Timeframe (MTF) Top-Down Strategy
### C. The Cointegration + LLD Pro Synergy (Single-Leg Trading)
To achieve high-probability execution, traders should apply a structured top-down filter across three timeframes:
By pairing `PairsTrading_Pro` (Cointegration Spread) and `LLD_Pro` (Lead-Lag) together, traders can execute highly efficient **single-leg directional trades**:
1. **The Compass (H1 Timeframe):** Verify if the H1 LLDI is strongly Blue (Second symbol leads) or Red (Chart symbol leads) to identify the major intraday driver.
2. **The Sniper Scope (M15 Timeframe):** Once the macro direction is confirmed, look at the M15 Optimal Lag to determine the delay window (e.g. `Optimal Lag = 3` = 45 minutes).
3. **The Trigger (M5 Timeframe):** Drop down to the M5 chart. Wait for the M5 LLDI to quickly spike into the dominant color, signaling that the lag is beginning to resolve on the micro-level.
1. `PairsTrading_Pro` signals that the spread is extremely cheap (Z-Score $\le -2.0$), meaning Asset A is underpriced relative to Asset B.
2. `LLD_Pro` signals that **Asset B is dominant and leads Asset A**.
3. *Action:* Since the leader (Asset B) has already moved up, and the follower (Asset A) is mathematically guaranteed to follow to close the spread gap, you simply **BUY Asset A as a single-leg directional trade**. This reduces execution margin requirements by 50% and completely avoids double-leg commissions!