PulseStrike is a tick-driven scalper built around burst detection, not candle close. It maintains a rolling baseline over a short window (default 4s) and treats a move as tradeable only when it is a statistical outlier versus that baseline (z-score, default 3.0). This replaced a fixed ATR-fraction trigger that over-traded normal tick noise.

Two execution modes are supported: momentum (with the burst) and reversion (against it). Symbol fit is not assumed; both modes should be tested.

Entries are gated by spread-aware TP sizing, ATR-scaled SL/TP, a max hold-time force close, plus daily trade and daily loss caps. Single-position operation is used to behave correctly on netting accounts, cycling trades with tick-level checks and a short cooldown.

Backtest (M1, every tick, random delay, 2026-01-01 to 2026-09-08, balance 10k): EURUSD PF 1.30 DD 10.33%; AUDUSD PF ...

👉 Read | NeuroBook | @mql5dev

A new MQL5 signal class, CSignalIsotonicPNN, implements a two-stage confidence model for oscillator-based entries. Seven RSI/Stochastic plus price-context interpretations output a bounded directional score in [0,1], with 0.5 as neutral.

Isotonic regression calibrates the score into an ordered probability using a rolling calibration window and forecast horizon, with safeguards against lookahead. An optional PNN then blends a posterior based on similarity to historical bullish/bearish states, controlled by PNNSamples, PNNSigma, and PNNWeight.

The design keeps each mode independently testable and separates ranking quality from calibration. Invalid inputs and degenerate computations return 0.5 to avoid accidental directional bias.

👉 Read | Signals | @mql5dev

This MQL5 script turns closed deal history into trade-level analytics that avoid the common win-rate trap. Deals are grouped by position into a single record per round trip, with explicit “defined” flags so missing metrics never masquerade as zeros.

The core metric, Trade Quality Score, uses expectancy but substitutes the Wilson lower bound for win rate to penalize small samples. The conservative expectancy is normalized by average loss as a risk proxy, yielding a dimensionless score comparable across symbols and account sizes.

Architecture is modular: history reader, optional hour-of-day session filter (midnight-safe), pure calculator, and a CCanvas dashboard plus Experts-tab report. A dedicated test script validates pip conversion, edge cases, Wilson math, thresholds, and session boundaries with synthetic trades.

👉 Read | NeuroBook | @mql5dev

Multi-chart EAs often act as if they are the only process in the account. The broker enforces margin, margin level, and liquidation at account scope, so individually well-sized trades can still combine into a full drawdown.

A shared PortfolioRisk.mqh moves portfolio measurement out of any single EA. It supports account-wide scope or a magic-number filter, scans positions plus pending orders, and builds currency-leg exposure without parsing symbol names.

Before opening a trade, CanOpenPosition() evaluates six limits on the “would be” state: total trades, per-symbol count, distinct symbols, margin use, floating loss, and per-currency net exposure. Pearson correlation is noted but intentionally excluded; currency decomposition is deterministic, history-free, and consistent across EAs.

👉 Read | AlgoBook | @mql5dev

Portfolio eigenvalues describe how total variance is split across independent risk factors, but raw spectra don’t clearly indicate whether diversification is real or just cosmetic. This workflow turns the eigenvalue proportions into a single diversification score using spectral entropy (Shannon entropy normalized to [0,1]): high values mean variance is evenly distributed; low values indicate one dominant driver.

A reusable MQL5 script computes the covariance matrix, extracts and sorts eigenvalues safely (ArraySort over unreliable vector.Sort), converts them to variance shares, then outputs H_norm, dominant-factor percentage, an ASCII bar chart, and a thresholded concentration verdict for side-by-side portfolio comparison.

A key takeaway: diversification is governed by covariance structure, not instrument labels. Adding an uncorrelated but high-vola...

👉 Read | AppStore | @mql5dev

Work continued on hardening an automated MT5 optimization pipeline rather than adding trading logic. Core library and strategy-specific project code were further separated, so new ideas can be tested by changing project parameters without editing the shared library.

