Manual liquidity zone work on higher timeframes misses the internal structure of the base candle. Verifying whether a zone was built by a triangle, rectangle, or double top/bottom requires repeated lower-timeframe zooming, which is slow and produces inconsistent labeling.

An automation module is proposed to classify the lower-timeframe geometry inside each base candle as ascending triangle, descending triangle, symmetrical triangle, rectangle, M, W, or undefined, then annotate the zone and alerts with the result.

Implementation outline targets MQL5: isolate detection logic in an include file (CGeometryDetector), add configurable tolerances and swing-distance filtering, compute slopes for symmetry, and integrate by extracting intrabar data for each base candle interval and storing the detected shape per zone.

👉 Read | AppStore | @mql5dev

#MQL5 #MT5 #Indicator

Deterministic Oscillatory Search (DOS) is a reproducible metaheuristic for global optimization that removes randomness entirely. Particles start from deterministically distributed points, making runs repeatable under identical inputs—useful for research and trading system validation.

Search is driven by a “fitness slope” state: improving, worsening, or unknown. When fitness deteriorates, a particle reflects (reverses direction) and halves velocity, producing controlled oscillations that refine local extrema without derivatives.

If oscillations stall (worsening persists), DOS switches to a swarm step, pushing particles toward the current global best using a configurable movement factor. An MT5-style implementation centers on per-particle velocity vectors, range clamping, best-solution tracking, and adaptive velocity updates.

Tests show stronger behav...

👉 Read | Quotes | @mql5dev

#MQL5 #MT5 #algorithm

DADA (Adaptive Bottlenecks + Dual Adversarial Decoders) targets time-series anomaly detection with adaptive compression, dual reconstructions for normal vs anomalous regimes, and patching/masking to reduce noise and improve generalization.

The Adaptive Bottlenecks block is implemented as CNeuronAdaBN inheriting CNeuronMoE. It reuses top-k gating and expert routing, but populates experts with autoencoders at multiple latent sizes. A convolution stage produces a latent vector whose segments map to each autoencoder, followed by multi-window conv, tensor transpositions, and per-autoencoder decoder weights.

Instead of a monolith, the system is assembled from library components into three jointly trained models: a state encoder autoencoder with patching/masking and AdaBN, an Actor replacing the anomalous decoder for action selection, and a direction predictor as a deci...

👉 Read | Calendar | @mql5dev

#MQL5 #MT5 #AI

An Expert Advisor can automate session control by recalculating daily start, end, and close times, clearing internal state, and initializing the price range used for breakout logic.

During the range window, minute-level highs and lows are sampled to compute the session maximum and minimum. A chart rectangle is updated in real time to reflect the current consolidation zone.

After the range window completes, the close of the latest finished candle is compared to the stored boundaries, independent of the range end timestamp. On confirmation, a market order is placed in the breakout direction, with take-profit set to the measured range size and stop-loss set to the opposite boundary.

👉 Read | VPS | @mql5dev

#MQL5 #MT5 #EA

L1 trend filtering can be used to estimate piecewise-linear price trends while suppressing short-term noise. The method keeps turning points and slope changes that are often lost with moving averages or heavy smoothing.

An example implementation in MQL5 demonstrates L1 Trend Filter routines for float and double vectors, validated on random-walk simulated series. This setup helps verify numerical stability, parameter sensitivity, and output consistency across data types.

In trading workflows, the filtered trend can support regime detection, adaptive risk controls, and signal preconditioning by separating structural movement from micro-variations. The same code is packaged as a shared project under MQL5\Shared Projects\L1Trend.

👉 Read | Calendar | @mql5dev

#MQL5 #MT5 #Indicator

CATCH targets subtle anomalies in multivariate market series by moving analysis to the frequency domain, where volatility bursts and regime shifts separate into different bands. Its key idea is “frequency patching”: split the complex FFT spectrum into overlapping patches, learn normal behavior per band, then reconstruct with iFFT and flag anomalies via reconstruction error.

A channel-aware fusion stage uses a masked Transformer to model cross-asset dependencies without letting irrelevant instruments dilute attention. The mask is learned and refined with an explicit objective to strengthen meaningful inter-channel links while suppressing noise.

The article then maps these ideas into MQL5 with OpenCL acceleration: complex-valued convolution and masked attention are implemented using float2 buffers for real/imag parts, reusing existing layer infrastr...

👉 Read | AlgoBook | @mql5dev

#MQL5 #MT5 #AlgoTrading

Strategy Tester gaps around CalendarValueHistory() make news-driven EAs hard to verify. Historical events are often missing, so entry blocks, SL/TP suspension, and pre-news closes never trigger, producing misleading curves and no audit trail in logs.

A deterministic fix is a static CSV of events (time, currency, importance, name) loaded once in OnInit() when MQL_TESTER is true. Tester mode switches isUpcomingNews()/IsNewsTime() to an in-memory scan, while live trading keeps the terminal calendar API unchanged.

Implementation points: strict CSV format compatible with StringToTime(), FILE_COMMON access for tester sandbox, a symbol currency cache built at init, and an optional script that exports the terminal calendar to CSV for chosen date ranges.

👉 Read | NeuroBook | @mql5dev

#MQL5 #MT5 #AlgoTrading

Price action systems often fail on breakout filtering, with liquidity sweeps creating false continuation signals. A structured model is required to locate likely resting orders, validate breaks, and keep risk rules consistent.

An MQL5 automation is outlined for order block trading in consolidation zones, using higher-timeframe swing structure as the trend filter. Setups require inducement first, then break of structure, with fair value gap alignment inside the impulse leg.

Implementation details cover enums for trade mode, FVG state, trailing type, and mitigated-zone handling. Core structs track OB metadata, FVG links, and open-trade trailing state, plus chart rendering utilities for zones, labels, BoS lines, and mitigation marks. Swing and consolidation functions drive detection, followed by zone creation, mitigation tracking, and risk-based execu...

👉 Read | Docs | @mql5dev

#MQL5 #MT5 #AlgoTrading

MetaTrader 5 keeps trade history inside the terminal, so external analytics need an explicit export path. This article completes that link by adding a lightweight EA that emits a trade record the moment a position is closed.

The core design is event-driven: OnTradeTransaction filters only “deal added” events that represent exits, avoiding fragile OnTick polling and ensuring only final, complete trades are sent.

The EA reconstructs a full trade by pairing the closing deal with its opening deal, normalizes enums (reason, direction) into readable strings, builds JSON manually (no native MQL5 JSON), then POSTs it via WebRequest to a versioned Flask API endpoint.

Local testing covers server startup, MT5 WebRequest URL whitelisting, and verifying 200 responses and server logs. Known gaps include no retries, no payload validation, and no local queueing for outa...

👉 Read | AppStore | @mql5dev

#MQL5 #MT5 #EA

MQL5 EAs often persist input changes by rewriting a single settings file, which removes history and makes past test results hard to reproduce.

A practical alternative is an append-only binary log. Each parameter snapshot is stored as a fixed-size struct record, enabling fast reads of the latest configuration while keeping prior versions intact.

The struct is typically split into metadata (version, timestamp), adjustable trading inputs, and an integrity field. A checksum is computed only from adjustable inputs so timestamps, counters, and performance metrics do not trigger new versions.

On startup, create version 1 only if the file is empty. On later runs, read the last record, recompute the checksum from current inputs, and append a new record only when the checksum differs.

👉 Read | CodeBase | @mql5dev

#MQL5 #MT5 #EA

A confluence detector identifies price zones where Fibonacci retracement levels from multiple swing pairs align.

The algorithm selects 4–5 significant swing points, calculates 38.2%, 50%, 61.8%, and 78.6% levels for each swing pair, then scans for clusters where two or more levels converge within a configurable tolerance. Zones are ranked by density, with more overlapping levels treated as higher strength.

Confluence areas are rendered as shaded rectangles and color-coded by strength: yellow for 2 matches, orange for 3, and red for 4 or more. Labels list the contributing levels, while individual Fibonacci lines can be enabled or hidden. An alert triggers when price enters a strong zone.

Runs on all timeframes, with typical use on H1, H4, and D1.

👉 Read | Docs | @mql5dev

#MQL4 #MT4 #Indicator

Daily Risk Monitor Lite is a free, open-source MT5 indicator that renders intraday account risk directly on the chart with a small, explainable feature set. The panel shows Daily Realized P/L, Floating P/L, Daily Total, and current drawdown percent, plus SAFE/WARNING/DANGER status via configurable colors.

Daily Realized P/L counts only exit deals within the active day range, with optional commission and swap inclusion. Floating P/L uses the current result of all open positions, optionally including swap. Daily Total is realized plus floating. Drawdown is max((Balance-Equity)/Balance*100, 0), or N/A when Balance<=0.

Day boundaries can follow broker 00:00 or a manual server-hour start. This is read-only: no auto-close, no trade blocking, no enforcement engine. Intended as a lightweight sample for monitoring and further development.

👉 Read | Quotes | @mql5dev

#MQL5 #MT5 #Indicator

Self-Aware Trend System (SATS) is a multi-engine SuperTrend for MT5 built around four parts: adaptive ATR SuperTrend bands, a Trend Quality Index (TQI), composite signal scoring, and on-chart risk levels (entry, SL, TP1–TP3).

Band width adapts with Kaufman Efficiency Ratio: tighter in directional moves, wider in chop. TQI (0–1) combines efficiency, volatility regime (ATR vs baseline), pivot-based structure, and momentum persistence. A character-flip module flags sharp TQI drops during an active trend as an early warning before a line flip.

Signals are gated by a configurable score (momentum, ER, volume Z-score, RSI zone, pivots). The dashboard tracks TQI components plus rolling win rate, average R, drawdown, streaks, and cumulative R. Optional auto-calibration adjusts “Quality Influence” from recent R results. Non-repainting uses confirmed closed bar...