Optimization tasks now support time limits to cap end-to-end runtime and stop early once enough strong candidates exist. The optimization EA UI was expanded to show stage, symbol, timeframe, elapsed/remaining time, and overall progress.

CConsoleDialog was fixed to avoid duplicated windows after restarts by correcting OnDeinit cleanup. Chart elements behind the UI were disabled to avoid font rendering issues and remove the need for window minimization, requiring small local copies of standard library dialog classes.

Next focus: running multiple instances of the final multi-currency EA across dif...

👉 Read | Forum | @mql5dev

MT5 historical demonstration on XAUUSD (RoboForex-ECN), 1 May–8 Sep 2026, using 100% real ticks. Test settings: USD 10,000 deposit, 1:100 leverage, fixed 0.01 lot, default parameters (v0.11 defaults match the demonstrated set; trading logic unchanged vs v0.10). Result is optimized history, not a forward test.

Performance summary: 220 trades, net profit USD 1,293.99, profit factor 1.67, max equity drawdown USD 203.00 (1.91%), win rate 28.64%. Both long and short sides were net positive. Low hit rate included 13 consecutive losses, with expectancy driven by larger winners.

Inputs: brick size 6.0 (price units), momentum 8, threshold 5.0, TP 10 bricks, SL 2 bricks, max hold 1440 min, cooldown 3 bricks, max spread/brick 0.35, Magic 26091043.

Operational notes: validate in Strategy Tester. Brick size is price units, not points. Virtual SL/TP require terminal uptim...

👉 Read | Forum | @mql5dev

TQNet targets multivariate market forecasting by combining fast reaction to current conditions with a learned “global memory” of stable cross-asset relationships. Instead of building attention queries from raw prices, it uses trainable vectors that shift cyclically over time, while keys/values still come from the live input window. This balances persistent structure (seasonality, recurring liquidity cycles) with local shocks and noise.

Architecturally, it stays lightweight: one multi-head attention block plus a shallow MLP with residual connections, then a linear projection to any forecast horizon. RevIN normalization is used to handle distribution shifts common in finance, keeping the model focused on patterns rather than changing scale/volatility.

The article also outlines an MQL5-oriented implementation path and positions TQNet as a practical a...

👉 Read | AppStore | @mql5dev

AurumNeuro Vanguard is an Expert Advisor focused on XAUUSD, built around a hybrid neural risk design and a Unified Market Dynamics Engine. Signal generation combines causal price analysis, online neural learning, and ATR-based risk controls rather than relying only on standard indicators.

The UMDE layer evaluates direction using price velocity, entropy, and Causal Price Dynamics to filter weaker conditions. A 5-12-3 neural network trains online on prior bar data and is used for directional confirmation plus dynamic TP/SL guidance.

Risk handling supports fixed lot or risk-percent sizing with auto sizing based on SL distance, commission, and tick value. Trade management includes ATR trailing with optional aggressive behavior, volatility-aware stop widening, RR-based profit hard close, and early loss exits when neural confidence drops. Execution filters include ...

👉 Read | Freelance | @mql5dev

Butterfly Optimization Algorithm (BOA), proposed in 2019 by Arora and Singh, models movement using a fragrance term f = c·I^a and a switch p between global and local search. Fitness is mapped to stimulus intensity I, then converted to fragrance via the power law, with a increasing toward 1 over epochs to shift from broad search to stronger exploitation.

Implementation review found a critical issue in the paper’s update equations. Using x_new = x + (r²·g* - x)·f biases steps toward r²·g*, which tends to pull the population toward the origin and only looks correct when the optimum is at 0.

A corrected form preserves components but fixes geometry: x_new = x + r²·(g* - x)·f, and locally x_new = x + r²·(x_j - x_k)·f. Testing should include optima away from the origin to catch this class of error.

👉 Read | NeuroBook | @mql5dev

ZetaBurst is a tick-driven scalper that evaluates a short rolling burst window (default 4 seconds) and builds a live baseline from recent returns. Trades trigger only on statistically abnormal moves using a z-score against the last InpStatsSampleCount samples, with a default threshold of 3.0σ. A prior ATR-fraction trigger was removed after it over-fired on normal tick noise and produced broad losses.