👉 Read | Freelance | @mql5dev

#MQL5 #MT5 #Indicator

Options theory is framed around clear mechanics (call/put rights vs seller obligations), key parameters (underlying, strike, premium, expiration), and exercise/settlement styles. Pricing is anchored to Black-Scholes for European options, with practical caveats: shifting volatility, non-lognormal returns, and limited fit for American exercise. Sensitivities are handled via Greeks, especially Delta for replication.

The core idea is emulating options by trading the underlying to match an option portfolio’s Delta, enabling synthetic contracts with custom strikes/expirations—even when listed options don’t exist or are illiquid. Delta-hedged rebalancing (time-based or delta-step) keeps the synthetic payoff aligned, but requires systematic automation, low friction costs, and continuous monitoring to avoid tracking error and missed adjustments.

👉 Read | Signals | @mql5dev

#MQL5 #MT5 #EA

CAPM emerged from Markowitz’s 1952 portfolio theory and the efficient frontier. In 1964, Sharpe reduced the computational burden of covariance-heavy optimization into a market equilibrium model; Lintner and Mossin reached similar results. By the 1970s, CAPM became standard for cost of capital estimates.

Classical CAPM links expected return to the risk-free rate and beta, with beta defined via covariance with the market over market variance. Assumptions include frictionless markets, shared expectations, and diversified portfolios pricing only systematic risk.

A MetaTrader 5 adaptation for FX replaces beta with a volatility-driven dynamic risk premium, uses unbiased variance, handles low-data cases, and annualizes volatility with a 252 factor. Two buffers expose expected return and risk premium, using CopyClose for efficient data access. Limitations includ...

👉 Read | NeuroBook | @mql5dev

#MQL5 #MT5 #Indicator

Source text is missing.

Paste the text to convert, and note the target platform if needed (e.g., LinkedIn, X).

👉 Read | Freelance | @mql5dev

#MQL4 #MT4 #EA

AI agents expanded quickly in early 2026, with mature automation on the crypto side. MetaTrader 5 remains largely unsupported by agent toolchains, despite visible demand in OpenClaw feature requests and trader forums.

A practical approach is an MCP (Model Context Protocol) server that bridges AI clients to MT5 via stdio. MCP standardizes tool discovery and typed calls, avoiding per-client plugin formats while keeping execution local.

The proposed Python design uses the official MetaTrader5 library plus FastMCP, exposing 14 tools across account, market data, positions, orders, and history. A wrapper layer manages initialization, login checks, timeouts, constant mapping, and order request normalization for mt5.order_send().

👉 Read | VPS | @mql5dev

#MQL5 #MT5 #AI

MetaTrader 5 canvas work can exceed static rendering by adding a timer-driven animation state machine. A four-phase sequence can progressively draw the parametric outline (t from 0 to 12π), fade in wing fills, fade in veins/scales/body, then switch to continuous flight.

Flight motion can be composed from independent oscillators: sine-based wing flap scaling on X toward the body axis, vertical bobbing and horizontal sway offsets, and a small tilt shear. Visual enhancements can include multi-pass glow strokes and per-frame HSV hue rotation.

Implementation in MQL5 typically adds an animation enum, input groups for speeds/amplitudes, and global accumulators for phase, opacity, oscillator phases, and hue shift. Core helpers include ShiftHue (RGB↔HSV), glow pixel falloff, a single ApplyFlyingTransform, and centralized point collection feeding outlines, ...

👉 Read | CodeBase | @mql5dev

#MQL5 #MT5 #AlgoTrading

ARCH/GARCH models handle volatility clustering but are symmetric by construction. Squared shocks remove sign, so rallies and drawdowns have identical impact on forward variance. Empirical data shows leverage effects where negative returns increase conditional risk more than positive returns.

Two common fixes are GJR-GARCH and TARCH. GJR-GARCH adds an indicator term that activates on negative returns, scaling shock impact via an asymmetry parameter. TARCH applies the threshold idea to conditional standard deviation, using absolute shocks with a sign-dependent slope.

An MQL5 volatility library can implement both as CGarchProcess derivatives, selectable through an ENUM_VOLATILITY_MODEL flag. Example scripts and an indicator can recalibrate on a rolling window and emit one-step forecasts plus a volatility z-score for anomaly detection.

Model choice is pract...

👉 Read | Quotes | @mql5dev

#MQL5 #MT5 #GARCH

Manual horizontal support and resistance remains effective, but it is static and requires continuous chart watching. This creates a monitoring gap on fast markets or multi-chart setups.

An MQL5 Expert Advisor can track manually drawn OBJ_HLINE levels in real time, compare bid price to each level, and classify events as touch, breakout, reversal, and retest. Levels are treated as discrete prices for deterministic checks, with per-line state (side, touched, breakout flags) to avoid duplicate alerts.

Implementation uses OnInit/OnDeinit for setup and cleanup, OnChartEvent for sync and deletion handling, and OnTick as the main loop. UI buttons manage sync and reset. Signals include labels, arrows with cooldown and one-per-bar rules, plus alerts and optional candlestick reversal pattern filters.

👉 Read | Calendar | @mql5dev

#MQL5 #MT5 #EA