Two execution modes are provided: momentum (trade with the burst) and reversion (trade against it). Symbol fit is not inferred automatically; both modes require separate testing per instrument.

Order handling accounts for real execution delay. Positions open without an attached stop, then SL/TP are computed from the confirmed fill price, clamped to the symbol minimum stop level. If stop attachment fails, the position is closed immediately. Entries also require a...

👉 Read | AlgoBook | @mql5dev

Modern FX trading stacks still face a trade-off between deterministic indicators and ML models that overfit, especially when train/test boundaries are weak. Information leakage remains a primary source of inflated backtest results.

A hybrid design is described: a fine-tuned Llama 3.2 model paired with Self-Evolving Adversarial Learning (SEAL), deployed on MetaTrader 5. The pipeline enforces a forward-test split (last 7 days held out) to keep validation unbiased.

Data generation labels 24h direction on M15 across eight major pairs, applies a 0.05% move threshold, and uses active class balancing near 50/50. Prompts include RSI, MACD, ATR, Bollinger position, stochastic, and volume ratio with a strict parsable output format.

SEAL adds adversarial self-play, prioritized replay, curriculum difficulty, and evolutionary updates to improve robustness und...

👉 Read | Freelance | @mql5dev

MetaTrader 5 Strategy Tester can produce a strong equity curve that fails to reproduce in forward or live trading. A single backtest reflects one ordering of trades and hides the range of possible equity paths, including tail drawdowns.

Monte Carlo simulation reshuffles the closed-trade PnL sequence to generate many alternative equity curves. Monte Carlo analysis then extracts metrics from that distribution, including bust rate, profit rate, and worst observed drawdown.

A practical Python pipeline can parse the MT5 HTML report, extract initial balance and “Direction=out” PnL rows, run N shuffled paths, and output a mean curve with 5–95% bands plus summary stats. This supports position sizing based on tail risk rather than one curve.

👉 Read | Quotes | @mql5dev

Moving from point-to-point bridges to an event-bus design, this article builds a native Kafka producer directly in MQL5 so one terminal can publish signals to a topic while any number of consumers independently subscribe, replay, and track offsets.

The core work is implementing Kafka’s wire protocol over raw TCP: big-endian framing with a 4-byte length prefix, RecordBatch v2 encoding, base-128 varints with zigzag for deltas, and CRC32C (Castagnoli) with a runtime-built lookup table. RecordBatch encoding is handled as a two-pass “write placeholders then patch” process for lengths and checksums.

On the trading side, signals are versioned and schema-checked at init, keyed by symbol+timeframe for partition ordering, queued and flushed on a timer or batch size, with configurable acks and exponential-backoff retries. Logging via FILE_COMMON produces tester-friendly,...

👉 Read | Quotes | @mql5dev

Part 15’s decision-forest classifier is turned into a tradable EA by separating prediction from execution. The model produces three class scores (bearish/neutral/bullish) from normalized bar features, but a decision layer gates whether that output is allowed to influence positions.

The classifier is refactored into a reusable module that encapsulates feature building, training stats, ALGLIB forest objects, and safe lifecycle rules (no inference until training succeeds, completed-bar history only, reproducible seeding, OOB error exposed for diagnostics).

The EA runs once per new bar, converts raw votes into a stable “directional regime” using a minimum-confidence filter, multi-bar confirmation, and a post-change cooldown. Trade policy is intentionally simple: long-only in bullish, short-only in bearish, otherwise no-entry, with strict risk sizing, spread/...

👉 Read | CodeBase | @mql5dev

Most trading anomaly detection watches price; this design watches execution quality: slippage, fill latency, spread paid, partial fills, and clustered requotes. A native Isolation Forest is implemented entirely in MQL5 and scored on every confirmed entry inside OnTradeTransaction(), acting as a monitoring/circuit-breaker layer rather than a signal generator.

The EA logs a 5D feature vector per fill, trains on a rolling window, and flags multivariate anomalies that single thresholds miss. Isolation Forest is chosen for unsupervised training, low parameter load, and fast scoring (trees x height), with configurable timing logs to verify runtime.

Implementation details that matter: strict feature-count enforcement to prevent invalid scoring, flat-array tree storage for serialization, subsampling without replacement, leaf-size path-length correction, binary p...

👉 Read | Signals | @mql5dev

TQNet targets time-series problems where both short-term moves and long-range structure matter. It does this by maintaining a global correlation tensor that acts as periodic memory, letting relationships form across the sequence without a fixed direction.

The implementation continues in MQL5 with a TQ-MHA module: multi-head attention where queries come from stored correlation parameters while keys/values come from current market data. This merges global context with fresh price action more safely than standard cross-attention.

A new CNeuronTQMHA class reuses cross-attention internals but changes residual/normalization to avoid overwriting local signals. Init is explicit and OpenCL-backed: GeLU for smoother training on noisy quotes, bounded timeframe indexing for carousel switching, and zeroed correlation buffers to prevent biased starts.

👉 Read | Signals | @mql5dev

MQL5 signal experiment shifts focus from detecting setups to timing entries under volatility regime changes. A dual-engine design combines a GARCH(1,1) volatility expansion gate with an optional volatility-scaled LSTM that scores short feature sequences.

Seven execution modes cover breakout, squeeze release, re-entry, mid-band impulse, band walk, pullback continuation, and range escape. Modes share a graded 0.5-centered score, then pass common LongCondition/ShortCondition gates: minimum GARCH expansion, raw pattern threshold, and final entry probability.

LSTM inputs are volatility-normalized (returns scaled by GARCH sigma, ATR and band metrics, expansion ratio). The key test is redundancy: ATR/Bollinger and GARCH can overlap, so validation requires switching algorithm-only vs blended LSTM and comparing backtests plus forward walks.

👉 Read | CodeBase | @mql5dev

Running multiple MQL5 Expert Advisors in one terminal means no shared state beyond GlobalVariables. That namespace is untyped, schema-free, and cannot distinguish stale from current values.

A named-pipe message bus provides explicit message framing and a fixed, typed schema agreed by all participants. The broker EA owns the pipe server, message registry, risk aggregator, and a chart dashboard. Slave EAs connect as clients, report per-magic position state, and receive a centrally computed portfolio risk figure plus the symbol identified as the cause.

Implementation details matter: kernel32.dll imports, message-mode pipes, non-blocking accept via PIPE_NOWAIT, and connection confirmation via PeekNamedPipe due to unreliable last-error reads in MQL5. Messages serialize to a fixed 52-byte little-endian layout, with doubles copied bit-exact.

Limits remain expl...

👉 Read | CodeBase | @mql5dev

Supertrend MTF Indicator is a free multi-timeframe trend tool for MT4. It plots bullish and bearish conditions on the chart using colored Supertrend lines and signal arrows.

Key capabilities include multi-timeframe trend confirmation, Buy/Sell arrow signals, and parameter controls for adapting sensitivity. It can be applied to Forex, Gold, indices, and other MT4 instruments, across different chart timeframes.

This is a visual indicator only and does not place or manage orders. An Expert Advisor based on the same Supertrend MTF logic is available separately for automated execution, while the indicator source is suited to manual analysis, education, and custom modifications.

Any trading tool should be validated on a demo account before live use.

👉 Read | Signals | @mql5dev

A Renko indicator for MT5 that renders fixed-size bricks using BID ticks and outputs only completed bricks. Reversal logic follows the classic two-brick rule, with square, equal-width brick rendering.

The chart window is used only as a host. Visible brick count is calculated automatically, the Renko range is vertically centered, and the real price scale remains on the right.

Implementation avoids offline charts, custom symbols, DLLs, and external libraries. A single primary input is used: Brick Size. Original chart visual settings are restored after removal.

Published as a CodeBase example aimed at studying Renko construction and MQL5 Canvas rendering.

👉 Read | Calendar | @mql5dev